Tests for Py_UNICODE* string support.

tests/errors/charptr_from_temp.pyx

 # mode: error
-# tag: werror, charptr, conversion, temp
+# tag: werror, charptr, conversion, temp, py_unicode_strings
 
 cdef bytes c_s = b"abc"
 s = b"abc"
@@ -18,7 +18,28 @@ cptr = s
 # temp => error
 cptr = s + b"cba"
 
+
+cdef unicode  c_u = u"abc"
+u = u"abc"
+
+cdef Py_UNICODE* cuptr
+
+# constant => ok
+cuptr = u"xyz"
+
+# global cdef variable => ok
+cuptr = c_u
+
+# pyglobal => warning
+cuptr = u
+
+# temp => error
+cuptr = u + u"cba"
+
+
 _ERRORS = """
-16:8: Obtaining char* from externally modifiable global Python value
-19:9: Obtaining char* from temporary Python value
+16:8: Obtaining 'char *' from externally modifiable global Python value
+19:9: Obtaining 'char *' from temporary Python value
+34:9: Obtaining 'Py_UNICODE *' from externally modifiable global Python value
+37:10: Obtaining 'Py_UNICODE *' from temporary Python value
 """

tests/errors/e_strcoerce.pyx

@@ -15,5 +15,5 @@ _ERRORS = """
 4:14: Only single-character string literals can be coerced into ints.
 5:14: Only single-character string literals can be coerced into ints.
 8:15: Only single-character string literals can be coerced into ints.
-11:14: Unicode literals do not support coercion to C types other than Py_UNICODE or Py_UCS4.
+11:14: Unicode literals do not support coercion to C types other than Py_UNICODE/Py_UCS4 (for characters) or Py_UNICODE* (for strings).
 """

tests/errors/string_assignments.pyx

 # mode: error
 # coding: ASCII
+# tag: py_unicode_strings
 
 # ok:
 cdef char* c1   =  "abc"
 cdef str s1     =  "abc"
 
 cdef unicode u1 = u"abc"
+cdef Py_UNICODE* cu1 = u1
 
 cdef bytes b1 = b"abc"
 cdef char* c2 = b"abc"
@@ -21,12 +23,18 @@ o4 = c1
 o5 = b1
 o6 = s1
 o7 = u1
+o8 = cu1
 
 # errors:
 cdef char* c_f1   = u"abc"
 cdef char* c_f2   = u1
 cdef char* c_f3   = s1
 
+cdef Py_UNICODE* cu_f1 = c1
+cdef Py_UNICODE* cu_f2 = b1
+cdef Py_UNICODE* cu_f3 = s1
+cdef Py_UNICODE* cu_f4 = b"abc"
+
 cdef bytes b_f1   = u"abc"
 cdef bytes b_f2   = u1
 cdef bytes b_f3   = s1
@@ -56,31 +64,36 @@ print <unicode>c1
 print <unicode>c1[1:2]
 
 _ERRORS = u"""
-26:20: Unicode literals do not support coercion to C types other than Py_UNICODE or Py_UCS4.
-27:22: Unicode objects do not support coercion to C types.
-28:22: 'str' objects do not support coercion to C types (use 'bytes'?).
-
-30:20: Cannot convert Unicode string to 'bytes' implicitly, encoding required.
-31:22: Cannot convert Unicode string to 'bytes' implicitly, encoding required.
-32:22: Cannot convert 'str' to 'bytes' implicitly. This is not portable.
-
-34:17: Cannot convert 'bytes' object to str implicitly. This is not portable to Py3.
-35:19: Cannot convert 'bytes' object to str implicitly. This is not portable to Py3.
-36:17: Cannot convert Unicode string to 'str' implicitly. This is not portable and requires explicit encoding.
-37:19: Cannot convert Unicode string to 'str' implicitly. This is not portable and requires explicit encoding.
-
-39:20: str objects do not support coercion to unicode, use a unicode string literal instead (u'')
-40:22: str objects do not support coercion to unicode, use a unicode string literal instead (u'')
-41:20: Cannot convert 'bytes' object to unicode implicitly, decoding required
-42:22: Cannot convert 'bytes' object to unicode implicitly, decoding required
-43:22: Cannot convert 'char*' to unicode implicitly, decoding required
-
-45:19: Cannot assign type 'str object' to 'tuple object'
-46:18: Cannot assign type 'unicode object' to 'tuple object'
-47:18: Cannot assign type 'bytes object' to 'tuple object'
-
-53:13: default encoding required for conversion from 'char *' to 'str object'
-54:13: default encoding required for conversion from 'char *' to 'str object'
-55:17: Cannot convert 'char*' to unicode implicitly, decoding required
-56:17: default encoding required for conversion from 'char *' to 'unicode object'
+29:20: Unicode literals do not support coercion to C types other than Py_UNICODE/Py_UCS4 (for characters) or Py_UNICODE* (for strings).
+30:22: Unicode objects only support coercion to Py_UNICODE*.
+31:22: 'str' objects do not support coercion to C types (use 'bytes'?).
+
+33:27: Cannot assign type 'char *' to 'Py_UNICODE *'
+34:27: Cannot convert 'bytes' object to Py_UNICODE*, use 'unicode'.
+35:27: 'str' objects do not support coercion to C types (use 'unicode'?).
+36:25: Cannot convert 'bytes' object to Py_UNICODE*, use 'unicode'.
+
+38:20: Cannot convert Unicode string to 'bytes' implicitly, encoding required.
+39:22: Cannot convert Unicode string to 'bytes' implicitly, encoding required.
+40:22: Cannot convert 'str' to 'bytes' implicitly. This is not portable.
+
+42:17: Cannot convert 'bytes' object to str implicitly. This is not portable to Py3.
+43:19: Cannot convert 'bytes' object to str implicitly. This is not portable to Py3.
+44:17: Cannot convert Unicode string to 'str' implicitly. This is not portable and requires explicit encoding.
+45:19: Cannot convert Unicode string to 'str' implicitly. This is not portable and requires explicit encoding.
+
+47:20: str objects do not support coercion to unicode, use a unicode string literal instead (u'')
+48:22: str objects do not support coercion to unicode, use a unicode string literal instead (u'')
+49:20: Cannot convert 'bytes' object to unicode implicitly, decoding required
+50:22: Cannot convert 'bytes' object to unicode implicitly, decoding required
+51:22: Cannot convert 'char*' to unicode implicitly, decoding required
+
+53:19: Cannot assign type 'str object' to 'tuple object'
+54:18: Cannot assign type 'unicode object' to 'tuple object'
+55:18: Cannot assign type 'bytes object' to 'tuple object'
+
+61:13: default encoding required for conversion from 'char *' to 'str object'
+62:13: default encoding required for conversion from 'char *' to 'str object'
+63:17: Cannot convert 'char*' to unicode implicitly, decoding required
+64:17: default encoding required for conversion from 'char *' to 'unicode object'
 """

