| 1 |
#include "EXTERN.h" |
| 2 |
#include "perl.h" |
| 3 |
#include "XSUB.h" |
| 4 |
|
| 5 |
#include "assert.h" |
| 6 |
#include "string.h" |
| 7 |
#include "stdlib.h" |
| 8 |
|
| 9 |
#define F_ASCII 0x00000001 |
| 10 |
#define F_UTF8 0x00000002 |
| 11 |
#define F_INDENT 0x00000004 |
| 12 |
#define F_CANONICAL 0x00000008 |
| 13 |
#define F_SPACE_BEFORE 0x00000010 |
| 14 |
#define F_SPACE_AFTER 0x00000020 |
| 15 |
#define F_ALLOW_NONREF 0x00000080 |
| 16 |
#define F_SHRINK 0x00000100 |
| 17 |
|
| 18 |
// F_SKIPINVALID? |
| 19 |
// F_EXECCODEREF? |
| 20 |
// F_SELFCONVERT? <=> { &__class__ => } |
| 21 |
|
| 22 |
#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
| 23 |
#define F_DEFAULT 0 |
| 24 |
|
| 25 |
#define INIT_SIZE 32 // initial scalar size to be allocated |
| 26 |
#define INDENT_STEP 3 // spaces per indentation level |
| 27 |
|
| 28 |
#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character |
| 29 |
#define SHORT_STRING_LEN 512 // special-case strings of up to this size |
| 30 |
|
| 31 |
#define SB do { |
| 32 |
#define SE } while (0) |
| 33 |
|
| 34 |
static HV *json_stash; // JSON::XS:: |
| 35 |
|
| 36 |
///////////////////////////////////////////////////////////////////////////// |
| 37 |
// utility functions |
| 38 |
|
| 39 |
static UV * |
| 40 |
SvJSON (SV *sv) |
| 41 |
{ |
| 42 |
if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) |
| 43 |
croak ("object is not of type JSON::XS"); |
| 44 |
|
| 45 |
return &SvUVX (SvRV (sv)); |
| 46 |
} |
| 47 |
|
| 48 |
static void |
| 49 |
shrink (SV *sv) |
| 50 |
{ |
| 51 |
sv_utf8_downgrade (sv, 1); |
| 52 |
if (SvLEN (sv) > SvCUR (sv) + 1) |
| 53 |
{ |
| 54 |
#ifdef SvPV_shrink_to_cur |
| 55 |
SvPV_shrink_to_cur (sv); |
| 56 |
#elif defined (SvPV_renew) |
| 57 |
SvPV_renew (sv, SvCUR (sv) + 1); |
| 58 |
#endif |
| 59 |
} |
| 60 |
} |
| 61 |
|
| 62 |
// decode an utf-8 character and return it, or (UV)-1 in |
| 63 |
// case of an error. |
| 64 |
// we special-case "safe" characters from U+80 .. U+7FF, |
| 65 |
// but use the very good perl function to parse anything else. |
| 66 |
// note that we never call this function for a ascii codepoints |
| 67 |
static UV |
| 68 |
decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
| 69 |
{ |
| 70 |
if (s[0] > 0xdf || s[0] < 0xc2) |
| 71 |
return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
| 72 |
else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
| 73 |
{ |
| 74 |
*clen = 2; |
| 75 |
return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
| 76 |
} |
| 77 |
else |
| 78 |
return (UV)-1; |
| 79 |
} |
| 80 |
|
| 81 |
///////////////////////////////////////////////////////////////////////////// |
| 82 |
// encoder |
| 83 |
|
| 84 |
// structure used for encoding JSON |
| 85 |
typedef struct |
| 86 |
{ |
| 87 |
char *cur; // SvPVX (sv) + current output position |
| 88 |
char *end; // SvEND (sv) |
| 89 |
SV *sv; // result scalar |
| 90 |
UV flags; // F_* |
| 91 |
int indent; // indentation level |
| 92 |
int max_depth; // max. recursion level |
| 93 |
} enc_t; |
| 94 |
|
| 95 |
static void |
| 96 |
need (enc_t *enc, STRLEN len) |
| 97 |
{ |
| 98 |
if (enc->cur + len >= enc->end) |
| 99 |
{ |
| 100 |
STRLEN cur = enc->cur - SvPVX (enc->sv); |
| 101 |
SvGROW (enc->sv, cur + len + 1); |
| 102 |
enc->cur = SvPVX (enc->sv) + cur; |
| 103 |
enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); |
| 104 |
} |
| 105 |
} |
| 106 |
|
| 107 |
static void |
| 108 |
encode_ch (enc_t *enc, char ch) |
| 109 |
{ |
| 110 |
need (enc, 1); |
