|
| 1 | +// Copyright 2013 Yuriy Iskra <iskra.yw@gmail.com> |
| 2 | + |
| 3 | +#include <string.h> |
| 4 | +#include <ctype.h> |
| 5 | +#include <assert.h> |
| 6 | +#include <errno.h> |
| 7 | +#include "erl_nif.h" |
| 8 | +#include "jsonx_str.h" |
| 9 | + |
| 10 | +typedef struct{ |
| 11 | + ERL_NIF_TERM am_true; |
| 12 | + ERL_NIF_TERM am_false; |
| 13 | + ERL_NIF_TERM am_ok; |
| 14 | + |
| 15 | + ERL_NIF_TERM am_null; |
| 16 | + ERL_NIF_TERM am_start_map; |
| 17 | + ERL_NIF_TERM am_map_key; |
| 18 | + ERL_NIF_TERM am_end_map; |
| 19 | + ERL_NIF_TERM am_start_array; |
| 20 | + ERL_NIF_TERM am_end_array; |
| 21 | + ERL_NIF_TERM am_parse_buf; |
| 22 | + ERL_NIF_TERM am_parse_end; |
| 23 | + |
| 24 | + ERL_NIF_TERM am_error; |
| 25 | + ERL_NIF_TERM am_erange; |
| 26 | + ERL_NIF_TERM am_estr; |
| 27 | + ERL_NIF_TERM am_esyntax; |
| 28 | +}PrivData; |
| 29 | + |
| 30 | +#define ERROR 0 |
| 31 | +#define START 1 |
| 32 | +#define COMPLETTE 2 |
| 33 | +#define COMMA 3 |
| 34 | +#define KEY_COMPLETTE 4 |
| 35 | + |
| 36 | +typedef struct{ |
| 37 | + ErlNifEnv *env; |
| 38 | + ErlNifBinary bin; |
| 39 | + unsigned char *cur; |
| 40 | + unsigned char *top; |
| 41 | + PrivData *priv; |
| 42 | + int m_state; |
| 43 | +}State; |
| 44 | + |
| 45 | +ErlNifResourceType* stream_RSTYPE; |
| 46 | + |
| 47 | +static void |
| 48 | +stream_rt_dtor(ErlNifEnv* env, void* obj){ |
| 49 | + State *entry = (void*)obj; |
| 50 | + enif_release_binary(&entry->bin); |
| 51 | + entry->bin.data = NULL; |
| 52 | + entry->bin.size = 0; |
| 53 | +} |
| 54 | + |
| 55 | +static int |
| 56 | +load(ErlNifEnv* env, void** priv_data, ERL_NIF_TERM load_info){ |
| 57 | + |
| 58 | + PrivData *pdata = enif_alloc(sizeof(PrivData)); |
| 59 | + if(pdata == NULL) return 1; |
| 60 | + |
| 61 | + stream_RSTYPE = enif_open_resource_type(env, NULL, |
| 62 | + "stream_RSTYPE", |
| 63 | + stream_rt_dtor, |
| 64 | + ERL_NIF_RT_CREATE, NULL); |
| 65 | + if (stream_RSTYPE == NULL) return 1; |
| 66 | + |
| 67 | + if(!enif_make_existing_atom(env, "true", &(pdata->am_true), ERL_NIF_LATIN1)) return 1; |
| 68 | + if(!enif_make_existing_atom(env, "false", &(pdata->am_false), ERL_NIF_LATIN1)) return 1; |
| 69 | + if(!enif_make_existing_atom(env, "ok", &(pdata->am_ok), ERL_NIF_LATIN1)) return 1; |
| 70 | + |
| 71 | + if(!enif_make_existing_atom(env, "null", &(pdata->am_null), ERL_NIF_LATIN1)) return 1; |
| 72 | + if(!enif_make_existing_atom(env, "start_map", &(pdata->am_start_map), ERL_NIF_LATIN1)) return 1; |
| 73 | + if(!enif_make_existing_atom(env, "map_key", &(pdata->am_map_key), ERL_NIF_LATIN1)) return 1; |
| 74 | + if(!enif_make_existing_atom(env, "end_map", &(pdata->am_end_map), ERL_NIF_LATIN1)) return 1; |
| 75 | + if(!enif_make_existing_atom(env, "start_array", &(pdata->am_start_array), ERL_NIF_LATIN1)) return 1; |
