From bc6a5c23ccb16f47c9d444a894025bd1123e51b4 Mon Sep 17 00:00:00 2001
From: "Reece H. Dunn" <msclrhd@gmail.com>
Date: Sat, 22 Dec 2012 01:19:03 +0000
Subject: [PATCH] Remove the Zc class as it is not part of the UCD; special
 case Cc-based whitespace instead.

---
 src/categories.cpp    |  2 +-
 src/ctype.cpp         | 22 ++++++++++++++--------
 src/include/ucd/ucd.h |  1 -
 tools/categories.py   |  2 +-
 4 files changed, 16 insertions(+), 11 deletions(-)

diff --git a/src/categories.cpp b/src/categories.cpp
index 0be4a315..230a5058 100644
--- a/src/categories.cpp
+++ b/src/categories.cpp
@@ -3285,7 +3285,7 @@ ucd::category_group ucd::lookup_category_group(codepoint_t c)
 {
 	switch (lookup_category(c))
 	{
-	case Cc: case Cf: case Ci: case Cn: case Co: case Cs: case Zc:
+	case Cc: case Cf: case Ci: case Cn: case Co: case Cs:
 		return C;
 	case Ll: case Lm: case Lo: case Lt: case Lu:
 		return L;
diff --git a/src/ctype.cpp b/src/ctype.cpp
index 4488378e..6237ba73 100644
--- a/src/ctype.cpp
+++ b/src/ctype.cpp
@@ -46,12 +46,7 @@ int ucd::isalpha(codepoint_t c)
 
 int ucd::iscntrl(codepoint_t c)
 {
-	switch (lookup_category(c))
-	{
-	case Cc: case Zc:
-		return 1;
-	}
-	return 0;
+	return lookup_category(c) == Cc;
 }
 
 int ucd::isdigit(codepoint_t c)
@@ -69,7 +64,7 @@ int ucd::isgraph(codepoint_t c)
 	switch (lookup_category(c))
 	{
 	case Cc: case Cf: case Ci: case Cn: case Co: case Cs:
-	case Zc: case Zl: case Zp: case Zs:
+	case Zl: case Zp: case Zs:
 		return 0;
 	}
 	return 1;
@@ -104,8 +99,19 @@ int ucd::isspace(codepoint_t c)
 {
 	switch (lookup_category(c))
 	{
-	case Zc: case Zl: case Zp: case Zs:
+	case Zl: case Zp: case Zs:
 		return 1;
+	case Cc:
+		switch (c) // Some control characters are also whitespace characters:
+		{
+		case 0x09: // U+0009 : CHARACTER TABULATION
+		case 0x0A: // U+000A : LINE FEED
+		case 0x0B: // U+000B : LINE TABULATION
+		case 0x0C: // U+000C : FORM FEED
+		case 0x0D: // U+000D : CARRIAGE RETURN
+		case 0x85: // U+0085 : NEXT LINE
+			return 1;
+		}
 	}
 	return 0;
 }
diff --git a/src/include/ucd/ucd.h b/src/include/ucd/ucd.h
index 6a3c1f54..cb95d2ca 100644
--- a/src/include/ucd/ucd.h
+++ b/src/include/ucd/ucd.h
@@ -84,7 +84,6 @@ namespace ucd
 		Sm, /**< @brief Math Symbol */
 		So, /**< @brief Other Symbol */
 
-		Zc, /**< @brief Whitespace character in the Cc category */
 		Zl, /**< @brief Line Separator */
 		Zp, /**< @brief Paragraph Separator */
 		Zs, /**< @brief Space Separator */
diff --git a/tools/categories.py b/tools/categories.py
index 94ed932c..895ce747 100755
--- a/tools/categories.py
+++ b/tools/categories.py
@@ -159,7 +159,7 @@ ucd::category_group ucd::lookup_category_group(codepoint_t c)
 {
 	switch (lookup_category(c))
 	{
-	case Cc: case Cf: case Ci: case Cn: case Co: case Cs: case Zc:
+	case Cc: case Cf: case Ci: case Cn: case Co: case Cs:
 		return C;
 	case Ll: case Lm: case Lo: case Lt: case Lu:
 		return L;