| 111 |
*enc->cur++ = ch; |
| 112 |
} |
| 113 |
|
| 114 |
static void |
| 115 |
encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) |
| 116 |
{ |
| 117 |
char *end = str + len; |
| 118 |
|
| 119 |
need (enc, len); |
| 120 |
|
| 121 |
while (str < end) |
| 122 |
{ |
| 123 |
unsigned char ch = *(unsigned char *)str; |
| 124 |
|
| 125 |
if (ch >= 0x20 && ch < 0x80) // most common case |
| 126 |
{ |
| 127 |
if (ch == '"') // but with slow exceptions |
| 128 |
{ |
| 129 |
need (enc, len += 1); |
| 130 |
*enc->cur++ = '\\'; |
| 131 |
*enc->cur++ = '"'; |
| 132 |
} |
| 133 |
else if (ch == '\\') |
| 134 |
{ |
| 135 |
need (enc, len += 1); |
| 136 |
*enc->cur++ = '\\'; |
| 137 |
*enc->cur++ = '\\'; |
| 138 |
} |
| 139 |
else |
| 140 |
*enc->cur++ = ch; |
| 141 |
|
| 142 |
++str; |
| 143 |
} |
| 144 |
else |
| 145 |
{ |
| 146 |
switch (ch) |
| 147 |
{ |
| 148 |
case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; |
| 149 |
case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; |
| 150 |
case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; |
| 151 |
case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; |
| 152 |
case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; |
| 153 |
|
| 154 |
default: |
| 155 |
{ |
| 156 |
STRLEN clen; |
| 157 |
UV uch; |
| 158 |
|
| 159 |
if (is_utf8) |
| 160 |
{ |
| 161 |
//uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); |
| 162 |
uch = decode_utf8 (str, end - str, &clen); |
| 163 |
if (clen == (STRLEN)-1) |
| 164 |
croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
| 165 |
} |
| 166 |
else |
| 167 |
{ |
| 168 |
uch = ch; |
| 169 |
clen = 1; |
| 170 |
} |
| 171 |
|
| 172 |
if (uch > 0x10FFFFUL) |
| 173 |
croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
| 174 |
|
| 175 |
if (uch < 0x80 || enc->flags & F_ASCII) |
| 176 |
{ |
| 177 |
if (uch > 0xFFFFUL) |
| 178 |
{ |
| 179 |
need (enc, len += 11); |
| 180 |
sprintf (enc->cur, "\\u%04x\\u%04x", |
| 181 |
(int)((uch - 0x10000) / 0x400 + 0xD800), |
| 182 |
(int)((uch - 0x10000) % 0x400 + 0xDC00)); |
| 183 |
enc->cur += 12; |
| 184 |
} |
| 185 |
else |
| 186 |
{ |
| 187 |
static char hexdigit [16] = "0123456789abcdef"; |
| 188 |
need (enc, len += 5); |
| 189 |
*enc->cur++ = '\\'; |
| 190 |
*enc->cur++ = 'u'; |
| 191 |
*enc->cur++ = hexdigit [ uch >> 12 ]; |
| 192 |
*enc->cur++ = hexdigit [(uch >> 8) & 15]; |
| 193 |
*enc->cur++ = hexdigit [(uch >> 4) & 15]; |
| 194 |
*enc->cur++ = hexdigit [(uch >> 0) & 15]; |
| 195 |
} |
| 196 |
|
| 197 |
str += clen; |
| 198 |
} |
| 199 |
else if (is_utf8) |
| 200 |
{ |
| 201 |
need (enc, len += clen); |
| 202 |
do |
| 203 |
{ |
| 204 |
*enc->cur++ = *str++; |
| 205 |
} |
| 206 |
while (--clen); |
| 207 |
} |
| 208 |
else |
| 209 |
{ |
| 210 |
need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed |
| 211 |
enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
| 212 |
++str; |
| 213 |
} |
| 214 |
} |
| 215 |
} |
| 216 |
} |
| 217 |
|
| 218 |
--len; |
| 219 |
} |
| 220 |
} |
| 221 |
|
| 222 |
static void |
| 223 |
encode_indent (enc_t *enc) |
| 224 |
{ |
| 225 |
if (enc->flags & F_INDENT) |
| 226 |
{ |
| 227 |
int spaces = enc->indent * INDENT_STEP; |
| 228 |
|
| 229 |
need (enc, spaces); |
| 230 |
memset (enc->cur, ' ', spaces); |
| 231 |
enc->cur += spaces; |
| 232 |
} |
| 233 |
} |
| 234 |
|
| 235 |
static void |
| 236 |
encode_space (enc_t *enc) |
| 237 |
{ |
| 238 |
need (enc, 1); |
| 239 |
encode_ch (enc, ' '); |
| 240 |
} |
| 241 |
|
| 242 |
static void |
| 243 |
encode_nl (enc_t *enc) |
