Python-vm-class

  • reference: Python 源码剖析
  • Python-vm-class
    • Python中的对象模型
    • 用户自定义class
    • 从class对象到instance对象
    • 访问instance对象中的属性

1.Python虚拟机中的类机制

a.Python中的对象模型

  • Python中的三类对象

type对象:表示Python内置的类型

class对象:表示Python程序员定义的类型

instance对象:表示由class对象创建的实例

  • 对象间的关系

is-kind-of:基类与子类的关系

is-instance-of:类与实例之间的关系

  • (type ‘type’) 和 (type ‘object’)

  • 从type对象到class对象

可调用性,class对象中实现了”call“操作(在Python内部的PyTypeObject中tp_call不为空),则可调用

1
2
3
4
5
6
class A(object):
def __call__(self):
print 'Hello Python'

a = A()
a()
  • 处理基类和type信息
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
[typeobject.c]
int PyType_Ready(PyTypeObject* type){
PyObject* dict, *bases;
PyTypeObject* base;
Py_ssize_t i,n;

//尝试获得type的tp_base中指定基类(super type)
base = type->type_base;
if(base == NULL && type != &PyBaseObject_Type){
base = type->tp_base = &PyBaseObject_Type;
}

//如果基类没有初始化,先初始化基类
if(base && base->tp_dict == NULL){
PyType_Ready(base);
}

//设置type信息
if(type->ob_type == NULL && base != NULL){
type->ob_type = base->ob_type;
}
...
}
  • 处理基类列表
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
[typeobject.c]
int PyType_Ready(PyTypeObject* type){
PyObject* dict, *bases;
PyTypeObject* base;
Py_ssize_t i,n;
...
//尝试获得type的tp_base中指定基类(super type)
base = type->tp_base;
if(base == NULL && type != &PyBaseObject_Type){
base = type->tp_base = &PyBaseObject_Type;
}
...
//处理bases:基类列表
bases = type->tp_bases;
if(bases == NULL){
//如果bases为空,则根据base的情况设定bases
if(base == NULL)
bases = PyTuple_New(0);
else
bases = PyTuple_Pack(1,base);
type->tp_bases = bases;
}
}
  • 填充tp_dict
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
[typeobject.c]
int PyType_Ready(PyTypeObject* type){
PyObject* dict, *bases;
PyTypeObject* base;
Py_ssize_t i,n;
...
//设定tp_dict
dict = type->tp_dict;
if(dict == NULL){
dict = PyDict_New();
type->tp_dict = dict;
}

//将与type相关的descriptor加入到tp_dict中
add_operators(type);
if(type->tp_methods != NULL){
add_methods(type,type->tp_methods)
}
if(type->tp_members != NULL){
add_members(type,type->tp_members)
}
if(type->tp_getset != NULL){
add_getset(type,type->tp_getset);
}
....
}
  • slot与操作排序

slot可以视为表示PyTypeObject中定义的操作

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
[typeobject.c]
typedef struct wrapperbase slotdef;

[descrobject.h]
struct wrapperbase{
char* name; //操作名称,比如"__add__"
int offset; //操作的函数地址在PyHeapTypeObject中的偏移量
void* function; //slot function的函数
wrapperfunc wrapper;
char* doc;
int flags;
PyObject* name_strobj;
};

//定义slot
[typeobject.c]
#define TPSLOT(NAME,SLOT,FUNCTION,WRAPPER,DOC)
{NAME,offsetof(PyTypeObject,SLOT),(void *)(FUNCTION),WRAPPER,PyDoc_STR(DOC)}

#define ETSLOT(NAME,SLOT,FUNCTION,WRAPPER,DOC)
{NAME,offsetof(PyHeapTypeObject,SLOT),(void *)(FUNCTION),WRAPPER,PyDoc_STR(DOC)}

[structmember.h]
#define offsetof(type,member) ((int)&((type*)0)->member)

[object.h]
typedef struct _heaptypeobject {
PyTypeObject ht_type;
PyNumberMethods as_number;
PyMappingMethods as_mapping;
PySequenceMethods as_sequence;
PyBufferProcs as_buffer;
PyObject* ht_name,*ht_slots;
} PyHeapTypeObject; //定于顺序表示优先级

//slot集合:slotdefs
[typeobject.c]
....
#define SQSLOT(NAME,SLOT,FUNCTION,WRAPPER,DOC)
ETSLOT(NAME,as_sequence.SLOT,FUNCTION,WRAPPER,DOC)
....

