first try of registry allocation

This commit is contained in:
2025-12-23 15:30:03 +01:00
parent ce900632d3
commit e5d912b28e
13 changed files with 461 additions and 350 deletions

View File

@@ -1,155 +0,0 @@
// TODO: store all of values, allocated registers, stack offsets in single allocator
// to be able to find out which kind a specific temp register represent
#pragma once
#include <cstddef>
#include "string.hpp"
#include "ir.hpp"
template<typename Slot>
class Allocator
{
public:
virtual ~Allocator() {};
public:
virtual const Slot& Allocate(const StringView& addr) = 0;
virtual const Slot& Resolve(const StringView& addr) = 0;
};
struct StackSlot
{
size_t offset;
StringView addr;
};
class StackAllocator : public Allocator<StackSlot>
{
public:
StackAllocator() = default;
~StackAllocator() override = default;
public:
const StackSlot& Allocate(const StringView& addr)
{
m_offset_counter += 4;
m_slots.Push(StackSlot { m_offset_counter, addr });
return m_slots.data[m_slots.size - 1];
}
const StackSlot& Resolve(const StringView& addr)
{
for (size_t i = 0; i < m_slots.size; ++i)
{
if (strcmp(m_slots.data[i].addr.c_str(), addr.c_str()) == 0)
{
return m_slots.data[i];
}
}
assert(0 && "could not resolve stack offset for specified address");
}
private:
size_t m_offset_counter = 0;
Builder<StackSlot> m_slots;
};
// struct ConstSlot
// {
// long value;
// StringView addr;
// };
// class ConstAllocator : public Allocator<ConstSlot>
// {
// public:
// ConstAllocator() = default;
// ~ConstAllocator() override = default;
// public:
// const ConstSlot& Allocate(const StringView& addr)
// {
// m_slots.Push(ConstSlot { 0, addr });
// return m_slots.data[m_slots.size - 1];
// }
// const ConstSlot& StoreValue(const StringView& addr, long value)
// {
// for (size_t i = 0; i < m_slots.size; ++i)
// {
// if (strcmp(m_slots.data[i].addr.c_str(), addr.c_str()) == 0)
// {
// m_slots.data[i].value = value;
// return m_slots.data[i];
// }
// }
// assert(0 && "could not resolve const under specified address");
// }
// const ConstSlot& Resolve(const StringView& addr)
// {
// for (size_t i = 0; i < m_slots.size; ++i)
// {
// if (strcmp(m_slots.data[i].addr.c_str(), addr.c_str()) == 0)
// {
// return m_slots.data[i];
// }
// }
// assert(0 && "could not resolve const under specified address");
// }
// private:
// Builder<ConstSlot> m_slots;
// };
// struct RegisterSlot
// {
// const StringView& reg;
// StringView addr;
// };
// class RegisterAllocator : public Allocator<RegisterSlot>
// {
// public:
// RegisterAllocator()
// {
// m_regs.Push(std::move(StringView("eax")));
// m_regs.Push(std::move(StringView("ecx")));
// }
// ~RegisterAllocator() override = default;
// public:
// const RegisterSlot& Allocate(const StringView& addr)
// {
// assert(m_slots.size < m_regs.size && "no space available for allocating to register");
// m_slots.Push(RegisterSlot { m_regs.data[m_slots.size], addr });
// return m_slots.data[m_slots.size - 1];
// }
// const RegisterSlot& Resolve(const StringView& addr)
// {
// for (size_t i = 0; i < m_slots.size; ++i)
// {
// if (strcmp(m_slots.data[i].addr.c_str(), addr.c_str()) == 0)
// {
// return m_slots.data[i];
// }
// }
// assert(0 && "could not resolve register for specified address");
// }
// void Clear()
// {
// m_slots.size = 0;
// }
// private:
// Builder<RegisterSlot> m_slots;
// Builder<StringView> m_regs;
// };
class CodeGenerator
{
public:
virtual ~CodeGenerator() {};
virtual void Generate(const char* filename, View<IR::Op*> ops) = 0;
};