| 244 |
{ |
| 245 |
if (enc->flags & F_INDENT) |
| 246 |
{ |
| 247 |
need (enc, 1); |
| 248 |
encode_ch (enc, '\n'); |
| 249 |
} |
| 250 |
} |
| 251 |
|
| 252 |
static void |
| 253 |
encode_comma (enc_t *enc) |
| 254 |
{ |
| 255 |
encode_ch (enc, ','); |
| 256 |
|
| 257 |
if (enc->flags & F_INDENT) |
| 258 |
encode_nl (enc); |
| 259 |
else if (enc->flags & F_SPACE_AFTER) |
| 260 |
encode_space (enc); |
| 261 |
} |
| 262 |
|
| 263 |
static void encode_sv (enc_t *enc, SV *sv); |
| 264 |
|
| 265 |
static void |
| 266 |
encode_av (enc_t *enc, AV *av) |
| 267 |
{ |
| 268 |
int i, len = av_len (av); |
| 269 |
|
| 270 |
encode_ch (enc, '['); encode_nl (enc); |
| 271 |
++enc->indent; |
| 272 |
|
| 273 |
for (i = 0; i <= len; ++i) |
| 274 |
{ |
| 275 |
encode_indent (enc); |
| 276 |
encode_sv (enc, *av_fetch (av, i, 0)); |
| 277 |
|
| 278 |
if (i < len) |
| 279 |
encode_comma (enc); |
| 280 |
} |
| 281 |
|
| 282 |
encode_nl (enc); |
| 283 |
|
| 284 |
--enc->indent; |
| 285 |
encode_indent (enc); encode_ch (enc, ']'); |
| 286 |
} |
| 287 |
|
| 288 |
static void |
| 289 |
encode_he (enc_t *enc, HE *he) |
| 290 |
{ |
| 291 |
encode_ch (enc, '"'); |
| 292 |
|
| 293 |
if (HeKLEN (he) == HEf_SVKEY) |
| 294 |
{ |
| 295 |
SV *sv = HeSVKEY (he); |
| 296 |
STRLEN len; |
| 297 |
char *str; |
| 298 |
|
| 299 |
SvGETMAGIC (sv); |
| 300 |
str = SvPV (sv, len); |
| 301 |
|
| 302 |
encode_str (enc, str, len, SvUTF8 (sv)); |
| 303 |
} |
| 304 |
else |
| 305 |
encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
| 306 |
|
| 307 |
encode_ch (enc, '"'); |
| 308 |
|
| 309 |
if (enc->flags & F_SPACE_BEFORE) encode_space (enc); |
| 310 |
encode_ch (enc, ':'); |
| 311 |
if (enc->flags & F_SPACE_AFTER ) encode_space (enc); |
| 312 |
encode_sv (enc, HeVAL (he)); |
| 313 |
} |
| 314 |
|
| 315 |
// compare hash entries, used when all keys are bytestrings |
| 316 |
static int |
| 317 |
he_cmp_fast (const void *a_, const void *b_) |
| 318 |
{ |
| 319 |
int cmp; |
| 320 |
|
| 321 |
HE *a = *(HE **)a_; |
| 322 |
HE *b = *(HE **)b_; |
| 323 |
|
| 324 |
STRLEN la = HeKLEN (a); |
| 325 |
STRLEN lb = HeKLEN (b); |
| 326 |
|
| 327 |
if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) |
| 328 |
cmp = la - lb; |
| 329 |
|
| 330 |
return cmp; |
| 331 |
} |
| 332 |
|
| 333 |
// compare hash entries, used when some keys are sv's or utf-x |
| 334 |
static int |
| 335 |
he_cmp_slow (const void *a, const void *b) |
| 336 |
{ |
| 337 |
return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); |
| 338 |
} |
| 339 |
|
| 340 |
static void |
| 341 |
encode_hv (enc_t *enc, HV *hv) |
| 342 |
{ |
| 343 |
int count, i; |
| 344 |
|
| 345 |
encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; |
| 346 |
|
| 347 |
if ((count = hv_iterinit (hv))) |
| 348 |
{ |
| 349 |
// for canonical output we have to sort by keys first |
| 350 |
// actually, this is mostly due to the stupid so-called |
| 351 |
// security workaround added somewhere in 5.8.x. |
| 352 |
// that randomises hash orderings |
| 353 |
if (enc->flags & F_CANONICAL) |
| 354 |
{ |
| 355 |
HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode |
| 356 |
int fast = 1; |
| 357 |
|
| 358 |
i = 0; |
| 359 |
while ((he = hv_iternext (hv))) |
| 360 |
{ |
| 361 |
hes [i++] = he; |
| 362 |
if (HeKLEN (he) < 0 || HeKUTF8 (he)) |
| 363 |
fast = 0; |
| 364 |
} |
| 365 |
|
| 366 |
assert (i == count); |
| 367 |
|
| 368 |
if (fast) |
| 369 |
qsort (hes, count, sizeof (HE *), he_cmp_fast); |
| 370 |
else |
| 371 |
{ |
| 372 |
