From 9dcccfcba5dcb33d3271ac8681f0c65a0a6d44f5 Mon Sep 17 00:00:00 2001
From: Pavel Charvat <pavel.charvat@netcentrum.cz>
Date: Thu, 15 Mar 2007 21:56:46 +0100
Subject: [PATCH] libucw: added UTF-8 correctness checker

---
 lib/unicode-utf8.c | 29 +++++++++++++++++++++++++++++
 lib/unicode.h      |  1 +
 2 files changed, 30 insertions(+)

diff --git a/lib/unicode-utf8.c b/lib/unicode-utf8.c
index 5f7b8186..8c8ac1bd 100644
--- a/lib/unicode-utf8.c
+++ b/lib/unicode-utf8.c
@@ -36,6 +36,35 @@ utf8_strnlen(byte *str, uns n)
   return len;
 }
 
+uns
+utf8_check(byte *s)
+{
+#define UTF8_CHECK_NEXT if (unlikely((*s & 0xc0) != 0x80)) goto bad; s++
+  while (*s)
+    {
+      uns u = *s++;
+      if (u < 0x80)
+	;
+      else if (unlikely(u < 0xc0))
+        {
+bad:
+	  return 0;
+	}
+      else if (u < 0xe0)
+        {
+	  UTF8_CHECK_NEXT;
+	}
+      else if (likely(u < 0xf0))
+        {
+	  UTF8_CHECK_NEXT;
+	  UTF8_CHECK_NEXT;
+	}
+      else
+	goto bad;
+    }
+  return 1;
+}
+
 #ifdef TEST
 #include <string.h>
 #include <stdio.h>
diff --git a/lib/unicode.h b/lib/unicode.h
index 79934540..5e8dac8f 100644
--- a/lib/unicode.h
+++ b/lib/unicode.h
@@ -204,5 +204,6 @@ utf8_encoding_len(uns c)
 
 uns utf8_strlen(byte *str);
 uns utf8_strnlen(byte *str, uns n);
+uns utf8_check(byte *str);
 
 #endif
-- 
2.39.2