View File

@@ -0,0 +1,17 @@
// TODO: store all of values, allocated registers, stack offsets in single allocator
// to be able to find out which kind a specific temp register represent
#pragma once
#include <cstddef>
#include <cstdarg>
#include "prelude/string.hpp"
#include "ir/ir.hpp"
class CodeGenerator
{
public:
virtual ~CodeGenerator() {};
virtual void Generate(const char* filename, View<IR::Op*> ops) = 0;
};

View File

@@ -0,0 +1,170 @@
#pragma once
#include "codegen/codegen.hpp"
#include "ir/slot.hpp"
class StackFasmX86_64Generator : public CodeGenerator
{
public:
~StackFasmX86_64Generator() override = default;
private:
StringView GetTempAddr(IR::Tmp reg)
{
return std::move((StringBuilder() << 't' << reg).view());
}
StringView GetSlotAddr(const IR::IRSlot& slot)
{
switch (slot.GetType())
{
case IR::IRSlot::Type::REGISTRY:
{
StringBuilder sb;
sb.AppendFormat("r%d", slot.GetSlot() + 10); // for r10, r11, r12 etc.
return sb.view();
} break;
case IR::IRSlot::Type::STACK:
{
StringBuilder sb;
sb.AppendFormat("[rbp-%d]", slot.GetSlot()); // for r10, r11, r12 etc.
return sb.view();
} break;
default:
assert(0 && "TODO: either unreachable or handle properly");
}
}
private:
void GenerateOp(const IR::Op *op)
{
switch (op->GetType())
{
case IR::OpType::EXTERN:
{
auto extrn = reinterpret_cast<const IR::ExternOp *>(op);
auto symbol = extrn->symbol();
appendf("extrn '%s' as __%s\n", symbol.c_str(), symbol.c_str());
appendf("%s = PLT __%s\n", symbol.c_str(), symbol.c_str());
}
break;
case IR::OpType::FN:
{
m_allocator = new SlotAllocator();
auto fn = reinterpret_cast<const IR::FnOp *>(op);
auto name = fn->name();
appendf("public %s\n", name.c_str());
appendf("%s:\n", name.c_str());
appendf("push rbp\n");
appendf("mov rbp, rsp\n");
StringBuilder fnOutput;
StringBuilder* backup = m_output;
m_output = &fnOutput;
if (fn->params().size > 0)
{
// TODO: support multiple parameters
auto param_slot = m_allocator->Allocate(fn->params().data[0]);
appendf("mov %s, rdi\n", GetSlotAddr(param_slot).c_str());
}
for(auto &fOp : fn->ops())
{
GenerateOp(fOp);
}
int stackSize = m_allocator->GetStackSize();
m_output = backup;
appendf("sub rsp, %d\n", stackSize);
*m_output << fnOutput.c_str();
appendf("leave\nret\n");
m_allocator = nullptr;
}
break;
case IR::OpType::CALL:
{
auto call = reinterpret_cast<const IR::CallOp *>(op);
// TODO: support several arguments
if (call->args().size == 1)
{
auto slot = m_allocator->Resolve(GetTempAddr(call->args().data[0]));
appendf("mov rdi, %s\n", GetSlotAddr(slot).c_str());
}
appendf("call %s\n", call->callee().c_str());
auto result_slot = m_allocator->Allocate(GetTempAddr(call->result()));
appendf("mov %s, rax\n", GetSlotAddr(result_slot).c_str());
}
break;
case IR::OpType::LOAD_CONST:
{
auto lc = reinterpret_cast<const IR::LoadConstOp *>(op);
auto addr = GetTempAddr(lc->result());
auto slot = m_allocator->Allocate(addr);
appendf("mov %s, %ld\n", GetSlotAddr(slot).c_str(), lc->value());
}
break;
case IR::OpType::STORE:
{
auto s = reinterpret_cast<const IR::StoreOp *>(op);
appendf("; DEBUG: resolving stack slot at %s\n", s->addr().c_str());
auto slot = m_allocator->Allocate(s->addr());
auto value_slot = m_allocator->Resolve(GetTempAddr(s->src()));
appendf("mov rax, %s\n", GetSlotAddr(value_slot).c_str());
appendf("mov %s, rax\n", GetSlotAddr(slot).c_str());
}
break;
case IR::OpType::LOAD:
{
auto l = reinterpret_cast<const IR::LoadOp *>(op);
auto value_slot = m_allocator->Allocate(GetTempAddr(l->result()));
auto variable_slot = m_allocator->Resolve(l->addr());
appendf("mov rax, %s\n", GetSlotAddr(variable_slot).c_str());
appendf("mov %s, rax\n", GetSlotAddr(value_slot).c_str());
}
break;
case IR::OpType::ADD:
{
auto expr = reinterpret_cast<const IR::AddOp *>(op);
auto lhs_slot = m_allocator->Resolve(GetTempAddr(expr->lhs()));
appendf("mov rax, %s\n", GetSlotAddr(lhs_slot).c_str());
auto rhs_slot = m_allocator->Resolve(GetTempAddr(expr->rhs()));
appendf("add rax, %s\n", GetSlotAddr(rhs_slot).c_str());
auto result_slot = m_allocator->Allocate(GetTempAddr(expr->result()));
appendf("mov %s, rax\n", GetSlotAddr(result_slot).c_str());
}
break;
default: appendf("; NOT HANDLED\n; %s\n", op->Format(0).c_str()); break;
}
}
void appendf(const char * fmt, ...)
{
assert(m_output != nullptr && "nowhere to write");
va_list args;
va_start(args, fmt);
m_output->VAppendFormat(fmt, args);
va_end(args);
}
public:
StringView GetOutput() { return m_output->view(); }
private:
StringBuilder* m_output = nullptr;
public:
void Generate(const char* filename, View<IR::Op*> ops) override
{
m_output = new StringBuilder();
appendf("; fasm x86_64 linux generated assembly using pl\n");
appendf("format ELF64\n");
appendf("section '.text' executable\n");
for (auto& op : ops)
{
GenerateOp(op);
}
}
public:
// StackAllocator* m_stack = nullptr;
// TODO: handle sub-blocks
SlotAllocator* m_allocator = nullptr;
};

