Willy Tarreau | 4576424 | 2017-12-30 13:18:25 +0100 | [diff] [blame] | 1 | /* |
| 2 | * HPACK stream decoder. Takes a series of hex codes on stdin using one line |
| 3 | * per HEADERS frame. Spaces, tabs, CR, '-' and ',' are silently skipped. |
| 4 | * e.g. : |
| 5 | * echo 82864188f439ce75c875fa5784 | contrib/hpack/decode |
| 6 | * |
| 7 | * The DHT size may optionally be changed in argv[1]. |
| 8 | * |
| 9 | * Build like this : |
Willy Tarreau | fc80e30 | 2020-05-27 11:05:33 +0200 | [diff] [blame] | 10 | * gcc -I../../include -O0 -g -fno-strict-aliasing -fwrapv \ |
Willy Tarreau | 4576424 | 2017-12-30 13:18:25 +0100 | [diff] [blame] | 11 | * -o decode decode.c |
| 12 | */ |
Willy Tarreau | 318adf4 | 2020-05-22 12:09:16 +0200 | [diff] [blame] | 13 | |
| 14 | #define HPACK_STANDALONE |
| 15 | |
Willy Tarreau | 4576424 | 2017-12-30 13:18:25 +0100 | [diff] [blame] | 16 | #include <ctype.h> |
Willy Tarreau | a1bd1fa | 2019-03-29 17:26:33 +0100 | [diff] [blame] | 17 | #include <inttypes.h> |
Willy Tarreau | 4576424 | 2017-12-30 13:18:25 +0100 | [diff] [blame] | 18 | #include <stdio.h> |
| 19 | #include <stdlib.h> |
| 20 | #include <unistd.h> |
Willy Tarreau | c13ed53 | 2020-06-02 10:22:45 +0200 | [diff] [blame] | 21 | #include <haproxy/chunk.h> |
Willy Tarreau | be327fa | 2020-06-03 09:09:57 +0200 | [diff] [blame] | 22 | #include <haproxy/hpack-dec.h> |
Willy Tarreau | 4576424 | 2017-12-30 13:18:25 +0100 | [diff] [blame] | 23 | |
| 24 | #define MAX_RQ_SIZE 65536 |
| 25 | #define MAX_HDR_NUM 1000 |
| 26 | |
| 27 | char hex[MAX_RQ_SIZE*3+3]; // enough for "[ XX]* <CR> <LF> \0" |
| 28 | uint8_t buf[MAX_RQ_SIZE]; |
| 29 | |
| 30 | char trash_buf[MAX_RQ_SIZE]; |
| 31 | char tmp_buf[MAX_RQ_SIZE]; |
| 32 | |
Willy Tarreau | 83061a8 | 2018-07-13 11:56:34 +0200 | [diff] [blame] | 33 | struct buffer trash = { .area = trash_buf, .data = 0, .size = sizeof(trash_buf) }; |
| 34 | struct buffer tmp = { .area = tmp_buf, .data = 0, .size = sizeof(tmp_buf) }; |
Willy Tarreau | 4576424 | 2017-12-30 13:18:25 +0100 | [diff] [blame] | 35 | |
| 36 | /* displays a <len> long memory block at <buf>, assuming first byte of <buf> |
| 37 | * has address <baseaddr>. String <pfx> may be placed as a prefix in front of |
| 38 | * each line. It may be NULL if unused. The output is emitted to file <out>. |
| 39 | */ |
| 40 | void debug_hexdump(FILE *out, const char *pfx, const char *buf, |
| 41 | unsigned int baseaddr, int len) |
| 42 | { |
| 43 | unsigned int i; |
| 44 | int b, j; |
| 45 | |
| 46 | for (i = 0; i < (len + (baseaddr & 15)); i += 16) { |
| 47 | b = i - (baseaddr & 15); |
| 48 | fprintf(out, "%s%08x: ", pfx ? pfx : "", i + (baseaddr & ~15)); |
| 49 | for (j = 0; j < 8; j++) { |
