1 files changed, 106 insertions, 0 deletions
diff --git a/asm2c.py b/asm2c.py
new file mode 100755
index 0000000..f3ae2d5
--- /dev/null
+++ b/asm2c.py
@@ -0,0 +1,106 @@
+#!/usr/bin/env python
+
+import sys
+import re
+
+def index_pred(lst, pred):
+    for i, el in enumerate(lst):
+        if pred(el):
+            return i
+
+def is_ida_addr(s):
+    return re.match(r'(.*:)?[0-9a-fA-F]+$', s) is not None
+
+autonames = {
+  'db': ["byte",  0, "uint8_t"],
+  'dw': ["word",  0, "uint16_t"],
+  'dd': ["dword", 0, "uint32_t"],
+}
+
+def asm_int(s):
+    if s.endswith('h'):
+        return int(s[:-1], 16)
+    elif s.endswith('o'):
+        return int(s[:-1], 8)
+    else:
+        return int(s)
+
+class Symb(object):
+    def __init__(self, op, name):
+        self.name = name
+        self.op = op
+        self.values = []
+    def push(self, line):
+        orig_line = line
+        line = re.sub(";.*$", "", line)
+        op_rol = line.split(None, 1)
+        if len(op_rol) < 2:
+            print >> sys.stderr, "WARN dummy op in line: %s" % orig_line
+            return
+        op, rol = op_rol
+        if op != self.op:
+            raise ValueError("op != self.op (%s != %s)" % (op, self.op))
+        elems = [el.strip() for el in rol.split(",")]
+        for el in elems:
+            if " " in el:
+                parts = el.split()
+                if    len(parts) != 2 or not parts[0].isdigit() \
+                   or not parts[1].startswith("dup(") \
+                   or not parts[1].endswith(")"):
+                    raise ValueError("bad arg %s for op %s" % (el, op))
+                nb = int(parts[0])
+                val = asm_int(parts[1][4:-1])
+                self.values.extend([val] * nb)
+            else:
+                self.values.append(asm_int(el))
+    def close(self):
+        pass
+    def out_lines(self):
+        yield "static const %s %s[%d] = {" % (autonames[self.op][2],
+                                              self.name,
+                                              len(self.values))
+        for i, v in enumerate(self.values):
+            if not i or i % 16:
+                yield "\t0x%X," % v
+            else:
+                yield "\t0x%X,\t\t/* 0x%X %d */" % (v, i, i)
+        yield "};"
+        yield ""
+
+def new_symb(op, name):
+    if name is None:
+        auto = autonames[op]
+        name = auto[0] + '_' + str(auto[1])
+        auto[1] += 1
+    return Symb(op, name)
+
+symbols = []
+current_symb = None
+# dismiss_first_words = None
+
+lines = sys.stdin.readlines()
+for line in lines:
+    line = line.strip()
+    elems = line.split()
+    pos = index_pred(elems, lambda a: a in autonames)
+    if pos is None:
+        print >> sys.stderr, "unrecognize line: %s" % line
+        continue
+    op = elems[pos]
+    named = pos > 0 and not is_ida_addr(elems[pos-1])
+    # if dimiss_first_words is None and named:
+    #    dimiss_first_words = pos - 1
+    if named:
+        name = elems[pos-1]
+        if current_symb:
+            current_symb.close()
+        current_symb = new_symb(op, name)
+        symbols.append(current_symb)
+    current_symb.push(" ".join(elems[pos:]))
+
+if current_symb:
+    current_symb.close()
+
+for symb in symbols:
+    for line in symb.out_lines():
+        print line