View File

@@ -1,140 +0,0 @@
#pragma once
#include "codegen.hpp"
class StackFasmX86_64Generator : public CodeGenerator
{
public:
~StackFasmX86_64Generator() override = default;
private:
int GetStackSize(const IR::OpView ops)
{
int stackSize = 0;
for (auto &op : ops)
{
switch (op->GetType())
{
case IR::OpType::STORE:
case IR::OpType::LOAD:
case IR::OpType::LOAD_CONST:
case IR::OpType::ADD:
case IR::OpType::CALL:
stackSize += 4;
}
}
return stackSize;
}
StringView GetTempAddr(IR::Reg reg)
{
return std::move((StringBuilder() << 't' << reg).view());
}
private:
void GenerateOp(const IR::Op *op)
{
switch (op->GetType())
{
case IR::OpType::EXTERN:
{
auto extrn = reinterpret_cast<const IR::ExternOp *>(op);
auto symbol = extrn->symbol();
printf("extrn '%s' as __%s\n", symbol.c_str(), symbol.c_str());
printf("%s = PLT __%s\n", symbol.c_str(), symbol.c_str());
}
break;
case IR::OpType::FN:
{
auto fn = reinterpret_cast<const IR::FnOp *>(op);
auto name = fn->name();
printf("public %s\n", name.c_str());
printf("%s:\n", name.c_str());
printf("push rbp\n");
printf("mov rbp, rsp\n");
m_stack = new StackAllocator();
int stackSize = GetStackSize(fn->ops());
printf("sub rsp, %d\n", stackSize);
if (fn->params().size > 0)
{
auto param_slot = m_stack->Allocate(fn->params().data[0]);
printf("mov [rbp-%d], edi\n", param_slot.offset);
}
for(auto &fOp : fn->ops())
{
GenerateOp(fOp);
}
operator delete(m_stack);
m_stack = nullptr;
printf("leave\nret\n");
}
break;
case IR::OpType::CALL:
{
auto call = reinterpret_cast<const IR::CallOp *>(op);
// TODO: support several arguments
if (call->args().size == 1)
{
auto slot = m_stack->Resolve(GetTempAddr(call->args().data[0]));
printf("mov edi, [rbp-%d]\n", slot.offset);
}
printf("call %s\n", call->callee().c_str());
auto result_slot = m_stack->Allocate(GetTempAddr(call->result()));
printf("mov dword [rbp-%d], eax\n", result_slot.offset);
}
break;
case IR::OpType::LOAD_CONST:
{
auto lc = reinterpret_cast<const IR::LoadConstOp *>(op);
auto addr = GetTempAddr(lc->result());
auto slot = m_stack->Allocate(addr);
printf("mov dword [rbp-%d], %ld\n", slot.offset, lc->value());
}
break;
case IR::OpType::STORE:
{
auto s = reinterpret_cast<const IR::StoreOp *>(op);
printf("; DEBUG: resolving stack slot at %s\n", s->addr().c_str());
auto slot = m_stack->Allocate(s->addr());
auto value_slot = m_stack->Resolve(GetTempAddr(s->src()));
printf("mov eax, [rbp-%d]\n", value_slot.offset);
printf("mov dword [rbp-%d], eax\n", slot.offset);
}
break;
case IR::OpType::LOAD:
{
auto l = reinterpret_cast<const IR::LoadOp *>(op);
