some split

This commit is contained in:
blueloveTH 2024-06-07 00:53:18 +08:00
parent dfd0d0532b
commit 8b2b8ab83a
10 changed files with 272 additions and 238 deletions

View File

@ -1,203 +0,0 @@
#pragma once
#include "pocketpy/common/utils.hpp"
#include "pocketpy/common/str.hpp"
#include "pocketpy/common/config.h"
#include <stdexcept>
namespace pkpy {
template <typename T>
constexpr T default_invalid_value() {
if constexpr(std::is_same_v<int, T>)
return -1;
else
return nullptr;
}
template <typename T>
struct NameDictImpl {
PK_ALWAYS_PASS_BY_POINTER(NameDictImpl)
using Item = std::pair<StrName, T>;
constexpr static uint16_t kInitialCapacity = 16;
static_assert(is_pod_v<T>);
float _load_factor;
uint16_t _size;
uint16_t _capacity;
uint16_t _critical_size;
uint16_t _mask;
Item* _items;
#define HASH_PROBE_1(key, ok, i) \
ok = false; \
i = key.index & _mask; \
while(!_items[i].first.empty()) { \
if(_items[i].first == (key)) { \
ok = true; \
break; \
} \
i = (i + 1) & _mask; \
}
#define HASH_PROBE_0 HASH_PROBE_1
NameDictImpl(float load_factor = PK_INST_ATTR_LOAD_FACTOR) : _load_factor(load_factor), _size(0) {
_set_capacity_and_alloc_items(kInitialCapacity);
}
~NameDictImpl() { std::free(_items); }
uint16_t size() const { return _size; }
uint16_t capacity() const { return _capacity; }
void _set_capacity_and_alloc_items(uint16_t val) {
_capacity = val;
_critical_size = val * _load_factor;
_mask = val - 1;
_items = (Item*)std::malloc(_capacity * sizeof(Item));
std::memset(_items, 0, _capacity * sizeof(Item));
}
void set(StrName key, T val) {
bool ok;
uint16_t i;
HASH_PROBE_1(key, ok, i);
if(!ok) {
_size++;
if(_size > _critical_size) {
_rehash_2x();
HASH_PROBE_1(key, ok, i);
}
_items[i].first = key;
}
_items[i].second = val;
}
void _rehash_2x() {
Item* old_items = _items;
uint16_t old_capacity = _capacity;
_set_capacity_and_alloc_items(_capacity * 2);
for(uint16_t i = 0; i < old_capacity; i++) {
if(old_items[i].first.empty()) continue;
bool ok;
uint16_t j;
HASH_PROBE_1(old_items[i].first, ok, j);
assert(!ok);
_items[j] = old_items[i];
}
std::free(old_items);
}
T try_get(StrName key) const {
bool ok;
uint16_t i;
HASH_PROBE_0(key, ok, i);
if(!ok) return default_invalid_value<T>();
return _items[i].second;
}
T* try_get_2(StrName key) const {
bool ok;
uint16_t i;
HASH_PROBE_0(key, ok, i);
if(!ok) return nullptr;
return &_items[i].second;
}
T try_get_likely_found(StrName key) const {
uint16_t i = key.index & _mask;
if(_items[i].first == key) return _items[i].second;
i = (i + 1) & _mask;
if(_items[i].first == key) return _items[i].second;
return try_get(key);
}
T* try_get_2_likely_found(StrName key) const {
uint16_t i = key.index & _mask;
if(_items[i].first == key) return &_items[i].second;
i = (i + 1) & _mask;
if(_items[i].first == key) return &_items[i].second;
return try_get_2(key);
}
bool del(StrName key) {
bool ok;
uint16_t i;
HASH_PROBE_0(key, ok, i);
if(!ok) return false;
_items[i].first = StrName();
_items[i].second = nullptr;
_size--;
// tidy
uint16_t pre_z = i;
uint16_t z = (i + 1) & _mask;
while(!_items[z].first.empty()) {
uint16_t h = _items[z].first.index & _mask;
if(h != i) break;
std::swap(_items[pre_z], _items[z]);
pre_z = z;
z = (z + 1) & _mask;
}
return true;
}
template <typename __Func>
void apply(__Func func) const {
for(uint16_t i = 0; i < _capacity; i++) {
if(_items[i].first.empty()) continue;
func(_items[i].first, _items[i].second);
}
}
bool contains(StrName key) const {
bool ok;
uint16_t i;
HASH_PROBE_0(key, ok, i);
return ok;
}
T operator[] (StrName key) const {
T* val = try_get_2_likely_found(key);
if(val == nullptr) { throw std::runtime_error(_S("NameDict key not found: ", key.escape()).str()); }
return *val;
}
array<StrName> keys() const {
array<StrName> v(_size);
int j = 0;
for(uint16_t i = 0; i < _capacity; i++) {
if(_items[i].first.empty()) continue;
new (&v[j++]) StrName(_items[i].first);
}
return v;
}
array<Item> items() const {
array<Item> v(_size);
int j = 0;
apply([&](StrName key, T val) {
new (&v[j++]) Item(key, val);
});
return v;
}
void clear() {
for(uint16_t i = 0; i < _capacity; i++) {
_items[i].first = StrName();
_items[i].second = nullptr;
}
_size = 0;
}
#undef HASH_PROBE_0
#undef HASH_PROBE_1
};
} // namespace pkpy