| 50 | if (b + j >= 0 && b + j < len) |
| 51 | fprintf(out, "%02x ", (unsigned char)buf[b + j]); |
| 52 | else |
| 53 | fprintf(out, " "); |
| 54 | } |
| 55 | |
| 56 | if (b + j >= 0 && b + j < len) |
| 57 | fputc('-', out); |
| 58 | else |
| 59 | fputc(' ', out); |
| 60 | |
| 61 | for (j = 8; j < 16; j++) { |
| 62 | if (b + j >= 0 && b + j < len) |
| 63 | fprintf(out, " %02x", (unsigned char)buf[b + j]); |
| 64 | else |
| 65 | fprintf(out, " "); |
| 66 | } |
| 67 | |
| 68 | fprintf(out, " "); |
| 69 | for (j = 0; j < 16; j++) { |
| 70 | if (b + j >= 0 && b + j < len) { |
| 71 | if (isprint((unsigned char)buf[b + j])) |
| 72 | fputc((unsigned char)buf[b + j], out); |
| 73 | else |
| 74 | fputc('.', out); |
| 75 | } |
| 76 | else |
| 77 | fputc(' ', out); |
| 78 | } |
| 79 | fputc('\n', out); |
| 80 | } |
| 81 | } |
| 82 | |
| 83 | /* enable DEBUG_HPACK to show each individual hpack code */ |
| 84 | #define DEBUG_HPACK |
| 85 | #include "../src/hpack-huff.c" |
| 86 | #include "../src/hpack-tbl.c" |
| 87 | #include "../src/hpack-dec.c" |
| 88 | |
| 89 | /* display the message and exit with the code */ |
| 90 | __attribute__((noreturn)) void die(int code, const char *format, ...) |
| 91 | { |
| 92 | va_list args; |
| 93 | |
| 94 | if (format) { |
| 95 | va_start(args, format); |
| 96 | vfprintf(stderr, format, args); |
| 97 | va_end(args); |
| 98 | } |
| 99 | exit(code); |
| 100 | } |
| 101 | |
| 102 | /* reads <hex> and stops at the first LF, '#' or \0. Converts from hex to |
| 103 | * binary, ignoring spaces, tabs, CR, "-" and ','. The output is sent into |
| 104 | * <bin> for no more than <size> bytes. The number of bytes placed there is |
| 105 | * returned, or a negative value in case of parsing error. |
| 106 | */ |
| 107 | int hex2bin(const char *hex, uint8_t *bin, int size) |
| 108 | { |
| 109 | int a, b, c; |
| 110 | uint8_t code; |
| 111 | int len = 0; |
| 112 | |
| 113 | a = b = -1; |
| 114 | |
| 115 | for (; *hex; hex++) { |
| 116 | c = *hex; |
| 117 | if (c == ' ' || c == '\t' || c == '\r' || |
| 118 | c == '-' || c == ',') |
| 119 | continue; |
| 120 | |
| 121 | if (c == '\n' || c == '#') |
| 122 | break; |
| 123 | |
| 124 | if (c >= '0' && c <= '9') |
| 125 | c -= '0'; |
| 126 | else if (c >= 'a' && c <= 'f') |
| 127 | c -= 'a' - 10; |
| 128 | else if (c >= 'A' && c <= 'F') |
| 129 | c -= 'A' - 10; |
| 130 | else |
| 131 | return -1; |
| 132 | |
| 133 | if (a == -1) |
| 134 | a = c; |
| 135 | else |
| 136 | b = c; |
| 137 | |
| 138 | if (b == -1) |
| 139 | continue; |
| 140 | |
| 141 | code = (a << 4) | b; |
| 142 | a = b = -1; |
| 143 | if (len >= size) |
| 144 | return -2; |
| 145 | |
| 146 | bin[len] = code; |
| 147 | len++; |
| 148 | } |
| 149 | if (a >= 0 || b >= 0) |