static slotdef slotdefs[] = {
...
//[不同操作名对应相同的操作]
BINSLOT("__add__",nb_add,slot_nb_add,"+"),
RBINSLOT("__radd__",nb_add,slot_nb_add,"+"),
//相同操作名对应不同操作
SQSLOT("__getitem__",sq_item,slot_sq_item,wrap_sq_item,"x.__getitem__(y)<==>x[y]"),
MPSLOT("__getitem__",mp_subscript,slot_mp_subscript,wrap_binaryfunc,"x.___getitem__(y)<==>x[y]"),
...
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
//slotdefs的排序在init_slotdefs中完成
[typeobject.c]
static void init_slotdefs(void){
slotdefs* p;
static int initialized = 0;
//init_slotdefs只会进行一次
if(initialized)
return
for(p=slotdefs;p->name;p++){
//填充slotdef结构体中的name_strobj
p->name_strobj = PyString_InternFromString(p->name);
}
//对slotdefs中的slotdef进行排序
qsort((void *)slotdefs,(size_t)(p-slotdefs),sizeof(slotdef),slotdef_cmp);
initialized = 1;
}

//slot排序的比较策略
static int slotdef_cmp(const void* aa,const void* bb){
const slotdef* a = (const slotdef *)aa, *b = (const slotdef *)bb;
int c = a->offset - b->offset;
if(c != 0)
return c;
else
return (a > b) ? 1 : (a < b) ? -1 : 0;
}
  • 从slot到descriptor
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
[descrobject.h]
#define PyDescr_COMMON
PyObject_HEAD
PyTypeObject* d_type;
PyObject* d_name

typedef struct{
PyDescr_COMMON;
struct wrapperbase* d_base;
void* d_wrapped; /* This can be any function pointer */
} PyWrapperDescrObject;

[descrobject.c]
static PyDescrObject* descr_new(PyTypeObject* descrtype, PyTypeObject* type, char* name){
PyDescrObject* descr;
//申请空间
descr = (PyDescrObject *)PyType_GenericAlloc(descrtype,0);
descr->d_type = type;
descr->d_name = PyString_InternFromString(name);
return descr;
}

PyObject* PyDescr_NewWrapper(PyTypeObject* type,struct wrapperbase* base, void* wrapped){
PyWrapperDescrObject* descr;
descr = descr_new(&PyWrapperDescr_Type,type,base->name);
descr->d_base = base;
descr->d_wrapped = wrapped;
return (PyObject *)descr;
}
  • 建立联系

排序后的结果放在slotdefs中,Python虚拟机可以从头到尾遍历slotdefs,基于每一个slot建立一个descriptor,然后在tp_dict中建立从操作名到descriptor的关联

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
[typeobject.c]
static int add_operators(PyTypeObject* type){
PyObject* dict = type->tp_dict;
slotdef* p;
PyObject* descr;
void **ptr;
//对slotdefs进行排序
init_slotdefs();
for(p = slotdefs; p->name; p++){
//如果slot中没有指定wrapper,则不处理
if(p->wrapper == NULL)
continue;
//获得slot对应的操作在PyTypeObject中的函数指针
ptr = slotptr(type,p->offset);
//如果tp_dict中已经存在操作名,则放弃
if(PyDict_GetItem(dict,p->name_strobj))
continue;
//创建descriptor
descr = PyDescr_NewWrapper(type,p,*ptr);
//将(操作名,descriptor)放入tp_dict中
PyDict_SetItem(dict,p->name_strobj,descr);
}
return 0;
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
//slotptr完成转换
[typeobject.c]
static void** slotptr(PyTypeObject* type,int offset){
char *ptr;

//判断从PyHeapTypeObject中排在后面的PySequenceMethods开始
if(offset >= offsetof(PyHeapTypeObject,as_sequence)){
ptr = (void *)type->tp_as_sequence;
offset -= offsetof(PyHeapTypeObject,as_sequence);
}
else if(offset >= offsetof(PyHeapTypeObject,as_mapping)){
ptr = (void *)type->tp_as_mapping;
offset -= offsetof(PyHeapTypeObject,as_mapping);
}
else if(offset >= offsetof(PyHeapTypeObject,as_number)){
ptr = (void *)type->tp_as_number;
offset -= offsetof(PyHeapTypeObject, as_number);
}
else{
ptr = (void *)type;
}

if( ptr != NULL )
ptr += offset;

