From b7da8dbb0e2236f37e4b4c99b88598369da3a008 Mon Sep 17 00:00:00 2001
From: Rangi <remy.oukaour+rangi@gmail.com>
Date: Sun, 9 Aug 2020 13:49:34 -0400
Subject: tools/ contains what's needed to build the ROMs; utils/ contains
 disassembly utilites

---
 utils/read_charmap.py | 45 +++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 45 insertions(+)
 create mode 100644 utils/read_charmap.py

(limited to 'utils/read_charmap.py')

diff --git a/utils/read_charmap.py b/utils/read_charmap.py
new file mode 100644
index 0000000..77036f6
--- /dev/null
+++ b/utils/read_charmap.py
@@ -0,0 +1,45 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+import os, io
+from re import compile
+from sys import stderr
+
+charmap_regex = compile('[ \t]*charmap[ \t]+"(.*?)",[ \t]*(\$[0-9A-Fa-f]{2}|%[01]{8}|[0-9]{3})')
+# A charmap line is
+# [ \t]*   - zero or more space chars
+# charmap  - literal charmap
+# [ \t]+   - one or more space chars
+# "(.*?)"  - a lazily-matched text identifier in quotes
+# ,        - literal comma
+# [ \t]*   - zero or more space chars
+# (        - either of
+#   \$[0-9A-Fa-f]{2} - two hexadecimal digits preceeded by literal $
+#   %[01]{8}         - eight dual digits preceeded by literal %
+#   [0-9]{3}         - three decimal digits
+# )
+
+def parse_int(s):
+	# assumes integers are literal; no +-*/, etc
+	s = s.strip()
+	if s.startswith('$'):
+		return int(s[1:], 16)
+	if s.startswith('%'):
+		return int(s[1:], 2)
+	return int(s)
+
+def read_charmap(charmap_path):
+	charmap = {}
+	with io.open(charmap_path, 'r', encoding='utf-8') as f:
+		lines = f.readlines()
+	for line in lines:
+		m = charmap_regex.match(line)
+		if m is None:
+			continue
+		char = m.group(1)
+		value = parse_int(m.group(2))
+		if value in charmap:
+			print('Value {0:s} already in charmap, dropping it in favor of first charmap entry'.format(m.group(2)))
+			continue
+		charmap[value] = char
+	return charmap
-- 
cgit v1.2.3