// hack to forcefully disable "use bytes" |
| 373 |
COP cop = *PL_curcop; |
| 374 |
cop.op_private = 0; |
| 375 |
|
| 376 |
ENTER; |
| 377 |
SAVETMPS; |
| 378 |
|
| 379 |
SAVEVPTR (PL_curcop); |
| 380 |
PL_curcop = &cop; |
| 381 |
|
| 382 |
qsort (hes, count, sizeof (HE *), he_cmp_slow); |
| 383 |
|
| 384 |
FREETMPS; |
| 385 |
LEAVE; |
| 386 |
} |
| 387 |
|
| 388 |
for (i = 0; i < count; ++i) |
| 389 |
{ |
| 390 |
encode_indent (enc); |
| 391 |
encode_he (enc, hes [i]); |
| 392 |
|
| 393 |
if (i < count - 1) |
| 394 |
encode_comma (enc); |
| 395 |
} |
| 396 |
|
| 397 |
encode_nl (enc); |
| 398 |
} |
| 399 |
else |
| 400 |
{ |
| 401 |
SV *sv; |
| 402 |
HE *he = hv_iternext (hv); |
| 403 |
|
| 404 |
for (;;) |
| 405 |
{ |
| 406 |
encode_indent (enc); |
| 407 |
encode_he (enc, he); |
| 408 |
|
| 409 |
if (!(he = hv_iternext (hv))) |
| 410 |
break; |
| 411 |
|
| 412 |
encode_comma (enc); |
| 413 |
} |
| 414 |
|
| 415 |
encode_nl (enc); |
| 416 |
} |
| 417 |
} |
| 418 |
|
| 419 |
--enc->indent; encode_indent (enc); encode_ch (enc, '}'); |
| 420 |
} |
| 421 |
|
| 422 |
static void |
| 423 |
encode_sv (enc_t *enc, SV *sv) |
| 424 |
{ |
| 425 |
SvGETMAGIC (sv); |
| 426 |
|
| 427 |
if (SvPOKp (sv)) |
| 428 |
{ |
| 429 |
STRLEN len; |
| 430 |
char *str = SvPV (sv, len); |
| 431 |
encode_ch (enc, '"'); |
| 432 |
encode_str (enc, str, len, SvUTF8 (sv)); |
| 433 |
encode_ch (enc, '"'); |
| 434 |
} |
| 435 |
else if (SvNOKp (sv)) |
| 436 |
{ |
| 437 |
need (enc, NV_DIG + 32); |
| 438 |
Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
| 439 |
enc->cur += strlen (enc->cur); |
| 440 |
} |
| 441 |
else if (SvIOKp (sv)) |
| 442 |
{ |
| 443 |
need (enc, 64); |
| 444 |
enc->cur += |
| 445 |
SvIsUV(sv) |
| 446 |
? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) |
| 447 |
: snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); |
| 448 |
} |
| 449 |
else if (SvROK (sv)) |
| 450 |
{ |
| 451 |
SV *rv = SvRV (sv); |
| 452 |
|
| 453 |
if (enc->indent >= enc->max_depth) |
| 454 |
croak ("data structure too deep (hit recursion limit)"); |
| 455 |
|
| 456 |
switch (SvTYPE (rv)) |
| 457 |
{ |
| 458 |
case SVt_PVAV: encode_av (enc, (AV *)rv); break; |
| 459 |
case SVt_PVHV: encode_hv (enc, (HV *)rv); break; |
| 460 |
|
| 461 |
default: |
| 462 |
croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
| 463 |
SvPV_nolen (sv)); |
| 464 |
} |
| 465 |
} |
| 466 |
else if (!SvOK (sv)) |
| 467 |
encode_str (enc, "null", 4, 0); |
| 468 |
else |
| 469 |
croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
| 470 |
SvPV_nolen (sv), SvFLAGS (sv)); |
| 471 |
} |
| 472 |
|
| 473 |
static SV * |
| 474 |
encode_json (SV *scalar, UV flags) |
| 475 |
{ |
| 476 |
if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
| 477 |
croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
| 478 |
|
| 479 |
enc_t enc; |
| 480 |
enc.flags = flags; |
| 481 |
enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
| 482 |
enc.cur = SvPVX (enc.sv); |
| 483 |
enc.end = SvEND (enc.sv); |
| 484 |
enc.indent = 0; |
| 485 |
enc.max_depth = 0x7fffffffUL; |
| 486 |
|
| 487 |
SvPOK_only (enc.sv); |
| 488 |
encode_sv (&enc, scalar); |
| 489 |
|
| 490 |
if (!(flags & (F_ASCII | F_UTF8))) |
| 491 |
SvUTF8_on (enc.sv); |
| 492 |
|
| 493 |
SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
| 494 |
|
| 495 |
if (enc.flags & F_SHRINK) |
| 496 |
shrink (enc.sv); |
| 497 |
|
| 498 |
return enc.sv; |
| 499 |
} |
| 500 |
|
| 501 |
///////////////////////////////////////////////////////////////////////////// |
| 502 |
