ctype: return true in isupper/islower if there is a simple case mapping present

CHANGELOG.md

@@ -3,6 +3,7 @@
 ## 9.0.0.1 - (In Progress)
 
 *  Add `iswblank` and `iswxdigit` compatibility.
+*  Improve ctype compatibility.
 
 ## 9.0.0 - 2016-12-28
 

src/ctype.c

@@ -97,7 +97,8 @@ int ucd_isgraph(codepoint_t c)
 
 int ucd_islower(codepoint_t c)
 {
-	return ucd_lookup_category(c) == UCD_CATEGORY_Ll;
+	return ucd_lookup_category(c) == UCD_CATEGORY_Ll
+	    || ucd_toupper(c) != c;
 }
 
 int ucd_isprint(codepoint_t c)
@@ -159,7 +160,8 @@ int ucd_isspace(codepoint_t c)
 
 int ucd_isupper(codepoint_t c)
 {
-	return ucd_lookup_category(c) == UCD_CATEGORY_Lu;
+	return ucd_lookup_category(c) == UCD_CATEGORY_Lu
+	    || ucd_tolower(c) != c;
 }
 
 int ucd_isxdigit(codepoint_t c)

tests/printcdata.c

@@ -248,7 +248,7 @@ int main(int argc, char **argv)
 	else
 	{
 		for (codepoint_t c = 0; c <= 0x10FFFF; ++c)
-			uprintf(stdout, c, format ? format : "%pH %s %C %c %UH %LH %TH %is\n");
+			uprintf(stdout, c, format ? format : "%pH %s %C %c %UH %LH %TH %is %iu %il\n");
 	}
 	return 0;
 }

tests/printucddata.c

@@ -243,7 +243,7 @@ int main(int argc, char **argv)
 	else
 	{
 		for (codepoint_t c = 0; c <= 0x10FFFF; ++c)
-			uprintf(stdout, c, format ? format : "%pH %s %C %c %UH %LH %TH %is\n");
+			uprintf(stdout, c, format ? format : "%pH %s %C %c %UH %LH %TH %is %iu %il\n");
 	}
 	return 0;
 }

tests/printucddata_cpp.cpp

@@ -243,7 +243,7 @@ int main(int argc, char **argv)
 	else
 	{
 		for (ucd::codepoint_t c = 0; c <= 0x10FFFF; ++c)
-			uprintf(stdout, c, format ? format : "%pH %s %C %c %UH %LH %TH %is\n");
+			uprintf(stdout, c, format ? format : "%pH %s %C %c %UH %LH %TH %is %iu %il\n");
 	}
 	return 0;
 }

tools/printdata.py

@@ -24,6 +24,8 @@ import ucd
 ucd_rootdir = sys.argv[1]
 csur_rootdir = 'data/csur'
 
+null = ucd.CodePoint('0000')
+
 unicode_chars = {}
 for data in ucd.parse_ucd_data(ucd_rootdir, 'UnicodeData'):
 	for codepoint in data['CodePoint']:
@@ -44,7 +46,22 @@ if '--with-csur' in sys.argv:
 def isspace(data):
 	return data.get('White_Space', 0)
 
-null = ucd.CodePoint('0000')
+def isupper(data):
+	if data.get('LowerCase', null) != null:
+		return 1
+	elif data.get('GeneralCategory', 'Cn') == 'Lu':
+		return 1
+	else:
+		return 0
+
+def islower(data):
+	if data.get('UpperCase', null) != null:
+		return 1
+	elif data.get('GeneralCategory', 'Cn') == 'Ll':
+		return 1
+	else:
+		return 0
+
 if __name__ == '__main__':
 	for codepoint in ucd.CodeRange('000000..10FFFF'):
 		try:
@@ -58,8 +75,9 @@ if __name__ == '__main__':
 		if title == null: title = codepoint
 		if upper == null: upper = codepoint
 		if lower == null: lower = codepoint
-		print('%s %s %s %s %s %s %s %s' % (
+		print('%s %s %s %s %s %s %s %s %s %s' % (
 		      codepoint, script,
 		      data.get('GeneralCategory', 'Cn')[0], data.get('GeneralCategory', 'Cn'),
 		      upper, lower, title,
-		      isspace(data)))
+		      isspace(data),
+		      isupper(data), islower(data)))