Author: abock
Date: 2005-11-27 22:06:35 -0500 (Sun, 27 Nov 2005)
New Revision: 53534
Modified:
trunk/entagged-sharp/ChangeLog
trunk/entagged-sharp/src/EncodingInfo.cs
trunk/entagged-sharp/src/Util/Utils.cs
Log:
2005-11-27 Aaron Bockover <[EMAIL PROTECTED]>
* src/Util/Utils.cs: Added the UnicodeValidator class here
* src/EncodingInfo.cs: Ugh, reversed the previous patch... that class
really doesn't belong in this file ;)
Modified: trunk/entagged-sharp/ChangeLog
===================================================================
--- trunk/entagged-sharp/ChangeLog 2005-11-28 03:02:49 UTC (rev 53533)
+++ trunk/entagged-sharp/ChangeLog 2005-11-28 03:06:35 UTC (rev 53534)
@@ -1,5 +1,11 @@
2005-11-27 Aaron Bockover <[EMAIL PROTECTED]>
+ * src/Util/Utils.cs: Added the UnicodeValidator class here
+ * src/EncodingInfo.cs: Ugh, reversed the previous patch... that class
+ really doesn't belong in this file ;)
+
+2005-11-27 Aaron Bockover <[EMAIL PROTECTED]>
+
* src/EncodingInfo.cs: UnicodeValidator class to validate a string as
UTF-8; by Ruben Vermeersch
Modified: trunk/entagged-sharp/src/EncodingInfo.cs
===================================================================
--- trunk/entagged-sharp/src/EncodingInfo.cs 2005-11-28 03:02:49 UTC (rev
53533)
+++ trunk/entagged-sharp/src/EncodingInfo.cs 2005-11-28 03:06:35 UTC (rev
53534)
@@ -102,81 +102,4 @@
return sb.ToString().Substring(0,sb.Length-1);
}
}
-public static class UnicodeValidator
-{
- public static bool ValidateUtf8(byte [] str)
- {
- int i, min = 0, val = 0;
-
- try {
- for(i = 0; i < str.Length; i++) {
- if(str[i] < 128) {
- continue;
- }
-
- if((str[i] & 0xe0) == 0xc0) { /* 110xxxxx */
- if((str[i] & 0x1e) == 0) {
- return false;
- }
-
- if((str[++i] & 0xc0) != 0x80) { /* 10xxxxxx */
- return false;
- }
- } else {
- bool skip_next_continuation = false;
-
- if((str[i] & 0xf0) == 0xe0) { /* 1110xxxx */
- min = 1 << 11;
- val = str[i] & 0x0f;
- skip_next_continuation = true;
- } else if((str[i] & 0xf8) == 0xf0) { /* 11110xxx */
- min = 1 << 16;
- val = str[i] & 0x07;
- } else {
- return false;
- }
-
- if(!skip_next_continuation && !IsContinuationChar(str,
++i, ref val)) {
- return false;
- }
-
- if(!IsContinuationChar(str, ++i, ref val)) {
- return false;
- }
-
- if(!IsContinuationChar(str, ++i, ref val)) {
- return false;
- }
-
- if(val < min || !IsValidUnicode(val)) {
- return false;
- }
- }
- }
- } catch(IndexOutOfRangeException e) {
- return false;
- }
-
- return true;
- }
-
- private static bool IsContinuationChar(byte [] str, int i, ref int val)
- {
- if((str[i] & 0xc0) != 0x80) { /* 10xxxxxx */
- return false;
- }
-
- val <<= 6;
- val |= str[i] & 0x3f;
-
- return true;
- }
-
- private static bool IsValidUnicode(int b)
- {
- return (b < 0x110000 &&
- ((b & 0xFFFFF800) != 0xD800) &&
- (b < 0xFDD0 || b > 0xFDEF) &&
- (b & 0xFFFE) != 0xFFFE);
- }
}
Modified: trunk/entagged-sharp/src/Util/Utils.cs
===================================================================
--- trunk/entagged-sharp/src/Util/Utils.cs 2005-11-28 03:02:49 UTC (rev
53533)
+++ trunk/entagged-sharp/src/Util/Utils.cs 2005-11-28 03:06:35 UTC (rev
53534)
@@ -188,4 +188,83 @@
}
}
}
+
+ public static class UnicodeValidator
+ {
+ public static bool ValidateUtf8(byte [] str)
+ {
+ int i, min = 0, val = 0;
+
+ try {
+ for(i = 0; i < str.Length; i++) {
+ if(str[i] < 128) {
+ continue;
+ }
+
+ if((str[i] & 0xe0) == 0xc0) { /* 110xxxxx */
+ if((str[i] & 0x1e) == 0) {
+ return false;
+ }
+
+ if((str[++i] & 0xc0) != 0x80) { /* 10xxxxxx */
+ return false;
+ }
+ } else {
+ bool skip_next_continuation = false;
+
+ if((str[i] & 0xf0) == 0xe0) { /* 1110xxxx */
+ min = 1 << 11;
+ val = str[i] & 0x0f;
+ skip_next_continuation = true;
+ } else if((str[i] & 0xf8) == 0xf0) { /* 11110xxx */
+ min = 1 << 16;
+ val = str[i] & 0x07;
+ } else {
+ return false;
+ }
+
+ if(!skip_next_continuation && !IsContinuationChar(str,
++i, ref val)) {
+ return false;
+ }
+
+ if(!IsContinuationChar(str, ++i, ref val)) {
+ return false;
+ }
+
+ if(!IsContinuationChar(str, ++i, ref val)) {
+ return false;
+ }
+
+ if(val < min || !IsValidUnicode(val)) {
+ return false;
+ }
+ }
+ }
+ } catch(IndexOutOfRangeException e) {
+ return false;
+ }
+
+ return true;
+ }
+
+ private static bool IsContinuationChar(byte [] str, int i, ref int val)
+ {
+ if((str[i] & 0xc0) != 0x80) { /* 10xxxxxx */
+ return false;
+ }
+
+ val <<= 6;
+ val |= str[i] & 0x3f;
+
+ return true;
+ }
+
+ private static bool IsValidUnicode(int b)
+ {
+ return (b < 0x110000 &&
+ ((b & 0xFFFFF800) != 0xD800) &&
+ (b < 0xFDD0 || b > 0xFDEF) &&
+ (b & 0xFFFE) != 0xFFFE);
+ }
+ }
}
_______________________________________________
Mono-patches maillist - [email protected]
http://lists.ximian.com/mailman/listinfo/mono-patches