return (void **)ptr;
}

add_operator_pylist

  • repr重写
1
2
3
4
5
6
class A(list):
def __repr__(self):
return 'Python'

s = '%s' % A()
print s
1
2
3
4
5
6
7
8
9
[typeobject.c]
static PyObject* slot_tp_repr(PyObject* self){
PyObject* func, *res;
static PyObject* repr_str;
//查找"__repr__"属性
func = lookup_method(self,"__repr__",&repr_str);
//调用"__repr__"对象的对象
res = PyEval_CallObject(func,NULL);
}

initial_class_A

  • 确定MRO(Method Resolve Order)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
[mro.py]
class A(list):
def show(self):
print "A::show"


class B(list):
def show(value):
print 'B::show'

class C(A):
pass

class D(C,B):
pass

d = D()
d.show()

mro_list

  • 继承基类操作
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
[typeobject.c]
int PyType_Ready(PyTypeObject* type){
...
bases = type->tp_mro;
n = PyTuple_GET_SIZE(bases);
for( i = 1; i < n; i++){
PyObject* b = PyTuple_GET_ITEM(bases,i);
inherit_slots(type,(PyTypeObject *)b);
}
...
}

[typeobject.c]
static void inherit_slots(PyTypeObject* type,PyTypeObject* base){
PyTypeObject* basebase;

#define SLOTDEFINED(SLOT)
(base->SLOT != 0 && (basebase ==NULL || base->SLOT != basebase->SLOT))

#define COPYSLOT(SLOT)
if(!type->SLOT && SLOTDEFINED(SLOT)) type->SLOT = base->SLOT

#define COPYNUM(SLOT) COPYSLOT(tp_as_number->SLOT)
if(type->tp_as_number != NULL && base->tp_as_number != NULL){
basebase = base->tp_base;
if(basebase->tp_as_number == NULL)
basebase = NULL;
COPYNUM(nb_add);
...
}
....
}
  • 填充基类中的子类列表
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
[typeobject.c]
int PyType_Ready(PyTypeObject* type){
PyObject* dict, *bases;
PyTypeObject* base;
Py_ssize_t i,n;
...

//填充基类的子类列表
bases = type->tp_bases;
n = PyTuple_GET_SIZE(bases);
for( i = 0; i < n; i++){
PyObject* b = PyTuple_GET_ITEM(bases,i);
add_subclass((PyTypeObject *)b,type);
}
...
}

c.用户自定义class

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
[class_0.py]
class A(object):
name = 'python'
def __init__(self):
print 'A::__init__'

def f(self):
print 'A::f'

def g(self,aValue):
self.value = aValue
print self.value

a = A()
a.f()
a.g(10)
  • 创建class对象

  • class的动态元信息

1
2
3
4
5
6
7
8
9
10
11
12
[PyCodeObject for class_0.py]
class A(object):
0 LOAD_CONST 0 (A)
3 LOAD_NAME 0 (object)
6 BUILD_TUPLE 1
9 LOAD_CONST 1 (code object for A)
12 MAKE_FUNCTION 0
15 CALL_FUNCTION 0


18 BUILD_CLASS
19 STORE_NAME 1 (A)

make_function

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
[PyCodeObject for class A]
0 LOAD_NAME 0 (__name__)
3 STORE_NAME 1 (__module__)
name = 'Python'
6 LOAD_CONST 0 ('Python')
9 STORE_NAME 2 (name)
def __init__(self):
12 LOAD_CONST 1 (code object for function __init__)
15 MAKE_FUNCTION 0
18 STORE_NAME 3 (__init__)
def f(self):
21 LOAD_CONST 2 (code object for function f)
24 MAKE_FUNCTION 0
27 STORE_NAME 4 (f)
def g(self,aValue):
30 LOAD_CONST 3 (code object for function g)
33 MAKE_FUNCTION 0
36 STORE_NAME 5 (g)

39 LOAD_LOCALS
40 RETURN_VALUE
1
2
3
4
5
[LOAD_LOCALS]
if((x = f->f_locals) != NULL){
PUSH(x);
continue;
}
1
2
3
4
5
6
[CALL_FUNCTION]
PyObject **sp;
sp = stack_pointer;
x = call_function(&sp,oparg);
stack_pointer = sp;
PUSH(x);

call_function_stack

  • metaclass
1
2
3
4
5
6
7
8
9
10
[BUILD_CLASS]
u = TOP(); //class的动态元信息f_locals
v = SECOND(); // class 的基类列表
w = THIRD(); //class的名'A'
STACKADJ(-2);
x = build_class(u,v,w);
SET_TOP(x);
Py_DECREF(u);
Py_DECREF(v);
Py_DECREF(w);

class_A_build_class

  • 获得metaclass
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
[ceval.c]
PyObject* build_class(PyObject* methods, PyObject* bases, PyObject* name){
PyObject* metaclass = NULL,*result, *base;