auto value_slot = m_stack->Allocate(GetTempAddr(l->result()));
auto variable_slot = m_stack->Resolve(l->addr());
printf("mov eax, [rbp-%d]\n", variable_slot.offset);
printf("mov dword [rbp-%d], eax\n", value_slot.offset);
}
break;
case IR::OpType::ADD:
{
auto expr = reinterpret_cast<const IR::AddOp *>(op);
auto lhs_slot = m_stack->Resolve(GetTempAddr(expr->lhs()));
printf("mov eax, [rbp-%d]\n", lhs_slot.offset);
auto rhs_slot = m_stack->Resolve(GetTempAddr(expr->rhs()));
printf("add eax, [rbp-%d]\n", rhs_slot.offset);
auto result_slot = m_stack->Allocate(GetTempAddr(expr->result()));
printf("mov dword [rbp-%d], eax\n", result_slot.offset);
}
break;
default: printf("; NOT HANDLED\n; %s\n", op->Format(0).c_str()); break;
}
}
public:
void Generate(const char* filename, View<IR::Op*> ops) override
{
printf("; fasm x86_64 linux generated assembly using pl\n");
printf("format ELF64\n");
printf("section '.text' executable\n");
for (auto& op : ops)
{
GenerateOp(op);
}
}
public:
// TODO: handle sub-blocks
StackAllocator* m_stack = nullptr;
};

61
include/ir/allocator.hpp Normal file
View File

@@ -0,0 +1,61 @@
#pragma once
#include "prelude/string.hpp"
#include "ir/slot.hpp"
#define AVAILABLE_REGISTERS 4
#define AVAILABLE_STACK_SIZE 4096
template<typename Slot, typename SlotAddr = const StringView&>
class Allocator
{
public:
virtual ~Allocator() {};
public:
virtual const Slot& Allocate(SlotAddr addr) = 0;
virtual const Slot& Resolve(SlotAddr addr) = 0;
};
class SlotAllocator : public Allocator<IR::IRSlot>
{
public:
SlotAllocator() = default;
~SlotAllocator() = default;
public:
const IR::IRSlot& Allocate(const StringView& addr) override
{
if (m_regs < AVAILABLE_REGISTERS)
{
m_slots.Push(IR::IRSlot(addr, m_regs++, IR::IRSlot::Type::REGISTRY));
return m_slots.data[m_slots.size - 1];
}
if (m_offset_counter + 8 <= AVAILABLE_STACK_SIZE)
{
m_offset_counter += 8;
m_slots.Push(IR::IRSlot(addr, m_offset_counter, IR::IRSlot::Type::STACK));
return m_slots.data[m_slots.size - 1];
}
// TODO: proper error handling (stack overflow etc.)
assert(0 && "failed to allocate local");
}
const IR::IRSlot& Resolve(const StringView& addr) override
{
for (size_t i = 0; i < m_slots.size; ++i)
{
if (m_slots.data[i].GetAddr() == addr)
{
return m_slots.data[i];
}
}
assert(0 && "could not resolve stack offset for specified address");
}
public:
View<IR::IRSlot> slots() const { return m_slots.view(); }
public:
unsigned int GetStackSize() const { return m_offset_counter; }
public:
int m_regs = 0;
unsigned int m_offset_counter = 0;
Builder<IR::IRSlot> m_slots;
};