View File

@ -431,6 +431,12 @@ struct small_map {
return &it->second;
}
V get(const K& key, V default_value) const {
static_assert(is_pod_v<V>);
auto it = try_get(key);
return it ? *it : default_value;
}
bool contains(const K& key) const { return try_get(key) != nullptr; }
void clear() { _data.clear(); }

View File

@ -2,7 +2,7 @@
#include "pocketpy/common/any.hpp"
#include "pocketpy/objects/tuplelist.hpp"
#include "pocketpy/objects/object.hpp"
#include "pocketpy/objects/namedict.hpp"
#include "pocketpy/objects/sourcedata.hpp"
namespace pkpy {
@ -89,9 +89,9 @@ struct CodeObject {
small_vector_2<StrName, 8> varnames; // local variables
int nlocals; // varnames.size()
NameDictInt varnames_inv;
small_map<StrName, int> varnames_inv;
vector<CodeBlock> blocks;
NameDictInt labels;
small_map<StrName, int> labels;
vector<FuncDecl_> func_decls;
int start_line;
@ -131,10 +131,10 @@ struct FuncDecl {
FuncType type = FuncType::UNSET;
NameDictInt kw_to_index;
small_map<StrName, int> kw_to_index;
void add_kwarg(int index, StrName key, PyVar value) {
kw_to_index.set(key, index);
kw_to_index.insert(key, index);
kwargs.push_back(KwArg{index, key, value});
}

View File

@ -0,0 +1,64 @@
#pragma once
#include "pocketpy/common/config.h"
#include "pocketpy/common/str.hpp"
#include "pocketpy/common/utils.hpp"
#include "pocketpy/objects/object.hpp"
namespace pkpy {
struct NameDict {
PK_ALWAYS_PASS_BY_POINTER(NameDict)
using Item = std::pair<StrName, PyVar>;
constexpr static uint16_t kInitialCapacity = 16;
static_assert(is_pod_v<PyVar>);
float _load_factor;
uint16_t _size;
uint16_t _capacity;
uint16_t _critical_size;
uint16_t _mask;
Item* _items;
NameDict(float load_factor = PK_INST_ATTR_LOAD_FACTOR) : _load_factor(load_factor), _size(0) {
_set_capacity_and_alloc_items(kInitialCapacity);
}
~NameDict() { std::free(_items); }
uint16_t size() const { return _size; }
uint16_t capacity() const { return _capacity; }
void _set_capacity_and_alloc_items(uint16_t val);
void set(StrName key, PyVar val);
void _rehash_2x();
PyVar try_get(StrName key) const;
PyVar* try_get_2(StrName key) const;
PyVar try_get_likely_found(StrName key) const;
PyVar* try_get_2_likely_found(StrName key) const;
bool del(StrName key);
bool contains(StrName key) const;
PyVar operator[] (StrName key) const;
array<StrName> keys() const;
array<Item> items() const;
void clear();
void apply(void (*f)(StrName, PyVar, void*), void* data);
};
static_assert(sizeof(NameDict) <= 128);
using NameDict_ = std::shared_ptr<NameDict>;
} // namespace pkpy