// decoder |
| 503 |
|
| 504 |
// structure used for decoding JSON |
| 505 |
typedef struct |
| 506 |
{ |
| 507 |
char *cur; // current parser pointer |
| 508 |
char *end; // end of input string |
| 509 |
const char *err; // parse error, if != 0 |
| 510 |
UV flags; // F_* |
| 511 |
} dec_t; |
| 512 |
|
| 513 |
static void |
| 514 |
decode_ws (dec_t *dec) |
| 515 |
{ |
| 516 |
for (;;) |
| 517 |
{ |
| 518 |
char ch = *dec->cur; |
| 519 |
|
| 520 |
if (ch > 0x20 |
| 521 |
|| (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) |
| 522 |
break; |
| 523 |
|
| 524 |
++dec->cur; |
| 525 |
} |
| 526 |
} |
| 527 |
|
| 528 |
#define ERR(reason) SB dec->err = reason; goto fail; SE |
| 529 |
#define EXPECT_CH(ch) SB \ |
| 530 |
if (*dec->cur != ch) \ |
| 531 |
ERR (# ch " expected"); \ |
| 532 |
++dec->cur; \ |
| 533 |
SE |
| 534 |
|
| 535 |
static SV *decode_sv (dec_t *dec); |
| 536 |
|
| 537 |
static signed char decode_hexdigit[256]; |
| 538 |
|
| 539 |
static UV |
| 540 |
decode_4hex (dec_t *dec) |
| 541 |
{ |
| 542 |
signed char d1, d2, d3, d4; |
| 543 |
unsigned char *cur = (unsigned char *)dec->cur; |
| 544 |
|
| 545 |
d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); |
| 546 |
d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); |
| 547 |
d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); |
| 548 |
d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); |
| 549 |
|
| 550 |
dec->cur += 4; |
| 551 |
|
| 552 |
return ((UV)d1) << 12 |
| 553 |
| ((UV)d2) << 8 |
| 554 |
| ((UV)d3) << 4 |
| 555 |
| ((UV)d4); |
| 556 |
|
| 557 |
fail: |
| 558 |
return (UV)-1; |
| 559 |
} |
| 560 |
|
| 561 |
static SV * |
| 562 |
decode_str (dec_t *dec) |
| 563 |
{ |
| 564 |
SV *sv = 0; |
| 565 |
int utf8 = 0; |
| 566 |
|
| 567 |
do |
| 568 |
{ |
| 569 |
char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; |
| 570 |
char *cur = buf; |
| 571 |
|
| 572 |
do |
| 573 |
{ |
| 574 |
unsigned char ch = *(unsigned char *)dec->cur++; |
| 575 |
|
| 576 |
if (ch == '"') |
| 577 |
{ |
| 578 |
--dec->cur; |
| 579 |
break; |
| 580 |
} |
| 581 |
else if (ch == '\\') |
| 582 |
{ |
| 583 |
switch (*dec->cur) |
| 584 |
{ |
| 585 |
case '\\': |
| 586 |
case '/': |
| 587 |
case '"': *cur++ = *dec->cur++; break; |
| 588 |
|
| 589 |
case 'b': ++dec->cur; *cur++ = '\010'; break; |
| 590 |
case 't': ++dec->cur; *cur++ = '\011'; break; |
| 591 |
case 'n': ++dec->cur; *cur++ = '\012'; break; |
| 592 |
case 'f': ++dec->cur; *cur++ = '\014'; break; |
| 593 |
case 'r': ++dec->cur; *cur++ = '\015'; break; |
| 594 |
|
| 595 |
case 'u': |
| 596 |
{ |
| 597 |
UV lo, hi; |
| 598 |
++dec->cur; |
| 599 |
|
| 600 |
hi = decode_4hex (dec); |
| 601 |
if (hi == (UV)-1) |
| 602 |
goto fail; |
| 603 |
|
| 604 |
// possibly a surrogate pair |
| 605 |
if (hi >= 0xd800) |
| 606 |
if (hi < 0xdc00) |
| 607 |
{ |
| 608 |
if (dec->cur [0] != '\\' || dec->cur [1] != 'u') |
| 609 |
ERR ("missing low surrogate character in surrogate pair"); |
| 610 |
|
| 611 |
dec->cur += 2; |
| 612 |
|
| 613 |
lo = decode_4hex (dec); |
| 614 |
if (lo == (UV)-1) |
| 615 |
goto fail; |
| 616 |
|
| 617 |
if (lo < 0xdc00 || lo >= 0xe000) |
| 618 |
ERR ("surrogate pair expected"); |
| 619 |
|
| 620 |
hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; |
| 621 |
} |
| 622 |
else if (hi < 0xe000) |
| 623 |
ERR ("missing high surrogate character in surrogate pair"); |
| 624 |
|
| 625 |
if (hi >= 0x80) |
| 626 |
{ |
| 627 |
utf8 = 1; |
| 628 |
|
| 629 |
cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); |
| 630 |
} |
