feat: implement method calls, recursive structs, and pointer comparisons (#272)

This commit introduces several core enhancements to the LLVM backend and
the parser, including support for method-style function calls, recursive
struct definitions via opaque types, and robust pointer arithmetic/comparisons.

Changes:
- **Method Calls & Structs**:
  - Implemented method-style calls (`obj.method()`) by resolving to
    `struct_name_method_name` functions and automatically passing `self`.
  - Refactored struct type generation to use **Opaque Structs**, allowing
    for recursive data structures (e.g., linked lists).
  - Improved `FieldAccess` to use centralized address generation logic.
  - Added implicit type coercion for struct literal fields.
- **Pointer & Type Logic**:
  - Added support for binary comparisons between pointers (`ptr == ptr`)
    and mixed pointer-integer comparisons (`ptr == 0`).
  - Implemented implicit type coercion for assignments and return statements.
  - Added integer promotion (z-extend) for values smaller than 32-bit in
    formatted IO functions.
- **Lexer & Parser Enhancements**:
  - Added safety check to prevent unescaped newlines in string literals.
  - Updated type parser to skip newlines, improving formatting flexibility
    for generic types.
  - Registered `CharLiteral` as a valid expression start.
- **Formatting Engine**:
  - Enhanced the formatting logic to support specifiers within placeholders
    (e.g., `{x}` for hex, `{c}` for char, `{p}` for pointer).
  - Added automatic C-string (`%s`) mapping for `i8` pointers.
- **Testing & Utils**:
  - Significantly updated `test56.wave` with a robust TCP socket server
    implementation demonstrating new syscalls, structs, and methods.
  - Added a Python test runner helper for verifying server-side responses.
  - Added `parse_placeholders` to `utils` for advanced format specifier parsing.

This update significantly increases the language's expressiveness,
enabling more complex system-level programming patterns.

Signed-off-by: LunaStev <luna@lunastev.org>

Author: LunaStev

front/lexer/src/lexer/literals.rs

@@ -2,10 +2,12 @@ use super::Lexer;
 
 impl<'a> Lexer<'a> {
     pub(crate) fn string(&mut self) -> String {
-        if self.peek() == '"' { self.advance(); }
-
         let mut string_literal = String::new();
         while !self.is_at_end() && self.peek() != '"' {
+            if self.peek() == '\n' {
+                panic!("Unterminated string (newline in string literal).");
+            }
+
             let c = self.advance();
 
             if c == '\\' {

front/parser/src/parser/types.rs

@@ -63,6 +63,7 @@ pub fn is_expression_start(token_type: &TokenType) -> bool {
             | TokenType::Lbrack
             | TokenType::Asm
             | TokenType::Deref
+            | TokenType::CharLiteral(_)
     )
 }
 
@@ -218,7 +219,9 @@ pub fn parse_type_from_stream(tokens: &mut Peekable<Iter<Token>>) -> Option<Wave
     let type_token = tokens.next()?;
 
     if let TokenType::Identifier(name) = &type_token.token_type {
-        while matches!(tokens.peek().map(|t| &t.token_type), Some(TokenType::Whitespace)) {
+        while matches!(tokens.peek().map(|t| &t.token_type),
+            Some(TokenType::Whitespace | TokenType::Newline)
+        ) {
             tokens.next();
         }
 

llvm_temporary/src/llvm_temporary/expression/rvalue/assign.rs

@@ -3,6 +3,7 @@ use crate::llvm_temporary::llvm_codegen::generate_address_ir;
 use inkwell::types::{AnyTypeEnum, BasicTypeEnum};
 use inkwell::values::{BasicValue, BasicValueEnum};
 use parser::ast::{AssignOperator, Expression};
+use crate::llvm_temporary::statement::variable::{coerce_basic_value, CoercionMode};
 
 pub(crate) fn gen_assign_operation<'ctx, 'a>(
     env: &mut ExprGenEnv<'ctx, 'a>,
@@ -14,6 +15,34 @@ pub(crate) fn gen_assign_operation<'ctx, 'a>(
         env.context, env.builder, target, env.variables, env.module, env.struct_types, env.struct_field_indices
     );
 
