/src/botan/src/lib/codec/hex/hex.cpp
Line | Count | Source |
1 | | /* |
2 | | * Hex Encoding and Decoding |
3 | | * (C) 2010,2020 Jack Lloyd |
4 | | * |
5 | | * Botan is released under the Simplified BSD License (see license.txt) |
6 | | */ |
7 | | |
8 | | #include <botan/hex.h> |
9 | | |
10 | | #include <botan/exceptn.h> |
11 | | #include <botan/mem_ops.h> |
12 | | #include <botan/internal/charset.h> |
13 | | #include <botan/internal/fmt.h> |
14 | | #include <botan/internal/int_utils.h> |
15 | | #include <botan/internal/loadstor.h> |
16 | | |
17 | | namespace Botan { |
18 | | |
19 | | namespace { |
20 | | |
21 | 388k | uint16_t hex_encode_2nibble(uint8_t n8, bool uppercase) { |
22 | | // Offset for upper or lower case 'a' resp |
23 | 388k | const uint16_t a_mask = uppercase ? 0x0707 : 0x2727; |
24 | | |
25 | 388k | const uint16_t n = (static_cast<uint16_t>(n8 & 0xF0) << 4) | (n8 & 0x0F); |
26 | | // n >= 10? If so add offset |
27 | 388k | const uint16_t diff = swar_lt<uint16_t>(0x0909, n) & a_mask; |
28 | | // Can't overflow between bytes, so don't need explicit SWAR addition: |
29 | 388k | return n + 0x3030 + diff; |
30 | 388k | } |
31 | | |
32 | | } // namespace |
33 | | |
34 | 4.37k | void hex_encode(char output[], const uint8_t input[], size_t input_length, bool uppercase) { |
35 | 392k | for(size_t i = 0; i != input_length; ++i) { |
36 | 388k | const uint16_t h = hex_encode_2nibble(input[i], uppercase); |
37 | 388k | output[2 * i] = get_byte<0>(h); |
38 | 388k | output[2 * i + 1] = get_byte<1>(h); |
39 | 388k | } |
40 | 4.37k | } |
41 | | |
42 | 4.37k | std::string hex_encode(const uint8_t input[], size_t input_length, bool uppercase) { |
43 | 4.37k | std::string output(2 * input_length, 0); |
44 | | |
45 | 4.37k | if(input_length > 0) { |
46 | 4.37k | hex_encode(&output.front(), input, input_length, uppercase); |
47 | 4.37k | } |
48 | | |
49 | 4.37k | return output; |
50 | 4.37k | } |
51 | | |
52 | | namespace { |
53 | | |
54 | 644 | uint8_t hex_char_to_bin(char input) { |
55 | | // Starts of valid value ranges (v_lo) and their lengths (v_range) |
56 | 644 | constexpr uint64_t v_lo = make_uint64(0, '0', 'a', 'A', ' ', '\n', '\t', '\r'); |
57 | 644 | constexpr uint64_t v_range = make_uint64(0, 10, 6, 6, 1, 1, 1, 1); |
58 | | |
59 | 644 | const uint8_t x = static_cast<uint8_t>(input); |
60 | 644 | const uint64_t x8 = x * 0x0101010101010101; |
61 | | |
62 | 644 | const uint64_t v_mask = swar_in_range<uint64_t>(x8, v_lo, v_range) ^ 0x8000000000000000; |
63 | | |
64 | | // This is the offset added to x to get the value we need |
65 | 644 | const uint64_t val_v = 0xd0a9c960767773 ^ static_cast<uint64_t>(0xFF - x) << 56; |
66 | | |
67 | 644 | return x + static_cast<uint8_t>(val_v >> (8 * index_of_first_set_byte(v_mask))); |
68 | 644 | } |
69 | | |
70 | | } // namespace |
71 | | |
72 | 12 | size_t hex_decode(uint8_t output[], const char input[], size_t input_length, size_t& input_consumed, bool ignore_ws) { |
73 | 12 | uint8_t* out_ptr = output; |
74 | 12 | bool top_nibble = true; |
75 | | |