| 631 |
else |
| 632 |
*cur++ = hi; |
| 633 |
} |
| 634 |
break; |
| 635 |
|
| 636 |
default: |
| 637 |
--dec->cur; |
| 638 |
ERR ("illegal backslash escape sequence in string"); |
| 639 |
} |
| 640 |
} |
| 641 |
else if (ch >= 0x20 && ch <= 0x7f) |
| 642 |
*cur++ = ch; |
| 643 |
else if (ch >= 0x80) |
| 644 |
{ |
| 645 |
--dec->cur; |
| 646 |
|
| 647 |
STRLEN clen; |
| 648 |
UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); |
| 649 |
if (clen == (STRLEN)-1) |
| 650 |
ERR ("malformed UTF-8 character in JSON string"); |
| 651 |
|
| 652 |
do |
| 653 |
{ |
| 654 |
*cur++ = *dec->cur++; |
| 655 |
} |
| 656 |
while (--clen); |
| 657 |
|
| 658 |
utf8 = 1; |
| 659 |
} |
| 660 |
else if (!ch) |
| 661 |
ERR ("unexpected end of string while parsing json string"); |
| 662 |
else |
| 663 |
ERR ("invalid character encountered"); |
| 664 |
|
| 665 |
} |
| 666 |
while (cur < buf + SHORT_STRING_LEN); |
| 667 |
|
| 668 |
STRLEN len = cur - buf; |
| 669 |
|
| 670 |
if (sv) |
| 671 |
{ |
| 672 |
SvGROW (sv, SvCUR (sv) + len + 1); |
| 673 |
memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
| 674 |
SvCUR_set (sv, SvCUR (sv) + len); |
| 675 |
} |
| 676 |
else |
| 677 |
sv = newSVpvn (buf, len); |
| 678 |
} |
| 679 |
while (*dec->cur != '"'); |
| 680 |
|
| 681 |
++dec->cur; |
| 682 |
|
| 683 |
if (sv) |
| 684 |
{ |
| 685 |
SvPOK_only (sv); |
| 686 |
*SvEND (sv) = 0; |
| 687 |
|
| 688 |
if (utf8) |
| 689 |
SvUTF8_on (sv); |
| 690 |
} |
| 691 |
else |
| 692 |
sv = newSVpvn ("", 0); |
| 693 |
|
| 694 |
return sv; |
| 695 |
|
| 696 |
fail: |
| 697 |
return 0; |
| 698 |
} |
| 699 |
|
| 700 |
static SV * |
| 701 |
decode_num (dec_t *dec) |
| 702 |
{ |
| 703 |
int is_nv = 0; |
| 704 |
char *start = dec->cur; |
| 705 |
|
| 706 |
// [minus] |
| 707 |
if (*dec->cur == '-') |
| 708 |
++dec->cur; |
| 709 |
|
| 710 |
if (*dec->cur == '0') |
| 711 |
{ |
| 712 |
++dec->cur; |
| 713 |
if (*dec->cur >= '0' && *dec->cur <= '9') |
| 714 |
ERR ("malformed number (leading zero must not be followed by another digit)"); |
| 715 |
} |
| 716 |
else if (*dec->cur < '0' || *dec->cur > '9') |
| 717 |
ERR ("malformed number (no digits after initial minus)"); |
| 718 |
else |
| 719 |
do |
| 720 |
{ |
| 721 |
++dec->cur; |
| 722 |
} |
| 723 |
while (*dec->cur >= '0' && *dec->cur <= '9'); |
| 724 |
|
| 725 |
// [frac] |
| 726 |
if (*dec->cur == '.') |
| 727 |
{ |
| 728 |
++dec->cur; |
| 729 |
|
| 730 |
if (*dec->cur < '0' || *dec->cur > '9') |
| 731 |
ERR ("malformed number (no digits after decimal point)"); |
| 732 |
|
| 733 |
do |
| 734 |
{ |
| 735 |
++dec->cur; |
| 736 |
} |
| 737 |
while (*dec->cur >= '0' && *dec->cur <= '9'); |
| 738 |
|
| 739 |
is_nv = 1; |
| 740 |
} |
| 741 |
|
| 742 |
// [exp] |
| 743 |
if (*dec->cur == 'e' || *dec->cur == 'E') |
| 744 |
{ |
| 745 |
++dec->cur; |
| 746 |
|
| 747 |
if (*dec->cur == '-' || *dec->cur == '+') |
| 748 |
++dec->cur; |
| 749 |
|
| 750 |
if (*dec->cur < '0' || *dec->cur > '9') |
| 751 |
ERR ("malformed number (no digits after exp sign)"); |
| 752 |
|
| 753 |
do |
| 754 |
{ |
| 755 |
++dec->cur; |
| 756 |
} |
| 757 |
while (*dec->cur >= '0' && *dec->cur <= '9'); |
| 758 |
|
| 759 |
is_nv = 1; |
| 760 |
} |
| 761 |
|
| 762 |
if (!is_nv) |
| 763 |
{ |
| 764 |
UV uv; |
| 765 |
int numtype = grok_number (start, dec->cur - start, &uv); |
| 766 |
if (numtype & IS_NUMBER_IN_UV) |
| 767 |
if (numtype & IS_NUMBER_NEG) |
| 768 |
{ |
| 769 |
if (uv < (UV)IV_MIN) |
| 770 |
return newSViv (-(IV)uv); |
| 771 |
} |
| 772 |
else |
| 773 |
return newSVuv (uv); |