| 76 | + if(!enif_make_existing_atom(env, "end_array", &(pdata->am_end_array), ERL_NIF_LATIN1)) return 1; |
| 77 | + if(!enif_make_existing_atom(env, "parse_buf", &(pdata->am_parse_buf), ERL_NIF_LATIN1)) return 1; |
| 78 | + if(!enif_make_existing_atom(env, "parse_end", &(pdata->am_parse_end), ERL_NIF_LATIN1)) return 1; |
| 79 | + |
| 80 | + if(!enif_make_existing_atom(env, "error", &(pdata->am_error), ERL_NIF_LATIN1)) return 1; |
| 81 | + if(!enif_make_existing_atom(env, "big_num", &(pdata->am_erange), ERL_NIF_LATIN1)) return 1; |
| 82 | + if(!enif_make_existing_atom(env, "invalid_string", &(pdata->am_estr), ERL_NIF_LATIN1)) return 1; |
| 83 | + if(!enif_make_existing_atom(env, "invalid_json", &(pdata->am_esyntax), ERL_NIF_LATIN1)) return 1; |
| 84 | + |
| 85 | + *priv_data = (void*)pdata; |
| 86 | + |
| 87 | + return 0; |
| 88 | +} |
| 89 | + |
| 90 | +static int |
| 91 | +reload(ErlNifEnv* env, void** priv_data, ERL_NIF_TERM load_info){ |
| 92 | + return 0; |
| 93 | +} |
| 94 | + |
| 95 | +static int |
| 96 | +upgrade(ErlNifEnv* env, void** priv_data, void** old_priv_data, ERL_NIF_TERM load_info){ |
| 97 | + return 0; |
| 98 | +} |
| 99 | +static void |
| 100 | +unload(ErlNifEnv* env, void* priv_data){ |
| 101 | + return; |
| 102 | +} |
| 103 | + |
| 104 | + |
| 105 | +static inline unsigned char |
| 106 | +look_ah(State *st){ |
| 107 | + while(isspace(*st->cur)) |
| 108 | + st->cur++; |
| 109 | + return *(st->cur); |
| 110 | +} |
| 111 | + |
| 112 | +static inline ERL_NIF_TERM |
| 113 | +make_error(State* st, ERL_NIF_TERM resume){ |
| 114 | + st->m_state = ERROR; |
| 115 | + *st->top = 'e'; |
| 116 | + return enif_make_tuple2(st->env, st->priv->am_error, resume); |
| 117 | +} |
| 118 | + |
| 119 | +static inline ERL_NIF_TERM |
| 120 | +parse_string(State* st){ |
| 121 | + unsigned char *endptr; |
| 122 | + unsigned char *endstr; |
| 123 | + unsigned char *dst; |
| 124 | + ERL_NIF_TERM ret; |
| 125 | + if(check_noescaped_jstr(st->cur, &endptr)){ |
| 126 | + if(*endptr == '"'){ |
| 127 | + dst = enif_make_new_binary(st->env, endptr - st->cur - 1, &ret); |
| 128 | + memcpy(dst, st->cur + 1, endptr - st->cur - 1); |
| 129 | + st->cur = endptr + 1; |
| 130 | + return ret; |
| 131 | + }else if(*endptr == '\\'){ |
| 132 | + if(check_with_unescape_jstr(endptr, &endstr, &endptr)){ |
| 133 | + dst = enif_make_new_binary(st->env, endstr - st->cur - 1, &ret); |
| 134 | + memcpy(dst, st->cur + 1, endstr - st->cur - 1); |
| 135 | + st->cur = endptr + 1; |
| 136 | + return ret; |
| 137 | + } |
| 138 | + } |
| 139 | + } |
| 140 | + return make_error(st, st->priv->am_estr); |
| 141 | +} |
| 142 | + |
| 143 | +static inline ERL_NIF_TERM |
| 144 | +parse_number(State *st){ |
| 145 | + long long int_num; |
| 146 | + double float_num; |
| 147 | + char *endptr; |
| 148 | + errno = 0; |