//检查属性表中是否有指定的_metaclass_
if(PyDict_Check(methods))
metaclass = PyDict_GetItemString(methods,"__metaclass__"); //metaclass为静态元信息,methods为动态元信息
if(metaclass != NULL)
Py_INCREF(metaclass);
else if(PyTuple_Check(bases) && PyTuple_GET_SIZE(bases) > 0){
// 获得A的第一基类,object
base = PyTuple_GET_ITEM(bases,0)
//获得object.__class__
metaclass = PyObject_GetAttrString(base,"__class__");
}
else{
....
}
result = PyObject_CallFunctionObjArgs(metaclass,name,bases,methods,NULL);
...
return result;
}

explain_meta_dynamic

meta_dynamic

  • 调用metaclass
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
[object.h]
typedef PyObject* (*ternaryfunc)(PyObject *, PyObject *, PyObject *);

[abstract.c]
PyObject* PyObject_Call(PyObject* func,PyObject* arg,PyObject* kw){
//arg即是PyObject_CallFunctionObjArgs中打包得到的tuple对象
ternaryfunc call = func->ob_type->tp_call;
PyObject* result = (*call)(func,arg,kw);
return result;
}

[typeobject.c]
type_call(PyTypeObject* type, PyObject* args, PyObject* kwds){
PyObject* obj;

obj = type->tp_new(type,args,kwds);

...//如果创建的是实例对象,则调用"__init__" 进行初始化
return obj;
}

[typeobject.c]
static PyObject* type_new(PyTypeObject* metatype, PyObject* args, PyObject* kwds){
//metatype是PyType_Type<type 'type'>,args中包含了(类名,基类列表,属性表)
PyObject* name,*bases, *dict;
static char* kwlist[] = {"name","bases","dict",0};
PyTypeObject* type,*base,*tmptype,*winner;
PyHeapTypeObject* et;
Py_ssize_t slotoffset;

//将args中的(类名,基类列表,属性表)分别解析到name,bases,dict三个变量中
PyArg_ParseTupleAndKeywords(args,kwds,"SO!O!:type",kwlist,
&name,
&PyTuple_Type,&bases.
&PyDict_Type,&dict);
.....//确定最佳metaclass,存储在PyObject* metatype中
....//确定最佳base,存储在PyObject* base中

//为class对象申请内存
//尽管PyType_Type为0,但PyBaseObject_Type的为PyType_GenericAlloc,
//在PyType_Ready中被继承了
//创建的内存大小为tp_basicsize _ tp_itemsize
type = (PyTypeObject *)metatype->tp_alloc(metatype,nslots);
et = (PyHeapTypeObject *)type;
et->ht_name = name;

//设置PyTypeObject中的各个域
type->tp_as_number = &et->as_number;
type->tp_as_sequence = &et->as_sequence;
type->tp_as_mapping = &et->as_mapping;
type->tp_as_buffer = &et->as_buffer;
type->tp_name = PyString_AS_STRING(name);

//设置基类和基类列表
type->tp_bases = bases;
type->tp_base = base;

//设置属性表
type->tp_dict = dict = PyDict_Copy(dict);

//如果自定义class中重写了__new__,将__new__对应的函数改造为static函数
tmp = PyDict_GetItemString(dict,"__new__");
tmp = PyStaticMethod_New(tmp);
PyDict_SetItemString(dict,"__new__",tmp);

//为class对象对应的instance对象设置内存大小信息
slotoffset = base->basicsize;
type->tp_dictoffset = slotoffset;
slotoffset += sizeof(PyObject *);
type->tp_weaklistoffset = slotoffset;
slotoffset += sizeof(PyObject *);
type->tp_basicsize = slotoffset;
type->tp_itemsize = base->tp_itemsize;
......

