| 1 | #include "data_lang/j8.h"
|
| 2 |
|
| 3 | #include <string>
|
| 4 |
|
| 5 | #include "data_lang/j8_test_lib.h"
|
| 6 | #include "vendor/greatest.h"
|
| 7 |
|
| 8 | // Naive buffer
|
| 9 | struct Buf {
|
| 10 | unsigned char* data;
|
| 11 | int capacity;
|
| 12 | int len;
|
| 13 | };
|
| 14 |
|
| 15 | void EncodeNaive(char* s, int n, Buf* buf, int j8_fallback) {
|
| 16 | char* orig_s = s; // save for rewinding
|
| 17 |
|
| 18 | unsigned char* in = (unsigned char*)s;
|
| 19 | unsigned char* input_end = (unsigned char*)s + n;
|
| 20 |
|
| 21 | unsigned char* out = buf->data; // mutated
|
| 22 | unsigned char* orig_out = buf->data; // not mutated
|
| 23 |
|
| 24 | unsigned char** p_out = &out; // for J8_OUT()
|
| 25 |
|
| 26 | J8_OUT('"');
|
| 27 | // printf("*in %p *out %p\n", *in, *out);
|
| 28 |
|
| 29 | int invalid_utf8 = 0;
|
| 30 | while (in < input_end) {
|
| 31 | // printf("1 in %p *out %p\n", in, *out);
|
| 32 |
|
| 33 | // TODO: check *out vs. capacity and maybe grow buffer
|
| 34 | invalid_utf8 = J8EncodeOne(&in, &out, 0); // JSON escaping
|
| 35 |
|
| 36 | // Try again with J8 escaping
|
| 37 | if (invalid_utf8 && j8_fallback) {
|
| 38 | in = (unsigned char*)orig_s;
|
| 39 | out = orig_out;
|
| 40 |
|
| 41 | J8_OUT('b');
|
| 42 | J8_OUT('\'');
|
| 43 |
|
| 44 | // TODO: check *out vs. capacity and maybe grow buffer
|
| 45 |
|
| 46 | while (in < input_end) {
|
| 47 | // printf("2 in %p *out %p\n", in, *out);
|
| 48 | J8EncodeOne(&in, &out, 1); // Now with J8 escaping
|
| 49 | }
|
| 50 |
|
| 51 | J8_OUT('\'');
|
| 52 | buf->len = out - orig_out;
|
| 53 | return;
|
| 54 | }
|
| 55 | }
|
| 56 |
|
| 57 | J8_OUT('"');
|
| 58 | buf->len = out - orig_out;
|
| 59 | }
|
| 60 |
|
| 61 | void EncodeBString(char* s, int n, std::string* result) {
|
| 62 | uint8_t* in = reinterpret_cast<uint8_t*>(s);
|
| 63 | uint8_t* in_end = in + n;
|
| 64 |
|
| 65 | result->append("b'");
|
| 66 |
|
| 67 | while (in < in_end) {
|
| 68 | int chunk_pos = result->size(); // current position
|
| 69 |
|
| 70 | // Same logic as EncodeBString()
|
| 71 | int chunk_size = in_end - in + 3; // 3 for the quotes
|
| 72 | // clamp it to account for tiny gaps and huge strings
|
| 73 | if (chunk_size < J8_MIN_CAPACITY) {
|
| 74 | chunk_size = J8_MIN_CAPACITY;
|
| 75 | } else if (chunk_size > 4096) {
|
| 76 | chunk_size = 4096;
|
| 77 | }
|
| 78 | printf("\t[2] in %p chunk %d\n", in, chunk_size);
|
| 79 | result->append(chunk_size, '\0'); // "pre-allocated" bytes to overwrite
|
| 80 |
|
| 81 | // Need C-style pointers to call the helper function
|
| 82 | uint8_t* raw_data = (uint8_t*)result->data();
|
| 83 |
|
| 84 | uint8_t* out = raw_data + chunk_pos;
|
| 85 | uint8_t* orig_out = out;
|
| 86 |
|
| 87 | uint8_t* out_end = raw_data + result->size();
|
| 88 |
|
| 89 | // printf("\tEncodeChunk JSON\n");
|
| 90 | J8EncodeChunk(&in, in_end, &out, out_end, true);
|
| 91 |
|
| 92 | int bytes_this_chunk = out - orig_out;
|
| 93 | int end_index = chunk_pos + bytes_this_chunk;
|
| 94 | printf("\t bytes_this_chunk %d\n", bytes_this_chunk);
|
| 95 | printf("\t end_index %d\n", end_index);
|
| 96 |
|
| 97 | result->erase(end_index, std::string::npos);
|
| 98 | }
|
| 99 | result->append("'");
|
| 100 | }
|
| 101 |
|
| 102 | void EncodeString(char* s, int n, std::string* result, int j8_fallback) {
|
| 103 | uint8_t* in = reinterpret_cast<uint8_t*>(s);
|
| 104 | uint8_t* in_end = in + n;
|
| 105 |
|
| 106 | int begin_index = result->size(); // position before writing opening quote
|
| 107 |
|
| 108 | result->append("\"");
|
| 109 |
|
| 110 | printf("\t***str len %d\n", n);
|
| 111 |
|
| 112 | while (in < in_end) {
|
| 113 | int chunk_pos = result->size(); // current position
|
| 114 |
|
| 115 | // Compute chunk size assuming that we'll output about 5 bytes "foo" for
|
| 116 | // the string foo. Cases like \u{1f}\u{1e} blow it up by a factor of 6, in
|
| 117 | // which case we'll make more trips through the loop.