| 149 | + |
| 150 | + int_num = strtoll((char *)st->cur, &endptr, 10); |
| 151 | + if((char*)st->cur == endptr){ |
| 152 | + return (ERL_NIF_TERM)0; |
| 153 | + }else if(errno == ERANGE){ |
| 154 | + return make_error(st, st->priv->am_erange); |
| 155 | + } |
| 156 | + |
| 157 | + if(*endptr == '.' || *endptr == 'e' || *endptr == 'E'){ |
| 158 | + float_num = strtod((char *)st->cur, &endptr); |
| 159 | + if(errno != ERANGE){ |
| 160 | + st->cur = (unsigned char*)endptr; |
| 161 | + return enif_make_double(st->env, float_num); |
| 162 | + } |
| 163 | + else{ |
| 164 | + return make_error(st, st->priv->am_erange); |
| 165 | + } |
| 166 | + } |
| 167 | + else{ |
| 168 | + st->cur = (unsigned char*)endptr; |
| 169 | + return enif_make_int64(st->env, int_num); |
| 170 | + } |
| 171 | +} |
| 172 | + |
| 173 | +static inline ERL_NIF_TERM |
| 174 | +parse_true(State* st){ |
| 175 | + if(!(strncmp("rue", (char*)(++st->cur), 3))){ |
| 176 | + st->cur = st->cur + 3; |
| 177 | + return st->priv->am_true; |
| 178 | + } |
| 179 | + return make_error(st, st->priv->am_esyntax); |
| 180 | +} |
| 181 | + |
| 182 | +static inline ERL_NIF_TERM |
| 183 | +parse_false(State* st){ |
| 184 | + if(!(strncmp("alse", (char*)(++st->cur), 4))){ |
| 185 | + st->cur = st->cur + 4; |
| 186 | + return st->priv->am_false; |
| 187 | + } |
| 188 | + return make_error(st, st->priv->am_esyntax); |
| 189 | +} |
| 190 | + |
| 191 | +static inline ERL_NIF_TERM |
| 192 | +parse_null(State* st){ |
| 193 | + if(!(strncmp("ull", (char*)(++st->cur), 3))){ |
| 194 | + st->cur = st->cur + 3; |
| 195 | + return st->priv->am_null; |
| 196 | + } |
| 197 | + return make_error(st, st->priv->am_esyntax); |
| 198 | +} |
| 199 | + |
| 200 | + |
| 201 | +static inline ERL_NIF_TERM |
| 202 | +parse_key(State *st){ |
| 203 | + ERL_NIF_TERM key; |
| 204 | + switch(look_ah(st)){ |
| 205 | + case '\"' : |
| 206 | + key = parse_string(st); |
| 207 | + if(st->m_state == ERROR) |
| 208 | + return key; |
| 209 | + if(look_ah(st) == ':'){ |
| 210 | + st->m_state = KEY_COMPLETTE; |
| 211 | + st->cur++; |
| 212 | + return enif_make_tuple2(st->env, st->priv->am_map_key, key); |
| 213 | + } |
| 214 | + case '\0' : |
| 215 | + return st->priv->am_parse_buf; |
| 216 | + default : |
| 217 | + return make_error(st, st->priv->am_esyntax); |
| 218 | + } |
| 219 | +} |
| 220 | + |
| 221 | +static inline ERL_NIF_TERM |
| 222 | +parse0(State *st){ |
| 223 | + switch(look_ah(st)){ |
| 224 | + case '\"' : st->m_state = COMPLETTE; return parse_string(st); |
| 225 | + case 't' : st->m_state = COMPLETTE; return parse_true(st); |
| 226 | + case 'f' : st->m_state = COMPLETTE; return parse_false(st); |
| 227 | + case 'n' : st->m_state = COMPLETTE; return parse_null(st); |
| 228 | + case '-' : case '0' : case '1' : case '2' : |