//调用PyType_Ready(type)对class对象进行初始化
PyType_Ready(type);
return (PyObject *)type;
}

compare_user_bulitin

d.从class对象到instance对象

1
2
3
4
5
[PyCodeObject for class_0.py]
a = A()
22 LOAD_NAME 1 (A)
25 CALL_FUNCTION 0
28 STORE_NAME 2 (a)

instance_local

创建class对象,Python虚拟机使用的是type_new;而对于instance对象,Python虚拟机则使用object_new

1
2
3
4
5
6
7
8
9
10
[typeobject.c]
type_call(PyTypeObject* type, PyObject* args, PyObject* kwds){
PyObject* obj;

obj = type->tp_new(type,args,kwds);

type = obj->ob_type; //如果创建的是实例对象,则调用"__init__" 进行初始化
type->tp_init(obj,args,kwds);
return obj;
}
1
2
3
4
5
6
7
8
//由于A重写了__init__,所以在fixup_slot_dispatchers中,tp_init会指向slotdefs中指定的与"__init__"对应的slot_tp_init
[typeobject.c]
static int slot_tp_init(PyObject* self, PyObject* args, PyObject* kwds){
static PyObject* init_str;
PyObject* meth = lookup_method(self,"__init__",&init_str);
PyObject_Call(meth,args,kwds);
return 0;
}

class_to_instance

e.访问instance对象中的属性

1
2
3
4
5
6
[PyCodeObject for class_0.py]
a.f()
31 LOAD_NAME 2 (a)
34 LOAD_ATTR 3 (f)
37 CALL_FUNCTION 0
40 POP_TOP
1
2
3
4
5
6
[LOAD_ATTR]
w = GETITEM(names,oparg); //PyStringObject对象"f"
v = TOP(); //instance对象
x = PyObject_GetAttr(v,w);
Py_DECREF(v);
SET_TOP(x);
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
[object.c]
PyObject* PyObject_GetAttr(PyObject* v, PyObject* name){
PyTypeObject* tp = v->ob_type;
//通过tp_getattro获得属性对应对象
if(tp->tp_getattro != NULL) //优先调用
return (*tp->tp_getattro)(v,name);

//通过tp_getattr获得属性对应对象
if(tp->tp_getattr != NULL)
return (*tp->tp_getattr)(v,PyString_AS_STRING(name));

//属性不存在,排除异常
PyErr_Format(PyExc_AttributeError, "'%.50s' object has no attribute '%.400s'",tp->tp_name,PyString_AS_STRING(name));
return NULL;
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
#属性访问算法
#首先寻找'f'对应的descriptor
#注意:hasattr会在<class A>的mro列表中寻找符号'f'
if hasattr(A,'f'):
descriptor = A.f

type = descriptor.__class__
if hasattr(type,'__get__') and (hasattr(type,'__set__') or 'f' not in a.__dict__):
return type.__get__(descriptor,a,A)

# 通过descriptor访问失败,在instance对象自身__dict__中寻找属性
if 'f' in a.__dict__:
return a.__dict__['f']

#instance对象的__dict__中找不到属性,返回a的基类列表中某个基类里定义的函数
# 注意:这里的descriptor实际指向了一个普通函数
if descriptor:
return descriptor.__get__(descriptor,a,A)
  • instance对象中的dict

a__dict__

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
//PyObject_GenericGetAttr
[object.c]
PyObject* PyObject_GenericGetAttr(PyObject* obj, PyObject* name){
PyTypeObject* tp = obj->ob_type;
PyObject* res = NULL;
Py_ssize_t dictoffset;
PyObject** dictptr;

//inline _PyObject_GetDictPtr函数的代码
dictoffset = tp->tp_dictoffset;
if(dictoffset != 0){
PyObject* dict;
if(dictoffset < 0){
...//处理变长对象
}
dictptr = (PyObject **)((char *)obj + dictoffset);
dict = *dictptr;
res = PyDict_GetItem(dict,name);
}
....
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
//PyObject_GenericSetAttr
[object.c]
int PyObject_GenericSetAttr(PyObject* obj, PyObject* name, PyObject* value){
PyTypeObject* tp = obj->ob_type;
PyObject** dictptr;
....
dictptr = _PyObject_GetDictPtr(obj);
if(dictptr != NULL){
PyObject* dict = *dictptr;
if(dict == NULL && value != NULL){
//这里创建了instance对象中的__dict__
dict = PyDict_New();
*dictptr = dict;
}
....
}
....
}
  • 再论descriptor

一般而言,对于一个Python中的对象obj,如果obj.class 对应的class对象中存在getsetdelete三种操作,那么obj为Python的一个descriptor

1
2
3
4
5
[slotdefs in typeobject.c]
.....
TPSLOT("__get__",tp_descr_get,...),
TPSLOT("__set__",tp_descr_set,....),
TPSLOT("__delete__",tp_descr_set,....)

