make-usb.py: use octal escape sequences

Fixes errors like:

    epan/dissectors/usb.c:15220:42: error: hex escape sequence out of range
        { 0x0cad9001, "PowerPad Pocket PC\xc2\xa0Device" },

Change-Id: I8c120892c0d52aceb3f6767401e7944353495825
Fixes: v3.1.0rc0-524-g6f57aa72a8 ("Make a couple of scripts Python 3 only.")
Reviewed-on: https://code.wireshark.org/review/32854
Petri-Dish: Peter Wu <peter@lekensteyn.nl>
Reviewed-by: Gerald Combs <gerald@wireshark.org>

1 files changed, 10 insertions, 4 deletions

diff --git a/tools/make-usb.py b/tools/make-usb.py
index 19d776bd85..b63f477314 100755
--- a/tools/make-usb.py
+++ b/tools/make-usb.py
@@ -34,15 +34,21 @@ products = dict()
 vendors_str="static const value_string usb_vendors_vals[] = {\n"
 products_str="static const value_string usb_products_vals[] = {\n"
 
+# Escape backslashes, quotes, control characters and non-ASCII characters.
+escapes = {}
+for i in range(256):
+    if i in b'\\"':
+        escapes[i] = '\\%c' % i
+    elif i in range(0x20, 0x80) or i in b'\t':
+        escapes[i] = chr(i)
+    else:
+        escapes[i] = '\\%03o' % i
 
 for utf8line in lines:
     # Convert single backslashes to double (escaped) backslashes, escape quotes, etc.
     utf8line = utf8line.rstrip()
-    utf8line = utf8line.replace('\\', '\\\\')
-    utf8line = utf8line.replace('"', '\\"')
     utf8line = re.sub("\?+", "?", utf8line)
-    # Finally, convert non-ASCII UTF-8 sequences to C-style escapes
-    line = utf8line.encode('UTF-8').decode('ascii', 'backslashreplace')
+    line = ''.join(escapes[byte] for byte in utf8line.encode('utf8'))
 
     if line == "# Vendors, devices and interfaces. Please keep sorted.":
         mode = MODE_VENDOR_PRODUCT