tests/run/py_unicode_strings.pyx

+# tag: py_unicode_strings
+
+cimport cython
+from libc.string cimport memcpy, strcpy
+
+cdef bint Py_UNICODE_equal(const Py_UNICODE* u1, const Py_UNICODE* u2):
+    while u1[0] != 0 and u2[0] != 0 and u1[0] == u2[0]:
+        u1 += 1
+        u2 += 1
+    return u1[0] == u2[0]
+
+
+ctypedef Py_UNICODE* LPWSTR
+
+cdef unicode uobj = u'unicode\u1234'
+cdef unicode uobj1 = u'u'
+cdef Py_UNICODE* c_pu_str = u"unicode\u1234"
+cdef Py_UNICODE c_pu_arr[42]
+cdef LPWSTR c_wstr = u"unicode\u1234"
+cdef Py_UNICODE* c_pu_empty = u""
+cdef char* c_empty = ""
+
+memcpy(c_pu_arr, c_pu_str, sizeof(Py_UNICODE) * (len(uobj) + 1))
+
+
+def test_c_to_python():
+    """
+    >>> test_c_to_python()
+    """
+    assert c_pu_arr == uobj
+    assert c_pu_str == uobj
+    assert c_wstr == uobj
+
+    assert c_pu_arr[1:] == uobj[1:]
+    assert c_pu_str[1:] == uobj[1:]
+    assert c_wstr[1:] == uobj[1:]
+
+    assert c_pu_arr[:1] == uobj[:1]
+    assert c_pu_arr[:1] == uobj[:1]
+    assert c_pu_str[:1] == uobj[:1]
+    assert c_wstr[:1] == uobj[:1]
+
+    assert c_pu_arr[1:7] == uobj[1:7]
+    assert c_pu_str[1:7] == uobj[1:7]
+    assert c_wstr[1:7] == uobj[1:7]
+
+    assert c_pu_arr[1] == uobj[1]
+    assert c_pu_str[1] == uobj[1]
+    assert c_wstr[1] == uobj[1]
+
+    assert len(c_pu_str) == 8
+    assert len(c_pu_arr) == 8
+    assert len(c_wstr) == 8
+
+    assert sizeof(c_pu_arr) == sizeof(Py_UNICODE) * 42
+    assert sizeof(c_pu_str) == sizeof(void*)
+
+    assert u'unicode'
+    assert not u''
+    assert c_pu_str
+    assert c_pu_empty
+
+
+def test_python_to_c():
+    """
+    >>> test_python_to_c()
+    """
+    cdef unicode u
+
+    assert Py_UNICODE_equal(c_pu_arr, uobj)
+    assert Py_UNICODE_equal(c_pu_str, uobj)
+    assert Py_UNICODE_equal(c_pu_str, <LPWSTR>uobj)
+    u = uobj[1:]
+    assert Py_UNICODE_equal(c_pu_str + 1, u)
+    assert Py_UNICODE_equal(c_wstr + 1, u)
+    u = uobj[:1]
+    assert Py_UNICODE_equal(<Py_UNICODE*>u"u", u)
+    u = uobj[1:7]
+    assert Py_UNICODE_equal(<Py_UNICODE*>u"nicode", u)
+    u = uobj[1]
+    assert Py_UNICODE_equal(<Py_UNICODE*>u"n", u)
+
+    assert len(u"abc\0") == 4
+    assert len(<Py_UNICODE*>u"abc\0") == 3