|
| 118 | int chunk_size = in_end - in + 3; // 3 for the quotes
|
| 119 | // clamp it to account for tiny gaps and huge strings
|
| 120 | if (chunk_size < J8_MIN_CAPACITY) {
|
| 121 | chunk_size = J8_MIN_CAPACITY;
|
| 122 | } else if (chunk_size > 4096) {
|
| 123 | chunk_size = 4096;
|
| 124 | }
|
| 125 | printf("\t[1] in %p chunk %d\n", in, chunk_size);
|
| 126 |
|
| 127 | result->append(chunk_size, '\0'); // "pre-allocated" bytes to overwrite
|
| 128 |
|
| 129 | // Need C-style pointers to call the helper function
|
| 130 | uint8_t* raw_data = (uint8_t*)result->data();
|
| 131 |
|
| 132 | uint8_t* out = raw_data + chunk_pos;
|
| 133 | uint8_t* orig_out = out;
|
| 134 |
|
| 135 | uint8_t* out_end = raw_data + result->size();
|
| 136 |
|
| 137 | // printf("\tEncodeChunk JSON\n");
|
| 138 | int invalid_utf8 = J8EncodeChunk(&in, in_end, &out, out_end, false);
|
| 139 | if (invalid_utf8 && j8_fallback) {
|
| 140 | // printf("RETRY\n");
|
| 141 | result->erase(begin_index, std::string::npos);
|
| 142 | EncodeBString(s, n, result); // fall back to b''
|
| 143 | printf("\t[1] result len %d\n", static_cast<int>(result->size()));
|
| 144 | return;
|
| 145 | }
|
| 146 |
|
| 147 | int bytes_this_chunk = out - orig_out;
|
| 148 | int end_index = chunk_pos + bytes_this_chunk;
|
| 149 | printf("\t bytes_this_chunk %d\n", bytes_this_chunk);
|
| 150 | printf("\t end_index %d\n", end_index);
|
| 151 |
|
| 152 | result->erase(end_index, std::string::npos);
|
| 153 | }
|
| 154 | result->append("\"");
|
| 155 | printf("\t[1] result len %d\n", static_cast<int>(result->size()));
|
| 156 | }
|
| 157 |
|
| 158 | void EncodeAndPrint(char* s, int n, int j8_fallback) {
|
| 159 | #if 0
|
| 160 | Buf buf = {0};
|
| 161 | buf.data = (unsigned char*)malloc(64);
|
| 162 | buf.capacity = 64;
|
| 163 |
|
| 164 | EncodeNaive(s, n, &buf, j8_fallback);
|
| 165 | buf.data[buf.len] = '\0'; // NUL terminate
|
| 166 |
|
| 167 | printf("out = %s\n", buf.data);
|
| 168 | free(buf.data);
|
| 169 | #else
|
| 170 |
|
| 171 | std::string result;
|
| 172 | EncodeString(s, n, &result, j8_fallback);
|
| 173 | printf("out = %s\n", result.c_str());
|
| 174 |
|
| 175 | #endif
|
| 176 | }
|
| 177 |
|
| 178 | TEST encode_test() {
|
| 179 | #if 1
|
| 180 | const char* mixed = "hi \x01 \u4000\xfe\u4001\xff\xfd ' \" new \n \\ \u03bc";
|
| 181 | EncodeAndPrint(const_cast<char*>(mixed), strlen(mixed), 0);
|
| 182 | EncodeAndPrint(const_cast<char*>(mixed), strlen(mixed), 1);
|
| 183 | #endif
|
| 184 |
|
| 185 | const char* a = "ab";
|
| 186 | EncodeAndPrint(const_cast<char*>(a), strlen(a), 0);
|
| 187 | EncodeAndPrint(const_cast<char*>(a), strlen(a), 1);
|
| 188 |
|
| 189 | const char* b = "0123456789";
|
| 190 | EncodeAndPrint(const_cast<char*>(b), strlen(b), 0);
|
| 191 | EncodeAndPrint(const_cast<char*>(b), strlen(b), 1);
|
| 192 |
|
| 193 | const char* u = "hi \u4000 \u03bc";
|
| 194 | EncodeAndPrint(const_cast<char*>(b), strlen(u), 0);
|
| 195 | EncodeAndPrint(const_cast<char*>(b), strlen(u), 1);
|
| 196 |
|
| 197 | // Internal NUL
|
| 198 | const char* bin = "\x00\x01\xff";
|
| 199 | EncodeAndPrint(const_cast<char*>(bin), 3, 0);
|
| 200 | EncodeAndPrint(const_cast<char*>(bin), 3, 1);
|
| 201 |
|
| 202 | // Blow up size
|
| 203 | const char* blowup =
|
| 204 | "\x01\x02\x03\x04\x05\x06\x07\x08\x09\x0A\x0B\x0C\x0D\x0e\x0f\x10\xfe";
|
| 205 | EncodeAndPrint(const_cast<char*>(blowup), strlen(blowup), 0);
|
| 206 | EncodeAndPrint(const_cast<char*>(blowup), strlen(blowup), 1);
|
| 207 |
|
| 208 | PASS();
|
| 209 | }
|
| 210 |
|
| 211 | GREATEST_MAIN_DEFS();
|
| 212 |
|
| 213 | int main(int argc, char** argv) {
|
| 214 | GREATEST_MAIN_BEGIN();
|
| 215 |
|
| 216 | RUN_TEST(encode_test);
|
| 217 |
|
| 218 | GREATEST_MAIN_END();
|
| 219 | return 0;
|
| 220 | }
|