View File

@@ -1,6 +1,9 @@
#pragma once
#include "string.hpp"
#include "ast.hpp"
#include <unordered_map>
#include "prelude/string.hpp"
#include "ir/slot.hpp"
#include "parser/ast.hpp"
#include "ir/allocator.hpp"
namespace IR
{
@@ -20,10 +23,10 @@ enum class OpType
#define OP_TYPE(x) \
OpType GetType() const override { return OpType::x; }
using Reg = int;
using Tmp = int;
using RegBuilder = Builder<Reg>;
using RegView = View<Reg>;
using TmpView = View<Tmp>;
using TmpBuilder = Builder<Tmp>;
class Op
{
@@ -40,13 +43,13 @@ using OpBuilder = Builder<Op*>;
class Valued
{
public:
Valued(Reg dest)
Valued(Tmp dest)
: m_dest(dest) {}
~Valued() = default;
public:
Reg result() const { return m_dest; }
Tmp result() const { return m_dest; }
private:
Reg m_dest;
Tmp m_dest;
};
class ExternOp : public Op
@@ -97,13 +100,14 @@ public:
private:
StringView m_name;
OpView m_ops;
View<Tmp> m_slots;
View<StringView> m_params;
};
class LoadConstOp : public Op, public Valued
{
public:
LoadConstOp(Reg dest, long value)
LoadConstOp(Tmp dest, long value)
: Valued(dest), m_value(value) {}
~LoadConstOp() {}
@@ -125,7 +129,7 @@ private:
class LoadOp : public Op, public Valued
{
public:
LoadOp(Reg dest, StringView addr)
LoadOp(Tmp dest, StringView addr)
: Valued(dest), m_addr(addr) {}
~LoadOp() {}
@@ -147,7 +151,7 @@ private:
class StoreOp : public Op
{
public:
StoreOp(StringView addr, Reg src)
StoreOp(StringView addr, Tmp src)
: m_addr(addr), m_src(src) {}
~StoreOp() {}
@@ -162,16 +166,16 @@ public:
}
public:
const StringView& addr() const { return m_addr; }
Reg src() const { return m_src; }
Tmp src() const { return m_src; }
private:
StringView m_addr;
Reg m_src;
Tmp m_src;
};
class AddOp : public Op, public Valued
{
public:
AddOp(Reg dest, Reg lhs, Reg rhs)
AddOp(Tmp dest, Tmp lhs, Tmp rhs)
: Valued(dest), m_lhs(lhs), m_rhs(rhs) {}
~AddOp() {}
@@ -181,21 +185,21 @@ public:
{
StringBuilder sb;
sb.AppendIndent(indent);
sb << 't' << result() << " = ADD t" << m_lhs << ", t" << m_rhs;
sb << 't' << result() << " = ADD " << m_lhs << ", " << m_rhs;
return sb.view();
}
public:
Reg lhs() const { return m_lhs; }
Reg rhs() const { return m_rhs; }
Tmp lhs() const { return m_lhs; }
Tmp rhs() const { return m_rhs; }
private:
Reg m_lhs;
Reg m_rhs;
Tmp m_lhs;
Tmp m_rhs;
};
class CallOp : public Op, public Valued
{
public:
CallOp(Reg dest, StringView callee, RegView args)
CallOp(Tmp dest, StringView callee, TmpView args)
: Valued(dest), m_callee(callee), m_args(args) {}
~CallOp() {}
@@ -215,10 +219,10 @@ public:
}
public:
const StringView& callee() const { return m_callee; }
const RegView& args() const { return m_args; }
const TmpView& args() const { return m_args; }
private:
StringView m_callee;
RegView m_args;
TmpView m_args;
};
class IRBuilder
@@ -228,24 +232,29 @@ public:
: m_root(root) {}
public:
// TODO: support other literals
