#1477: ur'\U0010FFFF' used to raise in narrow unicode builds.
Corrected the raw-unicode-escape codec to use UTF-16 surrogates in
this case, like the unicode-escape codec does.
Backport of r61793 and r61853
diff --git a/Objects/unicodeobject.c b/Objects/unicodeobject.c
index c749ac5..e2f1ed3 100644
--- a/Objects/unicodeobject.c
+++ b/Objects/unicodeobject.c
@@ -2273,8 +2273,22 @@
else
x += 10 + c - 'A';
}
-#ifndef Py_UNICODE_WIDE
- if (x > 0x10000) {
+ if (x <= 0xffff)
+ /* UCS-2 character */
+ *p++ = (Py_UNICODE) x;
+ else if (x <= 0x10ffff) {
+ /* UCS-4 character. Either store directly, or as
+ surrogate pair. */
+#ifdef Py_UNICODE_WIDE
+ *p++ = (Py_UNICODE) x;
+#else
+ x -= 0x10000L;
+ *p++ = 0xD800 + (Py_UNICODE) (x >> 10);
+ *p++ = 0xDC00 + (Py_UNICODE) (x & 0x03FF);
+#endif
+ } else {
+ endinpos = s-starts;
+ outpos = p-PyUnicode_AS_UNICODE(v);
if (unicode_decode_call_errorhandler(
errors, &errorHandler,
"rawunicodeescape", "\\Uxxxxxxxx out of range",
@@ -2282,8 +2296,6 @@
(PyObject **)&v, &outpos, &p))
goto onError;
}
-#endif
- *p++ = x;
nextByte:
;
}
@@ -2337,6 +2349,32 @@
*p++ = hexdigit[ch & 15];
}
else
+#else
+ /* Map UTF-16 surrogate pairs to '\U00xxxxxx' */
+ if (ch >= 0xD800 && ch < 0xDC00) {
+ Py_UNICODE ch2;
+ Py_UCS4 ucs;
+
+ ch2 = *s++;
+ size--;
+ if (ch2 >= 0xDC00 && ch2 <= 0xDFFF) {
+ ucs = (((ch & 0x03FF) << 10) | (ch2 & 0x03FF)) + 0x00010000;
+ *p++ = '\\';
+ *p++ = 'U';
+ *p++ = hexdigit[(ucs >> 28) & 0xf];
+ *p++ = hexdigit[(ucs >> 24) & 0xf];
+ *p++ = hexdigit[(ucs >> 20) & 0xf];
+ *p++ = hexdigit[(ucs >> 16) & 0xf];
+ *p++ = hexdigit[(ucs >> 12) & 0xf];
+ *p++ = hexdigit[(ucs >> 8) & 0xf];
+ *p++ = hexdigit[(ucs >> 4) & 0xf];
+ *p++ = hexdigit[ucs & 0xf];
+ continue;
+ }
+ /* Fall through: isolated surrogates are copied as-is */
+ s--;
+ size++;
+ }
#endif
/* Map 16-bit characters to '\uxxxx' */
if (ch >= 256) {