如果细分,那么descriptor还可以分为如下两种:

  • data descriptor: type中定义了getset的descriptor

  • non data descriptor: type中只定义了get的descriptor

get_attr

如果待访问的属性是一个descriptor,若它存在于class对象的tp_dict中,会调用其get方法;若它存在于instance对象的tp_dict中,则不会调用其get方法

descriptor

  • 函数变身
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
//PyFunction_Type中,"__get__"对应的tp_descr_get被设置成了&func_descr_get,意味着A.f实际上是一个descriptor
//又由于PyFunction_Type并没有设置tp_descr_set,所以A.f是一个non data descriptor
//并且a.__dict__中没有符号'f'的存在,所以a.f的返回值将被descriptor改变,结果将是A.f.__get__,也就是func_descr_get(A.f,a,A)
[funcobject.c]
/* Bind a function to an object */
static PyObject* func_descr_get(PyObject* func,PyObject* obj, PyObject* type){
if(obj == Py_None)
obj = NULL;
return PyMethod_New(func,obj,type);
}

[classobject.c]
PyObject* PyMethod_New(PyObject* func, PyObject* self, PyObject* class){
register PyMethodObject* im;
im = free_list;
if(im != NULL){
//使用缓冲池
free_list = (PyMethodObject *)(im->im_self);
PyObject_INIT(im,&PyMethod_Type);
}
else{
//不使用缓冲池,直接创建PyMethodObject
im = PyObject_GC_New(PyMethodObject,&PyMethod_Type);
}

im->im_weakreflist = NULL;
im->im_func = func;
//这里就是"self" ~~~!!!
im->im_self = self;
im->im_class = class;
_PyObject_GC_TRACK(im);
return (PyObject *)im;
}
1
2
3
4
5
6
7
8
9
//PyMethodObject
[classobject.h]
typedef struct {
PyObject_HEAD
PyObject* im_func; //可调用的PyFunctionObject对象,'f'
PyObject* im_self; //用于成员函数调用的self参数,instance对象(a)
PyObject* im_class; //class对象(A)
PyObject* im_weakreflist;
} PyMethodObject;
  • 无参函数的调用
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
[ceval.c]
static PyObject* call_function(PyObject** pp_stack, int oparg){
int na = oparg & 0xff;
int nk = (oparg>>8) & 0xff;
int n = na + 2*nk;
PyObject** pfunc = (*pp_stack) - n - 1;
PyObject* func = *pfunc;
PyObject* x, *w;

if(PyCFunction_Check(func) && nk == 0){
....
}else{
//从PyMethodObject对象中抽取PyFunctionObject对象和self参数
if(PyMethod_Check(func) && PyMethod_GET_SELF(func) != NULL){
PyObject* self = PyMethod_GET_SELF(func);
func = PyMethod_GET_FUNCTION(func);
//self参数入栈,调整参数信息变量
*pfunc = self;
na++;
n++;
}
if(PyFunction_Check(func))
x = fast_function(func,pp_stack,n,na,nk);
else
x = do_call(func,pp_stack,na,nk);
}
....
return x;
}

self

  • 带参函数的调用

  • Bound Method(a.f) 和 Unbound Method(A.f)

本质区别在于PyFunctionObject有没有与instance对象绑定在PyMethodObject中, Bound Method 完成了绑定动作,而Unbound Method 没有完成绑定动作

  • 千变万化的descriptor

  • 用descriptor实现static method

1
2
3
4
5
[class_2.py]
class A(object):
def g(value):
print value
g = staticmethod(g) #staticmethod存在于Python启动并进行初始化设置的builtin名字空间中
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
[funcobject.c]
typedef struct {
PyObject_HEAD
PyObject* sm_callable;
} staticmethod;

[funcobject.c]
static int sm_init(PyObject* self,PyObject* args, PyObject* kwds){
staticmethod* sm = (staticmethod *)self;
PyObject* callable;

PyArg_UnpackTuple(args,"staticmethod",1,1,&callable);
sm->sm_callable = callable;
return 0;
}

PyStaticMethod_Type,创建的staticmethod实际上也是一个descriptor,在PyStaticMethod_Type中,tp_descr_get指向了sm_descr_get

1
2
3
4
5
[funcobject.c]
static PyObject* sm_descr_get(PyObject* self, PyObject* obj, PyObject* type){
staticmethod* sm = (staticmethod *)self;
return sm->sm_callable;
}
Share