View File

@ -1,14 +1,12 @@
#pragma once
#include "pocketpy/common/namedict.hpp"
#include "pocketpy/common/str.hpp"
#include "pocketpy/common/config.h"
#include "pocketpy/objects/base.hpp"
namespace pkpy {
using NameDict = NameDictImpl<PyVar>;
using NameDict_ = std::shared_ptr<NameDict>;
using NameDictInt = NameDictImpl<int>;
static_assert(sizeof(NameDict) <= 128);
struct NameDict;
struct PyObject final {
bool gc_marked; // whether this object is marked
@ -30,13 +28,11 @@ struct PyObject final {
return *_attr;
}
PyVar attr(StrName name) const {
assert(is_attr_valid());
return (*_attr)[name];
}
PyObject(Type type) : gc_marked(false), type(type), _attr(nullptr) {}
PyVar attr(StrName name) const;
static NameDict* __init_namedict(float lf);
template <typename T, typename... Args>
void placement_new(Args&&... args) {
static_assert(std::is_same_v<T, std::decay_t<T>>);
@ -44,11 +40,11 @@ struct PyObject final {
// backdoor for important builtin types
if constexpr(std::is_same_v<T, DummyInstance>) {
_attr = new NameDict();
_attr = __init_namedict(PK_INST_ATTR_LOAD_FACTOR);
} else if constexpr(std::is_same_v<T, Type>) {
_attr = new NameDict(PK_TYPE_ATTR_LOAD_FACTOR);
_attr = __init_namedict(PK_TYPE_ATTR_LOAD_FACTOR);
} else if constexpr(std::is_same_v<T, DummyModule>) {
_attr = new NameDict(PK_TYPE_ATTR_LOAD_FACTOR);
_attr = __init_namedict(PK_TYPE_ATTR_LOAD_FACTOR);
}
}
};

View File

@ -99,18 +99,18 @@ void CodeEmitContext::patch_jump(int index) {
bool CodeEmitContext::add_label(StrName name) {
if(co->labels.contains(name)) return false;
co->labels.set(name, co->codes.size());
co->labels.insert(name, co->codes.size());
return true;
}
int CodeEmitContext::add_varname(StrName name) {
// PK_MAX_CO_VARNAMES will be checked when pop_context(), not here
int index = co->varnames_inv.try_get(name);
int index = co->varnames_inv.get(name, -1);
if(index >= 0) return index;
co->varnames.push_back(name);
co->nlocals++;
index = co->varnames.size() - 1;
co->varnames_inv.set(name, index);
co->varnames_inv.insert(name, index);
return index;
}
@ -156,7 +156,7 @@ void CodeEmitContext::emit_store_name(NameScope scope, StrName name, int line) {
}
void NameExpr::emit_(CodeEmitContext* ctx) {
int index = ctx->co->varnames_inv.try_get(name);
int index = ctx->co->varnames_inv.get(name, -1);
if(scope == NAME_LOCAL && index >= 0) {
ctx->emit_(OP_LOAD_FAST, index, line);
} else {

