ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/JSON-XS/XS.xs
(Generate patch)

Comparing JSON-XS/XS.xs (file contents):
Revision 1.127 by root, Fri Feb 26 21:46:45 2016 UTC vs.
Revision 1.137 by root, Sun Feb 24 04:21:05 2019 UTC

… …
6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
12#if defined(__BORLANDC__) || defined(_MSC_VER) 13#if defined(__BORLANDC__) || defined(_MSC_VER)
13# define snprintf _snprintf // C compilers have this in stdio.h 14# define snprintf _snprintf // C compilers have this in stdio.h
14#endif 15#endif
15 16
… …
49#define F_ALLOW_TAGS 0x00004000UL 50#define F_ALLOW_TAGS 0x00004000UL
50#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 51#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
51 52
52#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 53#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
53 54
54#define INIT_SIZE 32 // initial scalar size to be allocated 55#define INIT_SIZE 64 // initial scalar size to be allocated
55#define INDENT_STEP 3 // spaces per indentation level 56#define INDENT_STEP 3 // spaces per indentation level
56 57
57#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 58#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
58 59
59#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8) 60#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
… …
77 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 78 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
78 79
79#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)" 80#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
80 81
81#ifdef USE_ITHREADS 82#ifdef USE_ITHREADS
82# define JSON_SLOW 1
83# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) 83# define JSON_STASH (expect_true (json_stash) ? json_stash : gv_stashpv ("JSON::XS", 1))
84# define BOOL_STASH (bool_stash ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1)) 84# define BOOL_STASH (expect_true (bool_stash) ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1))
85# define GET_BOOL(value) (expect_true (bool_ ## value) ? bool_ ## value : get_bool ("Types::Serialiser::" # value))
85#else 86#else
86# define JSON_SLOW 0
87# define JSON_STASH json_stash 87# define JSON_STASH json_stash
88# define BOOL_STASH bool_stash 88# define BOOL_STASH bool_stash
89# define GET_BOOL(value) bool_ ## value
89#endif 90#endif
90 91
91// the amount of HEs to allocate on the stack, when sorting keys 92// the amount of HEs to allocate on the stack, when sorting keys
92#define STACK_HES 64 93#define STACK_HES 64
93 94
94static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean:: 95static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean::
95static SV *bool_true, *bool_false, *sv_json; 96static SV *bool_false, *bool_true;
97static SV *sv_json;
96 98
97enum { 99enum {
98 INCR_M_WS = 0, // initial whitespace skipping, must be 0 100 INCR_M_WS = 0, // initial whitespace skipping, must be 0
101 INCR_M_TFN, // inside true/false/null
102 INCR_M_NUM, // inside number
99 INCR_M_STR, // inside string 103 INCR_M_STR, // inside string
100 INCR_M_BS, // inside backslash 104 INCR_M_BS, // inside backslash
101 INCR_M_C0, // inside comment in initial whitespace sequence 105 INCR_M_C0, // inside comment in initial whitespace sequence
102 INCR_M_C1, // inside comment in other places 106 INCR_M_C1, // inside comment in other places
103 INCR_M_JSON // outside anything, count nesting 107 INCR_M_JSON // outside anything, count nesting
… …
116 // for the incremental parser 120 // for the incremental parser
117 SV *incr_text; // the source text so far 121 SV *incr_text; // the source text so far
118 STRLEN incr_pos; // the current offset into the text 122 STRLEN incr_pos; // the current offset into the text
119 int incr_nest; // {[]}-nesting level 123 int incr_nest; // {[]}-nesting level
120 unsigned char incr_mode; 124 unsigned char incr_mode;
125
126 SV *v_false, *v_true;
121} JSON; 127} JSON;
122 128
123INLINE void 129INLINE void
124json_init (JSON *json) 130json_init (JSON *json)
125{ 131{
126 Zero (json, 1, JSON); 132 static const JSON init = { F_ALLOW_NONREF, 512 };
127 json->max_depth = 512; 133
134 *json = init;
128} 135}
129 136
130///////////////////////////////////////////////////////////////////////////// 137/////////////////////////////////////////////////////////////////////////////
131// utility functions 138// utility functions
132 139
… …
154 SvPV_renew (sv, SvCUR (sv) + 1); 161 SvPV_renew (sv, SvCUR (sv) + 1);
155#endif 162#endif
156 } 163 }
157} 164}
158 165
166/* adds two STRLENs together, slow, and with paranoia */
167STRLEN
168strlen_sum (STRLEN l1, STRLEN l2)
169{
170 size_t sum = l1 + l2;
171
172 if (sum < (size_t)l2 || sum != (size_t)(STRLEN)sum)
173 croak ("JSON::XS: string size overflow");
174
175 return sum;
176}
177
178/* similar to SvGROW, but somewhat safer and guarantees exponential realloc strategy */
179static char *
180json_sv_grow (SV *sv, size_t len1, size_t len2)
181{
182 len1 = strlen_sum (len1, len2);
183 len1 = strlen_sum (len1, len1 >> 1);
184
185 if (len1 > 4096 - 24)
186 len1 = (len1 | 4095) - 24;
187
188 return SvGROW (sv, len1);
189}
190
159// decode an utf-8 character and return it, or (UV)-1 in 191// decode a utf-8 character and return it, or (UV)-1 in
160// case of an error. 192// case of an error.