+    let element_type = match ptr.get_type().get_element_type() {
+        AnyTypeEnum::IntType(t) => BasicTypeEnum::IntType(t),
+        AnyTypeEnum::FloatType(t) => BasicTypeEnum::FloatType(t),
+        AnyTypeEnum::PointerType(t) => BasicTypeEnum::PointerType(t),
+        AnyTypeEnum::ArrayType(t) => BasicTypeEnum::ArrayType(t),
+        AnyTypeEnum::StructType(t) => BasicTypeEnum::StructType(t),
+        AnyTypeEnum::VectorType(t) => BasicTypeEnum::VectorType(t),
+        _ => panic!("Unsupported LLVM element type"),
+    };
+
+    if matches!(operator, AssignOperator::Assign) {
+        let mut rhs = env.gen(value, Some(element_type));
+
+        if rhs.get_type() != element_type {
+            rhs = coerce_basic_value(
+                env.context,
+                env.builder,
+                rhs,
+                element_type,
+                "assign_cast",
+                CoercionMode::Implicit,
+            );
+        }
+
+        env.builder.build_store(ptr, rhs).unwrap();
+        return rhs;
+    }
+
     let current_val = env.builder.build_load(ptr, "load_current").unwrap();
 
     let new_val = env.gen(value, Some(current_val.get_type()));
@@ -55,17 +84,7 @@ pub(crate) fn gen_assign_operation<'ctx, 'a>(
             AssignOperator::RemAssign => env.builder.build_float_rem(lhs, rhs, "rem_assign").unwrap().as_basic_value_enum(),
         },
 
-        _ => panic!("Type mismatch or unsupported type in AssignOperation"),
-    };
-
-    let element_type = match ptr.get_type().get_element_type() {
-        AnyTypeEnum::IntType(t) => BasicTypeEnum::IntType(t),
-        AnyTypeEnum::FloatType(t) => BasicTypeEnum::FloatType(t),
-        AnyTypeEnum::PointerType(t) => BasicTypeEnum::PointerType(t),
-        AnyTypeEnum::ArrayType(t) => BasicTypeEnum::ArrayType(t),
-        AnyTypeEnum::StructType(t) => BasicTypeEnum::StructType(t),
-        AnyTypeEnum::VectorType(t) => BasicTypeEnum::VectorType(t),
-        _ => panic!("Unsupported LLVM element type"),
+        _ => panic!("AssignOperation (+=, -=, ...) only supports numeric types"),
     };
 
     let result_casted = match (result, element_type) {

llvm_temporary/src/llvm_temporary/expression/rvalue/binary.rs

@@ -144,6 +144,75 @@ pub(crate) fn gen<'ctx, 'a>(
                 _ => panic!("Unsupported mixed-type operator (float + int)"),
             }
         }
+        (BasicValueEnum::PointerValue(lp), BasicValueEnum::PointerValue(rp)) => {
+            let i64_ty = env.context.i64_type();
+            let li = env.builder.build_ptr_to_int(lp, i64_ty, "l_ptr2int").unwrap();
+            let ri = env.builder.build_ptr_to_int(rp, i64_ty, "r_ptr2int").unwrap();
+
+            let mut result = match operator {
+                Operator::Equal => env.builder.build_int_compare(IntPredicate::EQ, li, ri, "ptreq").unwrap(),
+                Operator::NotEqual => env.builder.build_int_compare(IntPredicate::NE, li, ri, "ptrne").unwrap(),
+                _ => panic!("Unsupported pointer operator: {:?}", operator),
+            };
+
+            if let Some(inkwell::types::BasicTypeEnum::IntType(target_ty)) = expected_type {
+                if result.get_type() != target_ty {
+                    result = env.builder.build_int_cast(result, target_ty, "cast_result").unwrap();
+                }
+            }
+
+            return result.as_basic_value_enum();
+        }
+
+        (BasicValueEnum::PointerValue(lp), BasicValueEnum::IntValue(ri)) => {
+            let i64_ty = env.context.i64_type();
+            let li = env.builder.build_ptr_to_int(lp, i64_ty, "l_ptr2int").unwrap();
+
+            let ri = if ri.get_type().get_bit_width() == 64 {
+                ri
+            } else {
+                env.builder.build_int_cast(ri, i64_ty, "r_i64").unwrap()
+            };
+
+            let mut result = match operator {
+                Operator::Equal => env.builder.build_int_compare(IntPredicate::EQ, li, ri, "ptreq0").unwrap(),
+                Operator::NotEqual => env.builder.build_int_compare(IntPredicate::NE, li, ri, "ptrne0").unwrap(),
+                _ => panic!("Unsupported ptr/int operator: {:?}", operator),
+            };
+
+            if let Some(inkwell::types::BasicTypeEnum::IntType(target_ty)) = expected_type {
+                if result.get_type() != target_ty {
+                    result = env.builder.build_int_cast(result, target_ty, "cast_result").unwrap();
+                }
+            }
+
+            return result.as_basic_value_enum();
+        }
+
+        (BasicValueEnum::IntValue(li), BasicValueEnum::PointerValue(rp)) => {
+            let i64_ty = env.context.i64_type();
+            let li = if li.get_type().get_bit_width() == 64 {
+                li
+            } else {
+                env.builder.build_int_cast(li, i64_ty, "l_i64").unwrap()
+            };
+
+            let ri = env.builder.build_ptr_to_int(rp, i64_ty, "r_ptr2int").unwrap();
+
+            let mut result = match operator {
+                Operator::Equal => env.builder.build_int_compare(IntPredicate::EQ, li, ri, "ptreq0").unwrap(),
+                Operator::NotEqual => env.builder.build_int_compare(IntPredicate::NE, li, ri, "ptrne0").unwrap(),
+                _ => panic!("Unsupported int/ptr operator: {:?}", operator),
+            };
+
+            if let Some(inkwell::types::BasicTypeEnum::IntType(target_ty)) = expected_type {
+                if result.get_type() != target_ty {
+                    result = env.builder.build_int_cast(result, target_ty, "cast_result").unwrap();
+                }
+            }
+
+            return result.as_basic_value_enum();
+        }
 