View File

@ -780,7 +780,7 @@ PyVar VM::__run_top_frame() {
case OP_JUMP_ABSOLUTE_TOP: DISPATCH_JUMP_ABSOLUTE(_CAST(int, POPX()))
case OP_GOTO: {
StrName _name(byte.arg);
int target = frame->co->labels.try_get_likely_found(_name);
int target = frame->co->labels.get(_name, -1);
if(target < 0) RuntimeError(_S("label ", _name.escape(), " not found"));
frame->prepare_jump_break(&s_data, target);
DISPATCH_JUMP_ABSOLUTE(target)

View File

@ -5,17 +5,17 @@
namespace pkpy {
PyVar* FastLocals::try_get_name(StrName name) {
int index = co->varnames_inv.try_get(name);
int index = co->varnames_inv.get(name, -1);
if(index == -1) return nullptr;
return &a[index];
}
NameDict_ FastLocals::to_namedict() {
NameDict_ dict = std::make_shared<NameDict>();
co->varnames_inv.apply([&](StrName name, int index) {
for(auto [name, index]: co->varnames_inv){
PyVar value = a[index];
if(value) dict->set(name, value);
});
}
return dict;
}

View File

@ -437,9 +437,10 @@ void VM::__obj_gc_mark(PyObject* obj) {
const PyTypeInfo* ti = _tp_info(obj->type);
if(ti->vt._gc_mark) ti->vt._gc_mark(obj->_value_ptr(), this);
if(obj->is_attr_valid()) {
obj->attr().apply([this](StrName _, PyVar obj) {
obj->attr().apply([](StrName _, PyVar obj, void* userdata) {
VM* vm = (VM*)userdata;
if(obj.is_ptr) vm->__obj_gc_mark((obj).get());
});
}, vm);
}
}
@ -602,16 +603,16 @@ PyVar VM::__py_exec_internal(const CodeObject_& code, PyVar globals, PyVar local
if(globals_dict) {
globals_dict->clear();
globals_obj->attr().apply([&](StrName k, PyVar v) {
for(auto [k, v]: globals_obj->attr().items()){
globals_dict->set(vm, VAR(k.sv()), v);
});
}
}
if(locals_dict) {
locals_dict->clear();
locals_closure->apply([&](StrName k, PyVar v) {
for(auto [k, v]: locals_closure->items()){
locals_dict->set(vm, VAR(k.sv()), v);
});
}
}
return retval;
}
@ -1024,7 +1025,7 @@ void VM::__prepare_py_call(PyVar* buffer, ArgsView args, ArgsView kwargs, const
for(int j = 0; j < kwargs.size(); j += 2) {
StrName key(_CAST(uint16_t, kwargs[j]));
int index = decl->kw_to_index.try_get_likely_found(key);
int index = decl->kw_to_index.get(key, -1);
// if key is an explicit key, set as local variable
if(index >= 0) {
buffer[index] = kwargs[j + 1];
@ -1844,9 +1845,10 @@ void VM::__breakpoint() {
void Function::_gc_mark(VM* vm) const {
decl->_gc_mark(vm);
if(_closure) {
_closure->apply([=](StrName _, PyVar obj) {
_closure->apply([](StrName _, PyVar obj, void* userdata) {
VM* vm = (VM*)userdata;
vm->obj_gc_mark(obj);
});
}, vm);
}
}