| 774 |
} |
| 775 |
|
| 776 |
return newSVnv (Atof (start)); |
| 777 |
|
| 778 |
fail: |
| 779 |
return 0; |
| 780 |
} |
| 781 |
|
| 782 |
static SV * |
| 783 |
decode_av (dec_t *dec) |
| 784 |
{ |
| 785 |
AV *av = newAV (); |
| 786 |
|
| 787 |
decode_ws (dec); |
| 788 |
if (*dec->cur == ']') |
| 789 |
++dec->cur; |
| 790 |
else |
| 791 |
for (;;) |
| 792 |
{ |
| 793 |
SV *value; |
| 794 |
|
| 795 |
value = decode_sv (dec); |
| 796 |
if (!value) |
| 797 |
goto fail; |
| 798 |
|
| 799 |
av_push (av, value); |
| 800 |
|
| 801 |
decode_ws (dec); |
| 802 |
|
| 803 |
if (*dec->cur == ']') |
| 804 |
{ |
| 805 |
++dec->cur; |
| 806 |
break; |
| 807 |
} |
| 808 |
|
| 809 |
if (*dec->cur != ',') |
| 810 |
ERR (", or ] expected while parsing array"); |
| 811 |
|
| 812 |
++dec->cur; |
| 813 |
} |
| 814 |
|
| 815 |
return newRV_noinc ((SV *)av); |
| 816 |
|
| 817 |
fail: |
| 818 |
SvREFCNT_dec (av); |
| 819 |
return 0; |
| 820 |
} |
| 821 |
|
| 822 |
static SV * |
| 823 |
decode_hv (dec_t *dec) |
| 824 |
{ |
| 825 |
HV *hv = newHV (); |
| 826 |
|
| 827 |
decode_ws (dec); |
| 828 |
if (*dec->cur == '}') |
| 829 |
++dec->cur; |
| 830 |
else |
| 831 |
for (;;) |
| 832 |
{ |
| 833 |
SV *key, *value; |
| 834 |
|
| 835 |
decode_ws (dec); EXPECT_CH ('"'); |
| 836 |
|
| 837 |
key = decode_str (dec); |
| 838 |
if (!key) |
| 839 |
goto fail; |
| 840 |
|
| 841 |
decode_ws (dec); EXPECT_CH (':'); |
| 842 |
|
| 843 |
value = decode_sv (dec); |
| 844 |
if (!value) |
| 845 |
{ |
| 846 |
SvREFCNT_dec (key); |
| 847 |
goto fail; |
| 848 |
} |
| 849 |
|
| 850 |
//TODO: optimise |
| 851 |
hv_store_ent (hv, key, value, 0); |
| 852 |
|
| 853 |
decode_ws (dec); |
| 854 |
|
| 855 |
if (*dec->cur == '}') |
| 856 |
{ |
| 857 |
++dec->cur; |
| 858 |
break; |
| 859 |
} |
| 860 |
|
| 861 |
if (*dec->cur != ',') |
| 862 |
ERR (", or } expected while parsing object/hash"); |
| 863 |
|
| 864 |
++dec->cur; |
| 865 |
} |
| 866 |
|
| 867 |
return newRV_noinc ((SV *)hv); |
| 868 |
|
| 869 |
fail: |
| 870 |
SvREFCNT_dec (hv); |
| 871 |
return 0; |
| 872 |
} |
| 873 |
|
| 874 |
static SV * |
| 875 |
decode_sv (dec_t *dec) |
| 876 |
{ |
| 877 |
decode_ws (dec); |
| 878 |
switch (*dec->cur) |
| 879 |
{ |
| 880 |
case '"': ++dec->cur; return decode_str (dec); |
| 881 |
case '[': ++dec->cur; return decode_av (dec); |
| 882 |
case '{': ++dec->cur; return decode_hv (dec); |
| 883 |
|
| 884 |
case '-': |
| 885 |
case '0': case '1': case '2': case '3': case '4': |
| 886 |
case '5': case '6': case '7': case '8': case '9': |
| 887 |
return decode_num (dec); |
| 888 |
|
| 889 |
case 't': |
| 890 |
if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
| 891 |
{ |
| 892 |
dec->cur += 4; |
| 893 |
return newSViv (1); |
| 894 |
} |
| 895 |
else |
| 896 |
ERR ("'true' expected"); |
| 897 |
|
| 898 |
break; |
| 899 |
|
| 900 |
case 'f': |
| 901 |
if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
| 902 |
{ |
| 903 |
dec->cur += 5; |
| 904 |
return newSViv (0); |
| 905 |
} |
| 906 |
else |
| 907 |
ERR ("'false' expected"); |
| 908 |
|
| 909 |
break; |
| 910 |
|
| 911 |
case 'n': |
| 912 |
if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
| 913 |
{ |
| 914 |
dec->cur += 4; |
| 915 |
return newSVsv (&PL_sv_undef); |
| 916 |
} |
| 917 |
else |
| 918 |
ERR ("'null' expected"); |
| 919 |
|
| 920 |
break; |
| 921 |
|
| 922 |
default: |
| 923 |
ERR ("malformed json string, neither array, object, number, string or atom"); |
| 924 |
break; |
| 925 |
} |
| 926 |
|
| 927 |
fail: |
| 928 |
return 0; |
| 929 |
