PDCLIB-3 Add XDIGIT to list of bits in Unicode character data

diff --git a/functions/locale/UnicodeData.py b/functions/locale/UnicodeData.py
index 42a8f9f3bad916923fe1013ad8afd3ade0863164..e31ec2e29f02fb2b5022a23a7cb0918273b20e40 100644 (file)
--- a/functions/locale/UnicodeData.py
+++ b/functions/locale/UnicodeData.py
@@ -26,6 +26,7 @@ BIT_SPACE =  32
 BIT_LOWER =  64\r
 BIT_UPPER = 128\r
 BIT_DIGIT = 256\r
+BIT_XDIGT = 512\r
 \r
 # Category to bitfield mapping\r
 categories = {\r
@@ -54,6 +55,38 @@ categories = {
     'Cc': BIT_CNTRL,                            # C0/C1 control codes\r
 }\r
 \r
+# Characters with special properties\r
+special = {\r
+    # Digits\r
+    0x0030: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,\r
+    0x0031: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,\r
+    0x0032: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,\r
+    0x0033: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,\r
+    0x0034: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,\r
+    0x0035: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,\r
+    0x0036: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,\r
+    0x0037: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,\r
+    0x0038: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,\r
+    0x0039: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,\r
+\r
+    # A-F (hex uppercase)\r
+    0x0041: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_UPPER,\r
+    0x0042: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_UPPER,\r
+    0x0043: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_UPPER,\r
+    0x0044: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_UPPER,\r
+    0x0045: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_UPPER,\r
+    0x0046: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_UPPER,\r
+\r
+\r
+    # a-f (hex lowercase)\r
+    0x0061: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_LOWER,\r
+    0x0062: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_LOWER,\r
+    0x0063: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_LOWER,\r
+    0x0064: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_LOWER,\r
+    0x0065: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_LOWER,\r
+    0x0066: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_LOWER,\r
+}\r
+\r
 in_file  = open('UnicodeData.txt', 'r')\r
 out_file = open('_PDCLIB_unicodedata.c', 'w')\r
 try:\r
@@ -83,7 +116,7 @@ try:
         num       = int(num_hex, 16)\r
         upper_case = int(upper_case_hex, 16) if len(upper_case_hex) else num\r
         lower_case = int(lower_case_hex, 16) if len(lower_case_hex) else num\r
-        bits = categories.get(category, 0)\r
+        bits = special.get(num, categories.get(category, 0))\r
 \r
         if upper_case == 0 and lower_case == 0 and bits == 0:\r
             continue\r

diff --git a/functions/locale/_PDCLIB_unicodedata.c b/functions/locale/_PDCLIB_unicodedata.c
index e4e9c83b48ac5f8c8de8792b3b459ecaec1d8f11..928226ce48e6dc95110f46b562191bde522b857d 100644 (file)
--- a/functions/locale/_PDCLIB_unicodedata.c
+++ b/functions/locale/_PDCLIB_unicodedata.c
@@ -63,16 +63,16 @@
     { 0x2D,    0x18,   0x2D,   0x2D }, // HYPHEN-MINUS\r
     { 0x2E,    0x18,   0x2E,   0x2E }, // FULL STOP\r
     { 0x2F,    0x18,   0x2F,   0x2F }, // SOLIDUS\r
-    { 0x30,    0x108,  0x30,   0x30 }, // DIGIT ZERO\r
-    { 0x31,    0x108,  0x31,   0x31 }, // DIGIT ONE\r
-    { 0x32,    0x108,  0x32,   0x32 }, // DIGIT TWO\r
