From: Pavel Charvat <pavel.charvat@netcentrum.cz>
Date: Fri, 23 Nov 2007 09:38:03 +0000 (+0100)
Subject: ucwlib: Added a function for checking "regular" unicode values.
X-Git-Tag: holmes-import~499^2~2
X-Git-Url: http://mj.ucw.cz/gitweb/?a=commitdiff_plain;h=647552e493529551c01cb764b26ee57cfc5a9eb7;p=libucw.git

ucwlib: Added a function for checking "regular" unicode values.
---

diff --git a/lib/unicode.h b/lib/unicode.h
index 52ea5e68..8f8fe0e2 100644
--- a/lib/unicode.h
+++ b/lib/unicode.h
@@ -208,6 +208,17 @@ utf8_encoding_len(uns c)
   return 6;
 }
 
+static uns
+unicode_sanitize_char(uns u)
+{
+  if (u >= 0x10000 ||			// We don't accept anything outside the basic plane
+      u >= 0xd800 && u < 0xf900 ||	// neither we do surrogates
+      u >= 0x80 && u < 0xa0 ||		// nor latin-1 control chars
+      u < 0x20)
+    return UNI_REPLACEMENT;
+  return u;
+}
+
 /* unicode-utf8.c */
 
 uns utf8_strlen(const byte *str);