} |
| 930 |
|
| 931 |
static SV * |
| 932 |
decode_json (SV *string, UV flags) |
| 933 |
{ |
| 934 |
SV *sv; |
| 935 |
|
| 936 |
if (flags & F_UTF8) |
| 937 |
sv_utf8_downgrade (string, 0); |
| 938 |
else |
| 939 |
sv_utf8_upgrade (string); |
| 940 |
|
| 941 |
SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
| 942 |
|
| 943 |
dec_t dec; |
| 944 |
dec.flags = flags; |
| 945 |
dec.cur = SvPVX (string); |
| 946 |
dec.end = SvEND (string); |
| 947 |
dec.err = 0; |
| 948 |
|
| 949 |
sv = decode_sv (&dec); |
| 950 |
|
| 951 |
if (!sv) |
| 952 |
{ |
| 953 |
IV offset = dec.flags & F_UTF8 |
| 954 |
? dec.cur - SvPVX (string) |
| 955 |
: utf8_distance (dec.cur, SvPVX (string)); |
| 956 |
SV *uni = sv_newmortal (); |
| 957 |
|
| 958 |
// horrible hack to silence warning inside pv_uni_display |
| 959 |
COP cop = *PL_curcop; |
| 960 |
cop.cop_warnings = pWARN_NONE; |
| 961 |
ENTER; |
| 962 |
SAVEVPTR (PL_curcop); |
| 963 |
PL_curcop = &cop; |
| 964 |
pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
| 965 |
LEAVE; |
| 966 |
|
| 967 |
croak ("%s, at character offset %d (%s)", |
| 968 |
dec.err, |
| 969 |
(int)offset, |
| 970 |
dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
| 971 |
} |
| 972 |
|
| 973 |
sv = sv_2mortal (sv); |
| 974 |
|
| 975 |
if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
| 976 |
croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
| 977 |
|
| 978 |
return sv; |
| 979 |
} |
| 980 |
|
| 981 |
///////////////////////////////////////////////////////////////////////////// |
| 982 |
// XS interface functions |
| 983 |
|
| 984 |
MODULE = JSON::XS PACKAGE = JSON::XS |
| 985 |
|
| 986 |
BOOT: |
| 987 |
{ |
| 988 |
int i; |
| 989 |
|
| 990 |
memset (decode_hexdigit, 0xff, 256); |
| 991 |
for (i = 10; i--; ) |
| 992 |
decode_hexdigit ['0' + i] = i; |
| 993 |
|
| 994 |
for (i = 7; i--; ) |
| 995 |
{ |
| 996 |
decode_hexdigit ['a' + i] = 10 + i; |
| 997 |
decode_hexdigit ['A' + i] = 10 + i; |
| 998 |
} |
| 999 |
|
| 1000 |
json_stash = gv_stashpv ("JSON::XS", 1); |
| 1001 |
} |
| 1002 |
|
| 1003 |
PROTOTYPES: DISABLE |
| 1004 |
|
| 1005 |
SV *new (char *dummy) |
| 1006 |
CODE: |
| 1007 |
RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); |
| 1008 |
OUTPUT: |
| 1009 |
RETVAL |
| 1010 |
|
| 1011 |
SV *ascii (SV *self, int enable = 1) |
| 1012 |
ALIAS: |
| 1013 |
ascii = F_ASCII |
| 1014 |
utf8 = F_UTF8 |
| 1015 |
indent = F_INDENT |
| 1016 |
canonical = F_CANONICAL |
| 1017 |
space_before = F_SPACE_BEFORE |
| 1018 |
space_after = F_SPACE_AFTER |
| 1019 |
pretty = F_PRETTY |
| 1020 |
allow_nonref = F_ALLOW_NONREF |
| 1021 |
shrink = F_SHRINK |
| 1022 |
CODE: |
| 1023 |
{ |
| 1024 |
UV *uv = SvJSON (self); |
| 1025 |
if (enable) |
| 1026 |
*uv |= ix; |
| 1027 |
else |
| 1028 |
*uv &= ~ix; |
| 1029 |
|
| 1030 |
RETVAL = newSVsv (self); |
| 1031 |
} |
| 1032 |
OUTPUT: |
| 1033 |
RETVAL |
| 1034 |
|
| 1035 |
void encode (SV *self, SV *scalar) |
| 1036 |
PPCODE: |
| 1037 |
XPUSHs (encode_json (scalar, *SvJSON (self))); |
| 1038 |
|
| 1039 |
void decode (SV *self, SV *jsonstr) |
| 1040 |
PPCODE: |
| 1041 |
XPUSHs (decode_json (jsonstr, *SvJSON (self))); |
| 1042 |
|
| 1043 |
PROTOTYPES: ENABLE |
| 1044 |
|
| 1045 |
void to_json (SV *scalar) |
| 1046 |
ALIAS: |
| 1047 |
objToJson = 0 |
| 1048 |
PPCODE: |
| 1049 |
XPUSHs (encode_json (scalar, F_UTF8)); |
| 1050 |
|
| 1051 |
void from_json (SV *jsonstr) |
| 1052 |
ALIAS: |
| 1053 |
jsonToObj = 0 |
| 1054 |
PPCODE: |
| 1055 |
XPUSHs (decode_json (jsonstr, F_UTF8)); |
| 1056 |
|