Reg ParseIntLiteral(const IntLiteralNode* literal)
Tmp ParseIntLiteral(const IntLiteralNode* literal)
{
auto dst = AllocateRegister();
m_ops.Push(new LoadConstOp(dst, literal->integer()));
return dst;
}
Reg ParseVariable(const VariableNode* var)
Tmp ParseVariable(const VariableNode* var)
{
auto dst = AllocateRegister();
m_ops.Push(new LoadOp(dst, var->name()));
return dst;
// auto dst = AllocateRegister();
// m_ops.Push(new LoadOp(dst, var->name()));
if (m_locals.find(var->name()) == m_locals.end())
{
// TODO: throw proper error
assert(0 && "ERROR: variable does not exist");
}
return m_locals[var->name()];
}
Reg ParseFnCall(const FnCallNode* fn)
Tmp ParseFnCall(const FnCallNode* fn)
{
// TODO: support multiple args
auto argRegs = RegBuilder();
auto argRegs = TmpBuilder();
if (fn->arg() != nullptr)
{
auto arg = ParseExpression(fn->arg());
@@ -256,7 +265,7 @@ public:
return dst;
}
Reg ParseFactor(const Node* factor)
Tmp ParseFactor(const Node* factor)
{
switch(factor->GetType())
{
@@ -267,10 +276,10 @@ public:
}
assert(0 && "unreachable");
return -1;
return Tmp();
}
Reg ParseExpression(const Node* expression)
Tmp ParseExpression(const Node* expression)
{
if (expression->GetType() == NodeType::Expression)
{
@@ -295,7 +304,8 @@ public:
void ParseVarDecl(const VarDeclNode* varDecl)
{
auto value = ParseExpression(varDecl->value());
m_ops.Push(new StoreOp(varDecl->name(), value));
// m_ops.Push(new StoreOp(varDecl->name(), value));
m_locals.insert(std::make_pair(varDecl->name(), value));
}
void ParseBlock(const CompoundNode* compound)
@@ -333,15 +343,17 @@ public:
// TODO: think about safety (copying m_ops.data before giving)
OpView ops() const { return OpView(m_ops.data, m_ops.size); }
private:
Reg AllocateRegister()
Tmp AllocateRegister()
{
return m_reg_counter++;
return m_tmp_counter++;
}
private:
OpBuilder m_ops;
const Node* m_root = nullptr;
Reg m_reg_counter = 0;
unsigned int m_tmp_counter = 0;
std::unordered_map<StringView, Tmp> m_locals;
};
} // namespace IR

48
include/ir/slot.hpp Normal file
View File

@@ -0,0 +1,48 @@
#pragma once
#include "prelude/string.hpp"
namespace IR
{
struct IRSlot
{
enum class Type
{
UNKNOWN = 0,
STACK,
REGISTRY,
};
public:
IRSlot(StringView addr, unsigned int slot, Type slotType) : m_addr(addr), m_slot(slot), m_slotType(slotType) {}
IRSlot() = default;
public:
Type GetType() const { return m_slotType; }
const StringView& GetAddr() const { return m_addr; }
unsigned int GetSlot() const { return m_slot; }
public:
StringView Format() const {
StringBuilder sb;
switch(GetType())
{
case Type::REGISTRY:
sb << "r" << GetSlot();
break;
case Type::STACK:
sb << "s[" << GetSlot() << "]";
break;
default:
sb << "(UNKNOWN_SLOT_TYPE)";
break;
}
return sb.view();
}
private:
StringView m_addr;
unsigned int m_slot;
Type m_slotType;
};
using IRSlotBuilder = Builder<IRSlot>;
using IRSlotView = View<IRSlot>;
} // namespace IR