         _ => panic!("Type mismatch in binary expression"),
     }

llvm_temporary/src/llvm_temporary/expression/rvalue/calls.rs

@@ -1,9 +1,32 @@
-use inkwell::types::BasicTypeEnum;
+use inkwell::types::{AsTypeRef, BasicTypeEnum};
 use super::ExprGenEnv;
 use inkwell::values::{BasicMetadataValueEnum, BasicValue, BasicValueEnum};
 use parser::ast::{Expression, WaveType};
 use crate::llvm_temporary::statement::variable::{coerce_basic_value, CoercionMode};
 
+fn normalize_struct_name(raw: &str) -> &str {
+    raw.strip_prefix("struct.").unwrap_or(raw).trim_start_matches('%')
+}
+
+fn resolve_struct_key<'ctx>(
+    st: inkwell::types::StructType<'ctx>,
+    struct_types: &std::collections::HashMap<String, inkwell::types::StructType<'ctx>>,
+) -> String {
+    if let Some(raw) = st.get_name().and_then(|n| n.to_str().ok()) {
+        return normalize_struct_name(raw).to_string();
+    }
+
+    let st_ref = st.as_type_ref();
+    for (name, ty) in struct_types {
+        if ty.as_type_ref() == st_ref {
+            return name.clone();
+        }
+    }
+
+    panic!("LLVM struct type has no name and cannot be matched to struct_types");
+}
+
+
 pub(crate) fn gen_method_call<'ctx, 'a>(
     env: &mut ExprGenEnv<'ctx, 'a>,
     object: &Expression,
@@ -59,6 +82,67 @@ pub(crate) fn gen_method_call<'ctx, 'a>(
         }
     }
 
