From 8d93ca138355a85a9dedc2abc225e984ea0e6d29 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Martin=20v=2E=20L=C3=B6wis?= Date: Sat, 23 Nov 2002 22:10:29 +0000 Subject: [PATCH] Verify that the code in CJK UNIFIED IDEOGRAPH- actually denotes an ideograph. --- Modules/unicodedata.c | 15 ++++++++++++--- 1 file changed, 12 insertions(+), 3 deletions(-) diff --git a/Modules/unicodedata.c b/Modules/unicodedata.c index 502b5bd8557..c5ba6fc4b44 100644 --- a/Modules/unicodedata.c +++ b/Modules/unicodedata.c @@ -580,6 +580,15 @@ static char *hangul_syllables[][3] = { { 0, 0, "H" } }; +static int +is_unified_ideograph(Py_UCS4 code) +{ + return ( + (0x3400 <= code && code <= 0x4DB5) || /* CJK Ideograph Extension A */ + (0x4E00 <= code && code <= 0x9FA5) || /* CJK Ideograph */ + (0x20000 <= code && code <= 0x2A6D6));/* CJK Ideograph Extension B */ +} + static int _getucname(Py_UCS4 code, char* buffer, int buflen) { @@ -610,9 +619,7 @@ _getucname(Py_UCS4 code, char* buffer, int buflen) return 1; } - if ((0x3400 <= code && code <= 0x4DB5) || /* CJK Ideograph Extension A */ - (0x4E00 <= code && code <= 0x9FA5) || /* CJK Ideograph */ - (0x20000 <= code && code <= 0x2A6D6)) {/* CJK Ideograph Extension B */ + if (is_unified_ideograph(code)) { if (buflen < 28) /* Worst case: CJK UNIFIED IDEOGRAPH-20000 */ return 0; @@ -743,6 +750,8 @@ _getcode(const char* name, int namelen, Py_UCS4* code) return 0; name++; } + if (!is_unified_ideograph(v)) + return 0; *code = v; return 1; }