1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
|
#!/usr/bin/env python3
#
# Copyright 2020 Ettus Research, a National Instruments Brand
#
# SPDX-License-Identifier: GPL-3.0-or-later
#
import sys
import re
if len(sys.argv) < 2:
print("Please supply an input filename!")
sys.exit()
input_arg = sys.argv[1]
filename_pat = re.compile(r"(.*)\.c")
m = filename_pat.match(input_arg)
if not m:
print("Please supply a .c file as input!")
sys.exit()
filename = m.group(1)
input_file = open(filename + ".c", "r")
cpp_file = open(filename + ".cpp", "w")
python_file = open(filename + ".py", "w")
OUTPUT_HEADER_CPP = """//
// Copyright 2020 Ettus Research, a National Instruments Brand
//
// SPDX-License-Identifier: GPL-3.0-or-later
//
#include <tuple>
"""
OUTPUT_HEADER_PYTHON = """#
# Copyright 2020 Ettus Research, a National Instruments Company
#
# SPDX-License-Identifier: GPL-3.0-or-later
#
"""
# Matches a line defining a new packet, e.g
# char peer1_19[] = { /* Packet 3153 */
# Group 1 is the array name, e.g. "peer1_19"
# Group 2 is the Commented Description, e.g. "Packet 3072"
define_pat = re.compile(r"^char (peer[0|1]_\d+)\[\] = \{ \/\* (Packet \d+) \*\/$")
# Matches a line in the middle of a packet or at the end, e.g.
# 0x01, 0x00, 0x18, 0x00, 0x02, 0x00, 0x80, 0x00,
# 0x08, 0x00, 0x0f, 0x00 };
# No groups, use full match
other_pat = re.compile(r"^((?:0x[0-9a-f]{2}, )*(?:0x[0-9a-f]{2}))(?: \};|, )$")
# Used to seperate the two parties in the conversation (peer0 and peer1)
var_sort_pat = re.compile(r"^peer0_\d+$")
cpp_file.write(OUTPUT_HEADER_CPP)
cpp_file.write("namespace {} {{\n\n".format(filename))
python_file.write(OUTPUT_HEADER_PYTHON)
var_names_peer_0 = []
var_names_peer_1 = []
while True:
line = input_file.readline()
if not line:
break
m = define_pat.match(line)
if m:
cpp_file.write("uint8_t {}[] = {{ // {}\n".format(m.group(1), m.group(2)))
python_file.write("{} = bytes([ # {}\n".format(m.group(1), m.group(2)))
var_sort_match = var_sort_pat.match(m.group(1))
if var_sort_match:
var_names_peer_0.append(m.group(1))
else:
var_names_peer_1.append(m.group(1))
continue
m = other_pat.match(line)
if m:
cpp_file.write(line)
python_file.write(line.replace("};", "])"))
continue
print("Encountered unexpected line:\n{}".format(line))
sys.exit()
for peer_name, var_names in [("peer0", var_names_peer_0), ("peer1", var_names_peer_1)]:
cpp_file.write("\n")
cpp_file.write("size_t {}_len = {};\n".format(peer_name, len(var_names)))
cpp_file.write("std::tuple<uint8_t*, size_t> {}[] = {{\n".format(peer_name))
python_file.write("\n")
python_file.write("{} = [\n\t".format(peer_name))
for var_name in var_names:
cpp_file.write("\tstd::make_tuple({0}, sizeof({0})),\n".format(var_name))
cpp_file.write("};\n")
for i, var_name in enumerate(var_names):
python_file.write(var_name)
if i + 1 < len(var_names):
python_file.write(",\t")
if (i + 1) % 10 == 0:
python_file.write("\n\t")
python_file.write("\n]\n")
cpp_file.write("\n")
cpp_file.write("}} // namespace {}\n".format(filename))
|