| 229 | + case '3' : case '4' : case '5' : case '6' : |
| 230 | + case '7' : case '8' : case '9' : |
| 231 | + st->m_state = COMPLETTE; return parse_number(st); |
| 232 | + case '[' : |
| 233 | + st->m_state = START; |
| 234 | + *(++st->top) = '['; st->cur++; |
| 235 | + return st->priv->am_start_array; |
| 236 | + case '{' : |
| 237 | + st->m_state = START; |
| 238 | + *(++st->top) = '{'; st->cur++; |
| 239 | + return st->priv->am_start_map; |
| 240 | + case '\0' : |
| 241 | + return st->priv->am_parse_buf; |
| 242 | + default: |
| 243 | + st->m_state = ERROR; |
| 244 | + *(st->top) = 'e'; |
| 245 | + return make_error(st, st->priv->am_esyntax); |
| 246 | + } |
| 247 | +} |
| 248 | + |
| 249 | +ERL_NIF_TERM |
| 250 | +get_event_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]){ |
| 251 | + State *st; |
| 252 | + assert(enif_get_resource(env, argv[0], stream_RSTYPE, (void**)&st)); |
| 253 | + switch(*st->top){ |
| 254 | + case '[' : |
| 255 | + if(st->m_state == COMPLETTE){ |
| 256 | + switch(look_ah(st)){ |
| 257 | + case ',' : st->m_state = COMMA; st->cur++; return parse0(st); |
| 258 | + case '\0': st->cur++; return st->priv->am_parse_buf; |
| 259 | + case ']' : st->m_state = COMPLETTE; st->cur++; st->top--; return st->priv->am_end_array; |
| 260 | + } |
| 261 | + }else if(st->m_state == COMMA){ |
| 262 | + return parse0(st); |
| 263 | + }else if(st->m_state == START){ |
| 264 | + switch(look_ah(st)){ |
| 265 | + case '\0': st->cur++; return st->priv->am_parse_buf; |
| 266 | + case ']' : st->m_state = COMPLETTE; st->cur++; st->top--; return st->priv->am_end_array; |
| 267 | + default: return parse0(st); |
| 268 | + } |
| 269 | + } |
| 270 | + break; |
| 271 | + case '{' : |
| 272 | + if(st->m_state == COMPLETTE){ |
| 273 | + switch(look_ah(st)){ |
| 274 | + case ',' : st->m_state = COMMA; st->cur++; return parse_key(st); |
| 275 | + case '}' : st->m_state = COMPLETTE; st->cur++; st->top--; return st->priv->am_end_map; |
| 276 | + case '\0': st->cur++; return st->priv->am_parse_buf; |
| 277 | + } |
| 278 | + }else if(st->m_state == KEY_COMPLETTE){ |
| 279 | + return parse0(st); |
| 280 | + }else if(st->m_state == COMMA){ |
| 281 | + return parse_key(st); |
| 282 | + }else if(st->m_state == START){ |
| 283 | + switch(look_ah(st)){ |
| 284 | + case '\0': st->cur++; return st->priv->am_parse_buf; |
| 285 | + case '}' : st->m_state = COMPLETTE; st->cur++; st->top--; return st->priv->am_end_map; |
| 286 | + default: return parse_key(st); |
| 287 | + } |
| 288 | + } |
| 289 | + break; |
| 290 | + case '\0': |
| 291 | + if(st->m_state == COMPLETTE){ |
| 292 | + ERL_NIF_TERM rest; |
| 293 | + look_ah(st); |
| 294 | + size_t size = strlen((char*)st->cur); |
| 295 | + unsigned char *nb = enif_make_new_binary(env, size, &rest); |
| 296 | + memcpy(nb, st->cur, size); |
