[Python-checkins] CVS: python/dist/src/Python bltinmodule.c,2.210,2.211

[Fredrik Lundh]

Update of /cvsroot/python/python/dist/src/Python
In directory usw-pr-cvs1:/tmp/cvs-serv30924/python

Modified Files:
	bltinmodule.c 
Log Message:


more unicode tweaks: make unichr(0xdddddddd) behave like u"\Udddddddd"
wrt surrogates.  (this extends the valid range from 65535 to 1114111)


Index: bltinmodule.c
===================================================================
RCS file: /cvsroot/python/python/dist/src/Python/bltinmodule.c,v
retrieving revision 2.210
retrieving revision 2.211
diff -C2 -r2.210 -r2.211
*** bltinmodule.c	2001/06/26 17:46:10	2.210
--- bltinmodule.c	2001/06/26 20:01:56	2.211
***************
*** 309,323 ****
  {
  	long x;
! 	Py_UNICODE s[1];
  
  	if (!PyArg_ParseTuple(args, "l:unichr", &x))
  		return NULL;
! 	if (x < 0 || x >= 65536) {
  		PyErr_SetString(PyExc_ValueError,
! 				"unichr() arg not in range(65536)");
  		return NULL;
  	}
- 	s[0] = (Py_UNICODE)x;
- 	return PyUnicode_FromUnicode(s, 1);
  }
  
--- 309,334 ----
  {
  	long x;
! 	Py_UNICODE s[2];
  
  	if (!PyArg_ParseTuple(args, "l:unichr", &x))
  		return NULL;
! 
! 	if (x < 0 || x > 0x10ffff) {
  		PyErr_SetString(PyExc_ValueError,
! 				"unichr() arg not in range(0x10ffff)");
  		return NULL;
+ 	}
+ 
+ 	if (x <= 0xffff) {
+ 		/* UCS-2 character */
+ 		s[0] = (Py_UNICODE) x;
+ 		return PyUnicode_FromUnicode(s, 1);
+ 	} else {
+ 		/* UCS-4 character.  store as two surrogate characters */
+ 		x -= 0x10000L;
+ 		s[0] = 0xD800 + (Py_UNICODE) (x >> 10);
+ 		s[1] = 0xDC00 + (Py_UNICODE) (x & 0x03FF);
+ 		return PyUnicode_FromUnicode(s, 2);
  	}
  }
  
***************
*** 325,329 ****
  "unichr(i) -> Unicode character\n\
  \n\
! Return a Unicode string of one character with ordinal i; 0 <= i < 65536.";
  
  
--- 336,340 ----
  "unichr(i) -> Unicode character\n\
  \n\
! Return a Unicode string of one character with ordinal i; 0 <= i <= 0x10ffff.";