X-Git-Url: https://pd.if.org/git/?a=blobdiff_plain;f=functions%2Flocale%2FUnicodeData.py;h=6fe74cb8d90ee263a1681274bf2faa5fc2255548;hb=c7496dafe5261c2f0d0c4becc036ac9c9e5f2b64;hp=42a8f9f3bad916923fe1013ad8afd3ade0863164;hpb=c6158c4e0aa0634d87ae800be0959e7a99a84b37;p=pdclib.old

diff --git a/functions/locale/UnicodeData.py b/functions/locale/UnicodeData.py
index 42a8f9f..6fe74cb 100644
--- a/functions/locale/UnicodeData.py
+++ b/functions/locale/UnicodeData.py
@@ -26,6 +26,7 @@ BIT_SPACE =  32
 BIT_LOWER =  64
 BIT_UPPER = 128
 BIT_DIGIT = 256
+BIT_XDIGT = 512
 
 # Category to bitfield mapping
 categories = {
@@ -48,12 +49,48 @@ categories = {
     'Sc': BIT_GRAPH,                            # Currency symbol
     'Sk': BIT_GRAPH,                            # Non-letterlike modifier symbol
     'So': BIT_GRAPH,                            # Other symbol
-    'Zs': BIT_SPACE | BIT_GRAPH | BIT_BLANK,    # Non-zero-width space character
-    'Zl': BIT_SPACE | BIT_GRAPH,                # Line separator
-    'Zp': BIT_SPACE | BIT_GRAPH,                # Paragraph separator
+    'Zs': BIT_SPACE,                            # Non-zero-width space character
+    'Zl': BIT_SPACE,                            # Line separator
+    'Zp': BIT_SPACE,                            # Paragraph separator
     'Cc': BIT_CNTRL,                            # C0/C1 control codes
 }
 
+# Characters with special properties
+special = {
+    # Blank characters
+    0x0020: BIT_SPACE | BIT_BLANK, # space
+    0x0009: BIT_SPACE | BIT_BLANK, # tab
+
+    # Digits
+    0x0030: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,
+    0x0031: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,
+    0x0032: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,
+    0x0033: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,
+    0x0034: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,
+    0x0035: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,
+    0x0036: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,
+    0x0037: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,
+    0x0038: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,
+    0x0039: BIT_XDIGT | BIT_DIGIT | BIT_GRAPH,
+
+    # A-F (hex uppercase)
+    0x0041: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_UPPER,
+    0x0042: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_UPPER,
+    0x0043: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_UPPER,
+    0x0044: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_UPPER,
+    0x0045: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_UPPER,
+    0x0046: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_UPPER,
+
+
+    # a-f (hex lowercase)
+    0x0061: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_LOWER,
+    0x0062: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_LOWER,
+    0x0063: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_LOWER,
+    0x0064: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_LOWER,
+    0x0065: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_LOWER,
+    0x0066: BIT_XDIGT | BIT_ALPHA | BIT_GRAPH | BIT_LOWER,
+}
+
 in_file  = open('UnicodeData.txt', 'r')
 out_file = open('_PDCLIB_unicodedata.c', 'w')
 try:
@@ -83,7 +120,7 @@ try:
         num       = int(num_hex, 16)
         upper_case = int(upper_case_hex, 16) if len(upper_case_hex) else num
         lower_case = int(lower_case_hex, 16) if len(lower_case_hex) else num
-        bits = categories.get(category, 0)
+        bits = special.get(num, categories.get(category, 0))
 
         if upper_case == 0 and lower_case == 0 and bits == 0:
             continue