| 150 | return -3; |
| 151 | return len; |
| 152 | } |
| 153 | |
| 154 | int main(int argc, char **argv) |
| 155 | { |
| 156 | struct hpack_dht *dht; |
| 157 | struct http_hdr list[MAX_HDR_NUM]; |
Willy Tarreau | 318adf4 | 2020-05-22 12:09:16 +0200 | [diff] [blame] | 158 | struct pool_head pool; |
Willy Tarreau | 4576424 | 2017-12-30 13:18:25 +0100 | [diff] [blame] | 159 | int outlen; |
| 160 | int dht_size = 4096; |
| 161 | int len, idx; |
| 162 | int line; |
| 163 | |
| 164 | /* first arg: dht size */ |
| 165 | if (argc > 1) { |
| 166 | dht_size = atoi(argv[1]); |
| 167 | argv++; argc--; |
| 168 | } |
| 169 | |
Willy Tarreau | 318adf4 | 2020-05-22 12:09:16 +0200 | [diff] [blame] | 170 | pool.size = dht_size; |
| 171 | pool_head_hpack_tbl = &pool; |
| 172 | dht = hpack_dht_alloc(); |
Willy Tarreau | 4576424 | 2017-12-30 13:18:25 +0100 | [diff] [blame] | 173 | if (!dht) { |
| 174 | die(1, "cannot initialize dht\n"); |
| 175 | return 1; |
| 176 | } |
| 177 | |
| 178 | for (line = 1; fgets(hex, sizeof(hex), stdin); line++) { |
| 179 | len = hex2bin(hex, buf, sizeof(buf)); |
| 180 | if (len <= 0) |
| 181 | continue; |
| 182 | printf("###### line %d : frame len=%d #######\n", line, len); |
| 183 | debug_hexdump(stdout, " ", (const char *)buf, 0, len); |
| 184 | |
| 185 | outlen = hpack_decode_frame(dht, buf, len, list, |
| 186 | sizeof(list)/sizeof(list[0]), &tmp); |
| 187 | if (outlen <= 0) { |
| 188 | printf(" HPACK decoding failed: %d\n", outlen); |
| 189 | continue; |
| 190 | } |
| 191 | |
| 192 | printf("<<< Found %d headers :\n", outlen); |
| 193 | for (idx = 0; idx < outlen - 1; idx++) { |
| 194 | //printf(" \e[1;34m%s\e[0m: ", |
| 195 | // list[idx].n.ptr ? istpad(trash.str, list[idx].n).ptr : h2_phdr_to_str(list[idx].n.len)); |
| 196 | |
| 197 | //printf("\e[1;35m%s\e[0m\n", istpad(trash.str, list[idx].v).ptr); |
| 198 | |
| 199 | printf(" %s: ", list[idx].n.ptr ? |
Willy Tarreau | 843b7cb | 2018-07-13 10:54:26 +0200 | [diff] [blame] | 200 | istpad(trash.area, list[idx].n).ptr : |
Willy Tarreau | 4576424 | 2017-12-30 13:18:25 +0100 | [diff] [blame] | 201 | h2_phdr_to_str(list[idx].n.len)); |
| 202 | |
Willy Tarreau | 843b7cb | 2018-07-13 10:54:26 +0200 | [diff] [blame] | 203 | printf("%s [n=(%p,%d) v=(%p,%d)]\n", |
| 204 | istpad(trash.area, list[idx].v).ptr, |
Willy Tarreau | 4576424 | 2017-12-30 13:18:25 +0100 | [diff] [blame] | 205 | list[idx].n.ptr, (int)list[idx].n.len, list[idx].v.ptr, (int)list[idx].v.len); |
| 206 | } |
| 207 | puts(">>>"); |
| 208 | #ifdef DEBUG_HPACK |
| 209 | printf("<<=== DHT dump [ptr=%p]:\n", dht); |
| 210 | hpack_dht_dump(stdout, dht); |
| 211 | puts("===>>"); |
| 212 | #endif |
| 213 | } |
| 214 | return 0; |
| 215 | } |