76 | 12 | clear_mem(output, input_length / 2); |
77 | | |
78 | 656 | for(size_t i = 0; i != input_length; ++i) { |
79 | 644 | const uint8_t bin = hex_char_to_bin(input[i]); |
80 | | |
81 | 644 | if(bin >= 0x10) { |
82 | 0 | if(bin == 0x80 && ignore_ws) { |
83 | 0 | continue; |
84 | 0 | } |
85 | | |
86 | 0 | throw Invalid_Argument(fmt("hex_decode: invalid character '{}'", format_char_for_display(input[i]))); |
87 | 0 | } |
88 | | |
89 | 644 | if(top_nibble) { |
90 | 322 | *out_ptr |= bin << 4; |
91 | 322 | } else { |
92 | 322 | *out_ptr |= bin; |
93 | 322 | } |
94 | | |
95 | 644 | top_nibble = !top_nibble; |
96 | 644 | if(top_nibble) { |
97 | 322 | ++out_ptr; |
98 | 322 | } |
99 | 644 | } |
100 | | |
101 | 12 | input_consumed = input_length; |
102 | 12 | size_t written = (out_ptr - output); |
103 | | |
104 | | /* |
105 | | * We only got half of a uint8_t at the end; zap the half-written |
106 | | * output and mark it as unread |
107 | | */ |
108 | 12 | if(!top_nibble) { |
109 | 0 | *out_ptr = 0; |
110 | 0 | input_consumed -= 1; |
111 | 0 | } |
112 | | |
113 | 12 | return written; |
114 | 12 | } |
115 | | |
116 | 12 | size_t hex_decode(uint8_t output[], const char input[], size_t input_length, bool ignore_ws) { |
117 | 12 | size_t consumed = 0; |
118 | 12 | size_t written = hex_decode(output, input, input_length, consumed, ignore_ws); |
119 | | |
120 | 12 | if(consumed != input_length) { |
121 | 0 | throw Invalid_Argument("hex_decode: input did not have full bytes"); |
122 | 0 | } |
123 | | |
124 | 12 | return written; |
125 | 12 | } |
126 | | |
127 | 0 | size_t hex_decode(uint8_t output[], std::string_view input, bool ignore_ws) { |
128 | 0 | return hex_decode(output, input.data(), input.length(), ignore_ws); |
129 | 0 | } |
130 | | |
131 | 0 | size_t hex_decode(std::span<uint8_t> output, std::string_view input, bool ignore_ws) { |
132 | 0 | return hex_decode(output.data(), input.data(), input.length(), ignore_ws); |
133 | 0 | } |
134 | | |
135 | 12 | secure_vector<uint8_t> hex_decode_locked(const char input[], size_t input_length, bool ignore_ws) { |
136 | 12 | secure_vector<uint8_t> bin(1 + input_length / 2); |
137 | | |
138 | 12 | size_t written = hex_decode(bin.data(), input, input_length, ignore_ws); |
139 | | |
140 | 12 | bin.resize(written); |
141 | 12 | return bin; |
142 | 12 | } |
143 | | |
144 | 0 | secure_vector<uint8_t> hex_decode_locked(std::string_view input, bool ignore_ws) { |
145 | 0 | return hex_decode_locked(input.data(), input.size(), ignore_ws); |
146 | 0 | } |
147 | | |
148 | 0 | std::vector<uint8_t> hex_decode(const char input[], size_t input_length, bool ignore_ws) { |
149 | 0 | std::vector<uint8_t> bin(1 + input_length / 2); |
150 | |
|
151 | 0 | size_t written = hex_decode(bin.data(), input, input_length, ignore_ws); |
152 | |
|
153 | 0 | bin.resize(written); |
154 | 0 | return bin; |
155 | 0 | } |
156 | | |
157 | 0 | std::vector<uint8_t> hex_decode(std::string_view input, bool ignore_ws) { |
158 | 0 | return hex_decode(input.data(), input.size(), ignore_ws); |
159 | 0 | } |
160 | | |
161 | | } // namespace Botan |