169
src/objects/namedict.cpp Normal file
View File

@ -0,0 +1,169 @@
#include "pocketpy/objects/namedict.hpp"
#include <stdexcept>
namespace pkpy {
#define HASH_PROBE_1(key, ok, i) \
ok = false; \
i = key.index & _mask; \
while(!_items[i].first.empty()) { \
if(_items[i].first == (key)) { \
ok = true; \
break; \
} \
i = (i + 1) & _mask; \
}
#define HASH_PROBE_0 HASH_PROBE_1
void NameDict::_set_capacity_and_alloc_items(uint16_t val) {
_capacity = val;
_critical_size = val * _load_factor;
_mask = val - 1;
_items = (Item*)std::malloc(_capacity * sizeof(Item));
std::memset(_items, 0, _capacity * sizeof(Item));
}
void NameDict::set(StrName key, PyVar val) {
bool ok;
uint16_t i;
HASH_PROBE_1(key, ok, i);
if(!ok) {
_size++;
if(_size > _critical_size) {
_rehash_2x();
HASH_PROBE_1(key, ok, i);
}
_items[i].first = key;
}
_items[i].second = val;
}
void NameDict::_rehash_2x() {
Item* old_items = _items;
uint16_t old_capacity = _capacity;
_set_capacity_and_alloc_items(_capacity * 2);
for(uint16_t i = 0; i < old_capacity; i++) {
if(old_items[i].first.empty()) continue;
bool ok;
uint16_t j;
HASH_PROBE_1(old_items[i].first, ok, j);
assert(!ok);
_items[j] = old_items[i];
}
std::free(old_items);
}
PyVar NameDict::try_get(StrName key) const {
bool ok;
uint16_t i;
HASH_PROBE_0(key, ok, i);
if(!ok) return nullptr;
return _items[i].second;
}
PyVar* NameDict::try_get_2(StrName key) const {
bool ok;
uint16_t i;
HASH_PROBE_0(key, ok, i);
if(!ok) return nullptr;
return &_items[i].second;
}
PyVar NameDict::try_get_likely_found(StrName key) const {
uint16_t i = key.index & _mask;
if(_items[i].first == key) return _items[i].second;
i = (i + 1) & _mask;
if(_items[i].first == key) return _items[i].second;
return try_get(key);
}
PyVar* NameDict::try_get_2_likely_found(StrName key) const {
uint16_t i = key.index & _mask;
if(_items[i].first == key) return &_items[i].second;
i = (i + 1) & _mask;
if(_items[i].first == key) return &_items[i].second;
return try_get_2(key);
}
bool NameDict::del(StrName key) {
bool ok;
uint16_t i;
HASH_PROBE_0(key, ok, i);
if(!ok) return false;
_items[i].first = StrName();
_items[i].second = nullptr;
_size--;
// tidy
uint16_t pre_z = i;
uint16_t z = (i + 1) & _mask;
while(!_items[z].first.empty()) {
uint16_t h = _items[z].first.index & _mask;
if(h != i) break;
std::swap(_items[pre_z], _items[z]);
pre_z = z;
z = (z + 1) & _mask;
}
return true;
}
bool NameDict::contains(StrName key) const {
bool ok;
uint16_t i;
HASH_PROBE_0(key, ok, i);
return ok;
}
PyVar NameDict::operator[] (StrName key) const {
PyVar* val = try_get_2_likely_found(key);
if(val == nullptr) { throw std::runtime_error(_S("NameDict key not found: ", key.escape()).str()); }
return *val;
}
array<StrName> NameDict::keys() const {
array<StrName> v(_size);
int j = 0;
for(uint16_t i = 0; i < _capacity; i++) {
if(_items[i].first.empty()) continue;
new (&v[j++]) StrName(_items[i].first);
}
return v;
}
array<NameDict::Item> NameDict::items() const {
array<Item> v(_size);
int j = 0;
for(uint16_t i = 0; i < _capacity; i++) {
if(_items[i].first.empty()) continue;
new (&v[j++]) Item(_items[i].first, _items[i].second);
}
return v;
}
void NameDict::clear() {
std::memset(_items, 0, _capacity * sizeof(Item));
_size = 0;
}
void NameDict::apply (void (*f)(StrName, PyVar, void*), void* data) {
for(uint16_t i = 0; i < _capacity; i++) {
if(_items[i].first.empty()) continue;
f(_items[i].first, _items[i].second, data);
}
}
#undef HASH_PROBE_0
#undef HASH_PROBE_1
PyVar PyObject::attr(StrName name) const {
assert(is_attr_valid());
return (*_attr)[name];
}
NameDict* PyObject::__init_namedict(float lf) {
return new NameDict(lf);
}
} // namespace pkpy