View File

@@ -2,7 +2,7 @@
#include <cstring>
#include <cassert>
#include <vector>
#include "lexer.hpp"
#include "parser/lexer.hpp"
enum class NodeType
{

View File

@@ -3,7 +3,7 @@
#include <cstring>
#include <cctype>
#include "string.hpp"
#include "prelude/string.hpp"
enum class TokenType
{
@@ -110,13 +110,13 @@ public:
c = m_code.data[m_pos++];
}
if (std::isalpha(c) != 0)
if (std::isalpha(c) != 0 || c == '_')
{
StringBuilder s;
long offset_start = m_pos - m_last_newline;
s.Push(c);
// id
while (std::isalpha(m_code.data[m_pos]) != 0)
while (std::isalpha(m_code.data[m_pos]) != 0 || m_code.data[m_pos] == '_')
{
s.Push(m_code.data[m_pos++]);
}

View File

@@ -2,9 +2,11 @@
#include <cstddef>
#include <cstdio>
#include <cassert>
#include <cstdarg>
#include <cstring>
#include <cstdlib>
#include <utility>
#include <functional>
template<typename T>
struct View
@@ -105,6 +107,48 @@ public:
const char* c_str() const { return data; }
};
namespace std
{
template<>
struct hash<StringView>
{
size_t operator()(const StringView& sv) const noexcept
{
// FNV-1a hash (simple, fast, good enough)
constexpr size_t fnv_offset = 14695981039346656037ull;
constexpr size_t fnv_prime = 1099511628211ull;
size_t hash = fnv_offset;
// Exclude the null terminator if present
size_t len = (sv.size > 0 && sv.data[sv.size - 1] == '\0')
? sv.size - 1
: sv.size;
for (size_t i = 0; i < len; ++i)
{
hash ^= static_cast<unsigned char>(sv.data[i]);
hash *= fnv_prime;
}
return hash;
}
};
}
inline bool operator==(const StringView& a, const StringView& b)
{
// Sizes include null terminator; logical string length is size-1
if (a.size != b.size) return false;
if (a.size == 0) return true;
return std::memcmp(a.data, b.data, a.size) == 0;
}
inline bool operator!=(const StringView& a, const StringView& b)
{
return !(a == b);
}
template<typename T>
struct Builder
{
@@ -217,6 +261,52 @@ public:
size += indent;
}
void VAppendFormat(const char* fmt, va_list args)
{
// Make a copy because vsnprintf consumes the va_list
va_list argsCopy;
va_copy(argsCopy, args);
int len = std::vsnprintf(nullptr, 0, fmt, argsCopy);
va_end(argsCopy);
if (len <= 0)
return;
this->grow(this->size + len + 1);
std::vsnprintf(this->data + this->size, len + 1, fmt, args);
this->size += len;
}
void AppendFormat(const char* fmt, ...)
{
va_list args;
va_start(args, fmt);
// 1) compute length (excluding terminator)
va_list argsCopy;
va_copy(argsCopy, args);
int len = std::vsnprintf(nullptr, 0, fmt, argsCopy);
va_end(argsCopy);
if (len <= 0)
{
va_end(args);
return;
}
// 2) reserve space
this->grow(this->size + len + 1);
// 3) write formatted string
std::vsnprintf(this->data + this->size, len + 1, fmt, args);
this->size += len;
va_end(args);
}
// Return a null-terminated string pointer owned by builder
const char* c_str()
{
@@ -272,4 +362,12 @@ public:
Extend(buf);
return *this;
}
StringBuilder& operator<<(unsigned int v)
{
char buf[32];
snprintf(buf, sizeof(buf), "%d", v);
Extend(buf);
return *this;
}
};