| 297 | + return enif_make_tuple2(env, st->priv->am_parse_end, rest); |
| 298 | + return st->priv->am_parse_end; |
| 299 | + }else{ |
| 300 | + return parse0(st); |
| 301 | + } |
| 302 | + break; |
| 303 | + case 'e': |
| 304 | + default : |
| 305 | + break; |
| 306 | + } |
| 307 | + return make_error(st, st->priv->am_esyntax); |
| 308 | +} |
| 309 | + |
| 310 | +ERL_NIF_TERM |
| 311 | +update_decoder_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]){ |
| 312 | + State *st; |
| 313 | + ErlNifBinary input; |
| 314 | + if(!enif_inspect_binary(env, argv[1], &input)){ |
| 315 | + return enif_make_badarg(env); |
| 316 | + } |
| 317 | + assert(enif_get_resource(env, argv[0], stream_RSTYPE, (void**)&st)); |
| 318 | + size_t stack_size = (st->top + 1 - st->bin.data); |
| 319 | + |
| 320 | + size_t free = st->bin.size - stack_size - sizeof(ERL_NIF_TERM); |
| 321 | + if(input.size > free){ |
| 322 | + enif_realloc_binary(&st->bin, input.size + stack_size + sizeof(ERL_NIF_TERM)); |
| 323 | + st->top = st->bin.data + stack_size - 1; |
| 324 | + st->cur = st->top + 1; |
| 325 | + memcpy(st->cur, input.data, input.size); |
| 326 | + *((ERL_NIF_TERM*)(st->cur + input.size)) = 0U; |
| 327 | + }else{ |
| 328 | + st->cur = st->top + 1; |
| 329 | + memcpy(st->cur, input.data, input.size); |
| 330 | + *((ERL_NIF_TERM*)(st->cur + input.size)) = 0U; |
| 331 | + } |
| 332 | + return st->priv->am_ok; |
| 333 | +} |
| 334 | + |
| 335 | +ERL_NIF_TERM |
| 336 | +make_stream_resource_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]){ |
| 337 | + ErlNifBinary input; |
| 338 | + if(!enif_inspect_binary(env, argv[0], &input)){ |
| 339 | + return enif_make_badarg(env); |
| 340 | + } |
| 341 | + size_t alloc_size = input.size + 2 * sizeof(ERL_NIF_TERM); |
| 342 | + State *st = (State*)enif_alloc_resource(stream_RSTYPE, sizeof(State)); |
| 343 | + st->env = env; |
| 344 | + st->priv = (PrivData*)enif_priv_data(env); |
| 345 | + st->m_state = START; |
| 346 | + |
| 347 | + enif_alloc_binary(alloc_size , &st->bin); |
| 348 | + st->cur = st->bin.data + sizeof(ERL_NIF_TERM); |
| 349 | + memcpy(st->cur, input.data, input.size); |
| 350 | + *((ERL_NIF_TERM*)st->bin.data) = 0U; |
| 351 | + *((ERL_NIF_TERM*)(st->bin.data + input.size + sizeof(ERL_NIF_TERM))) = 0U; |
| 352 | + st->top = st->cur - 1; |
| 353 | + |
| 354 | + ERL_NIF_TERM ret = enif_make_resource(env, (void *)st); |
| 355 | + enif_release_resource(st); |
| 356 | + return ret; |
| 357 | +} |
| 358 | + |
| 359 | +static ErlNifFunc |
| 360 | +nif_funcs[] = { |
| 361 | + {"new_decoder", 1, make_stream_resource_nif}, |
| 362 | + {"update_decoder", 2, update_decoder_nif}, |
| 363 | + {"get_event", 1, get_event_nif} |
| 364 | +}; |
| 365 | + |
| 366 | +ERL_NIF_INIT(jstream, nif_funcs, load, reload, upgrade, unload); |
| 367 | + |
0 commit comments