-    { 0x33,    0x108,  0x33,   0x33 }, // DIGIT THREE\r
-    { 0x34,    0x108,  0x34,   0x34 }, // DIGIT FOUR\r
-    { 0x35,    0x108,  0x35,   0x35 }, // DIGIT FIVE\r
-    { 0x36,    0x108,  0x36,   0x36 }, // DIGIT SIX\r
-    { 0x37,    0x108,  0x37,   0x37 }, // DIGIT SEVEN\r
-    { 0x38,    0x108,  0x38,   0x38 }, // DIGIT EIGHT\r
-    { 0x39,    0x108,  0x39,   0x39 }, // DIGIT NINE\r
+    { 0x30,    0x308,  0x30,   0x30 }, // DIGIT ZERO\r
+    { 0x31,    0x308,  0x31,   0x31 }, // DIGIT ONE\r
+    { 0x32,    0x308,  0x32,   0x32 }, // DIGIT TWO\r
+    { 0x33,    0x308,  0x33,   0x33 }, // DIGIT THREE\r
+    { 0x34,    0x308,  0x34,   0x34 }, // DIGIT FOUR\r
+    { 0x35,    0x308,  0x35,   0x35 }, // DIGIT FIVE\r
+    { 0x36,    0x308,  0x36,   0x36 }, // DIGIT SIX\r
+    { 0x37,    0x308,  0x37,   0x37 }, // DIGIT SEVEN\r
+    { 0x38,    0x308,  0x38,   0x38 }, // DIGIT EIGHT\r
+    { 0x39,    0x308,  0x39,   0x39 }, // DIGIT NINE\r
     { 0x3A,    0x18,   0x3A,   0x3A }, // COLON\r
     { 0x3B,    0x18,   0x3B,   0x3B }, // SEMICOLON\r
     { 0x3C,    0x8,    0x3C,   0x3C }, // LESS-THAN SIGN\r
@@ -80,12 +80,12 @@
     { 0x3E,    0x8,    0x3E,   0x3E }, // GREATER-THAN SIGN\r
     { 0x3F,    0x18,   0x3F,   0x3F }, // QUESTION MARK\r
     { 0x40,    0x18,   0x40,   0x40 }, // COMMERCIAL AT\r
-    { 0x41,    0x89,   0x61,   0x41 }, // LATIN CAPITAL LETTER A\r
-    { 0x42,    0x89,   0x62,   0x42 }, // LATIN CAPITAL LETTER B\r
-    { 0x43,    0x89,   0x63,   0x43 }, // LATIN CAPITAL LETTER C\r
-    { 0x44,    0x89,   0x64,   0x44 }, // LATIN CAPITAL LETTER D\r
-    { 0x45,    0x89,   0x65,   0x45 }, // LATIN CAPITAL LETTER E\r
-    { 0x46,    0x89,   0x66,   0x46 }, // LATIN CAPITAL LETTER F\r
+    { 0x41,    0x289,  0x61,   0x41 }, // LATIN CAPITAL LETTER A\r
+    { 0x42,    0x289,  0x62,   0x42 }, // LATIN CAPITAL LETTER B\r
+    { 0x43,    0x289,  0x63,   0x43 }, // LATIN CAPITAL LETTER C\r
+    { 0x44,    0x289,  0x64,   0x44 }, // LATIN CAPITAL LETTER D\r
+    { 0x45,    0x289,  0x65,   0x45 }, // LATIN CAPITAL LETTER E\r
+    { 0x46,    0x289,  0x66,   0x46 }, // LATIN CAPITAL LETTER F\r
     { 0x47,    0x89,   0x67,   0x47 }, // LATIN CAPITAL LETTER G\r
     { 0x48,    0x89,   0x68,   0x48 }, // LATIN CAPITAL LETTER H\r
     { 0x49,    0x89,   0x69,   0x49 }, // LATIN CAPITAL LETTER I\r
@@ -112,12 +112,12 @@
     { 0x5E,    0x8,    0x5E,   0x5E }, // CIRCUMFLEX ACCENT\r
     { 0x5F,    0x18,   0x5F,   0x5F }, // LOW LINE\r
     { 0x60,    0x8,    0x60,   0x60 }, // GRAVE ACCENT\r
-    { 0x61,    0x49,   0x61,   0x41 }, // LATIN SMALL LETTER A\r
-    { 0x62,    0x49,   0x62,   0x42 }, // LATIN SMALL LETTER B\r
-    { 0x63,    0x49,   0x63,   0x43 }, // LATIN SMALL LETTER C\r
-    { 0x64,    0x49,   0x64,   0x44 }, // LATIN SMALL LETTER D\r
-    { 0x65,    0x49,   0x65,   0x45 }, // LATIN SMALL LETTER E\r
-    { 0x66,    0x49,   0x66,   0x46 }, // LATIN SMALL LETTER F\r
+    { 0x61,    0x249,  0x61,   0x41 }, // LATIN SMALL LETTER A\r
+    { 0x62,    0x249,  0x62,   0x42 }, // LATIN SMALL LETTER B\r
+    { 0x63,    0x249,  0x63,   0x43 }, // LATIN SMALL LETTER C\r
+    { 0x64,    0x249,  0x64,   0x44 }, // LATIN SMALL LETTER D\r
+    { 0x65,    0x249,  0x65,   0x45 }, // LATIN SMALL LETTER E\r
+    { 0x66,    0x249,  0x66,   0x46 }, // LATIN SMALL LETTER F\r
     { 0x67,    0x49,   0x67,   0x47 }, // LATIN SMALL LETTER G\r
     { 0x68,    0x49,   0x68,   0x48 }, // LATIN SMALL LETTER H\r
     { 0x69,    0x49,   0x69,   0x49 }, // LATIN SMALL LETTER I\r