python源码学习（七）——String对象

2019独角兽企业重金招聘Python工程师标准>>>

http://www.androiddev.net/python7/

python源码学习（七）——String对象

By 李 志豪 | 2014/07/05

0 Comment

有了前面对Python中整数对象的学习，再学习String对象就回容易理解些，首先我们来看一下Python中的PyStringObject和PySting_Type，PyStringObject.h对象的定义如下：

1	typedef struct {

2	PyObject_VAR_HEAD

3	long ob_shash;

4	int ob_sstate;

5	char ob_sval[1];

6

7	/* Invariants:

8	*     ob_sval contains space for 'ob_size+1' elements.

9	*     ob_sval[ob_size] == 0.

10	*     ob_shash is the hash of the string or -1 if not computed yet.

11	*     ob_sstate != 0 iff the string object is in stringobject.c's

12	*       'interned' dictionary; in this case the two references

13	*       from 'interned' to this object are *not counted* in ob_refcnt.

14

*/

15	} PyStringObject;

PyObject_VAR_HEAD中有个ob_size变量，字符串的长度由ob_size决定，ob_size决定了这段内存的实际长度（字节），这个机制是Python中所有变长对象的实现机制。Python中字符串都是以结尾。ob_shash的初始值是-1，这个在Python的dict中有非常巨大的作用。hash值得算法如下：

1	static long

2	string_hash(PyStringObject *a)

3

{

4	register Py_ssize_t len;

5	register unsigned char *p;

6	register long x;

7

8	#ifdef Py_DEBUG

9	assert(_Py_HashSecret_Initialized);

10

#endif

11	if (a->ob_shash != -1)

12	return a->ob_shash;

13	len = Py_SIZE(a);

14

/*

15	We make the hash of the empty string be 0, rather than using

16	(prefix ^ suffix), since this slightly obfuscates the hash secret

17

*/

18	if (len == 0) {

19	a->ob_shash = 0;

20

return 0;

21

}

22	p = (unsigned char *) a->ob_sval;

23	x = _Py_HashSecret.prefix;

24	x ^= *p << 7;

25	while (--len >= 0)

26	x = (1000003*x) ^ *p++;

27	x ^= Py_SIZE(a);

28	x ^= _Py_HashSecret.suffix;

29	if (x == -1)

30

x = -2;

31	a->ob_shash = x;

32

return x;

33

}

字符串的预存hash值和intern机制使得Python的执行效率提升了20%左右。再看一下PyString_Type，代码如下：

1	PyTypeObject PyString_Type = {

2	PyVarObject_HEAD_INIT(&PyType_Type, 0)

3

"str",

4	PyStringObject_SIZE,

5	sizeof(char),

6	string_dealloc,                             /* tp_dealloc */

7	(printfunc)string_print,                    /* tp_print */

8	0,                                          /* tp_getattr */

9	0,                                          /* tp_setattr */

10	0,                                          /* tp_compare */

11	string_repr,                                /* tp_repr */

12	&string_as_number,                          /* tp_as_number */

13	&string_as_sequence,                        /* tp_as_sequence */

14	&string_as_mapping,                         /* tp_as_mapping */

15	(hashfunc)string_hash,                      /* tp_hash */

16	0,                                          /* tp_call */

17	string_str,                                 /* tp_str */

18	PyObject_GenericGetAttr,                    /* tp_getattro */

19	0,                                          /* tp_setattro */

20	&string_as_buffer,                          /* tp_as_buffer */

21	Py_TPFLAGS_DEFAULT | Py_TPFLAGS_CHECKTYPES |

22	Py_TPFLAGS_BASETYPE | Py_TPFLAGS_STRING_SUBCLASS |

23	Py_TPFLAGS_HAVE_NEWBUFFER,              /* tp_flags */

24	string_doc,                                 /* tp_doc */

25	0,                                          /* tp_traverse */

26	0,                                          /* tp_clear */

27	(richcmpfunc)string_richcompare,            /* tp_richcompare */

28	0,                                          /* tp_weaklistoffset */

29	0,                                          /* tp_iter */

30	0,                                          /* tp_iternext */

31	string_methods,                             /* tp_methods */

32	0,                                          /* tp_members */

33	0,                                          /* tp_getset */

34	&PyBaseString_Type,                         /* tp_base */

35	0,                                          /* tp_dict */

36	0,                                          /* tp_descr_get */

37	0,                                          /* tp_descr_set */

38	0,                                          /* tp_dictoffset */

39	0,                                          /* tp_init */

40	0,                                          /* tp_alloc */

41	string_new,                                 /* tp_new */

42	PyObject_Del,                               /* tp_free */

43

};

从上面的代码我们不难看出，String对象的tp_as_number、tp_as_sequence、tp_as_mapping三个域都被设置了，这说明了PyStringObject对数值操作
、序列操作和映射操作都支持。接下来我们看一下图片：



好了，通过以上两张图片我们就可以比较清楚的了解Python中string对象运行时的状态以及所占内存了，以后将会讲一个非常重要的机制，就是string的intern机制。

转载于:https://my.oschina.net/tplinuxhyh/blog/798309