+    {
+        let obj_preview = env.gen(object, None);
+        let obj_ty = obj_preview.get_type();
+
+        let struct_name_opt: Option<String> = match obj_ty {
+            BasicTypeEnum::StructType(st) => Some(resolve_struct_key(st, env.struct_types)),
+            BasicTypeEnum::PointerType(p) if p.get_element_type().is_struct_type() => {
+                Some(resolve_struct_key(p.get_element_type().into_struct_type(), env.struct_types))
+            }
+            _ => None,
+        };
+
+        if let Some(struct_name) = struct_name_opt {
+            let fn_name = format!("{}_{}", struct_name, name);
+
+            if let Some(function) = env.module.get_function(&fn_name) {
+                let fn_type = function.get_type();
+                let param_types = fn_type.get_param_types();
+                let expected_self = param_types.get(0).cloned();
+
+                let mut obj_val = obj_preview;
+                if let Some(et) = expected_self {
+                    obj_val = coerce_basic_value(
+                        env.context,
+                        env.builder,
+                        obj_val,
+                        et,
+                        "self_cast",
+                        CoercionMode::Implicit,
+                    );
+                }
+
+                let mut call_args: Vec<BasicMetadataValueEnum> = Vec::new();
+                call_args.push(obj_val.into());
+
+                for (i, arg_expr) in args.iter().enumerate() {
+                    let expected_ty = param_types.get(i + 1).cloned();
+                    let mut arg_val = env.gen(arg_expr, expected_ty);
+                    if let Some(et) = expected_ty {
+                        arg_val = coerce_basic_value(
+                            env.context, env.builder, arg_val, et, &format!("arg{}_cast", i),
+                            CoercionMode::Implicit
+                        );
+                    }
+                    call_args.push(arg_val.into());
+                }
+
+                let call_site = env
+                    .builder
+                    .build_call(function, &call_args, &format!("call_{}", fn_name))
+                    .unwrap();
+
+                if function.get_type().get_return_type().is_some() {
+                    return call_site.try_as_basic_value().left().unwrap();
+                } else {
+                    return env.context.i32_type().const_zero().as_basic_value_enum();
+                }
+            }
+        }
+    }
+
     // method-style call: fn(self, ...)
     let function = env
         .module

llvm_temporary/src/llvm_temporary/expression/rvalue/structs.rs

@@ -1,6 +1,7 @@
 use super::ExprGenEnv;
 use inkwell::values::{BasicValue, BasicValueEnum};
 use parser::ast::{Expression, WaveType};
+use crate::llvm_temporary::llvm_codegen::generate_address_ir;
 
 pub(crate) fn gen_struct_literal<'ctx, 'a>(
     env: &mut ExprGenEnv<'ctx, 'a>,
@@ -27,7 +28,11 @@ pub(crate) fn gen_struct_literal<'ctx, 'a>(
             .get(field_name)
             .unwrap_or_else(|| panic!("Field '{}' not found in struct '{}'", field_name, name));
 
-        let field_val = env.gen(field_expr, None);
+        let expected_field_ty = struct_ty
+            .get_field_type_at_index(*idx)
+            .unwrap_or_else(|| panic!("No field type at index {} for struct '{}'", idx, name));
+
+        let field_val = env.gen(field_expr, Some(expected_field_ty));
 
         let field_ptr = env
             .builder
@@ -48,40 +53,24 @@ pub(crate) fn gen_field_access<'ctx, 'a>(
     object: &Expression,
     field: &str,
 ) -> BasicValueEnum<'ctx> {
-    let var_name = match object {
-        Expression::Variable(name) => name,
-        other => panic!("FieldAccess on non-variable object not supported yet: {:?}", other),
-    };
-
-    let var_info = env
-        .variables
-        .get(var_name)
-        .unwrap_or_else(|| panic!("Variable '{}' not found for field access", var_name));
-
-    let struct_name = match &var_info.ty {
-        WaveType::Struct(name) => name,
-        other_ty => panic!(
-            "Field access on non-struct type {:?} for variable '{}'",
-            other_ty, var_name
-        ),
+    let full = Expression::FieldAccess {
+        object: Box::new(object.clone()),
+        field: field.to_string(),
     };
 
-    let field_indices = env
-        .struct_field_indices
-        .get(struct_name)
-        .unwrap_or_else(|| panic!("Field index map for struct '{}' not found", struct_name));
-
-    let idx = field_indices
-        .get(field)
-        .unwrap_or_else(|| panic!("Field '{}' not found in struct '{}'", field, struct_name));
-
-    let field_ptr = env
-        .builder
-        .build_struct_gep(var_info.ptr, *idx, &format!("{}.{}", var_name, field))
-        .unwrap();
+    let ptr = generate_address_ir(
+        env.context,
+        env.builder,
+        &full,
+        env.variables,
+        env.module,
+        env.struct_types,
+        env.struct_field_indices,
+    );
 
     env.builder
-        .build_load(field_ptr, &format!("load_{}_{}", var_name, field))
+        .build_load(ptr, &format!("load_field_{}", field))
         .unwrap()
         .as_basic_value_enum()
 }
+