161// we special-case "safe" characters from U+80 .. U+7FF, 193// we special-case "safe" characters from U+80 .. U+7FF,
162// but use the very good perl function to parse anything else. 194// but use the very good perl function to parse anything else.
163// note that we never call this function for a ascii codepoints 195// note that we never call this function for a ascii codepoints
164INLINE UV 196INLINE UV
… …
318 if (len == 1) 350 if (len == 1)
319 if (*pv == '1') 351 if (*pv == '1')
320 return 1; 352 return 1;
321 else if (*pv == '0') 353 else if (*pv == '0')
322 return 0; 354 return 0;
323
324 } 355 }
325 356
326 return -1; 357 return -1;
327} 358}
328 359
… …
333 if (!SvROK (scalar)) 364 if (!SvROK (scalar))
334 return 1; 365 return 1;
335 366
336 scalar = SvRV (scalar); 367 scalar = SvRV (scalar);
337 368
369 if (SvTYPE (scalar) >= SVt_PVMG)
370 {
338 if (SvSTASH (scalar) == bool_stash) 371 if (SvSTASH (scalar) == bool_stash)
339 return 1; 372 return 1;
340 373
341 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0) 374 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
342 return 1; 375 return 1;
376 }
343 377
344 return 0; 378 return 0;
345} 379}
346 380
347///////////////////////////////////////////////////////////////////////////// 381/////////////////////////////////////////////////////////////////////////////
… …
359} enc_t; 393} enc_t;
360 394
361INLINE void 395INLINE void
362need (enc_t *enc, STRLEN len) 396need (enc_t *enc, STRLEN len)
363{ 397{
364 if (expect_false (enc->cur + len >= enc->end)) 398 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
365 { 399 {
366 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 400 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
367 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 401 char *buf = json_sv_grow (enc->sv, cur, len);
368 enc->cur = SvPVX (enc->sv) + cur; 402 enc->cur = buf + cur;
369 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 403 enc->end = buf + SvLEN (enc->sv) - 1;
370 } 404 }
371} 405}
372 406
373INLINE void 407INLINE void
374encode_ch (enc_t *enc, char ch) 408encode_ch (enc_t *enc, char ch)
… …
390 424
391 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case 425 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
392 { 426 {
393 if (expect_false (ch == '"')) // but with slow exceptions 427 if (expect_false (ch == '"')) // but with slow exceptions
394 { 428 {
395 need (enc, len += 1); 429 need (enc, len + 1);
396 *enc->cur++ = '\\'; 430 *enc->cur++ = '\\';
397 *enc->cur++ = '"'; 431 *enc->cur++ = '"';
398 } 432 }
399 else if (expect_false (ch == '\\')) 433 else if (expect_false (ch == '\\'))
400 { 434 {
401 need (enc, len += 1); 435 need (enc, len + 1);
402 *enc->cur++ = '\\'; 436 *enc->cur++ = '\\';
403 *enc->cur++ = '\\'; 437 *enc->cur++ = '\\';
404 } 438 }
405 else 439 else
406 *enc->cur++ = ch; 440 *enc->cur++ = ch;
… …
409 } 443 }
410 else 444 else
411 { 445 {
412 switch (ch) 446 switch (ch)
413 { 447 {
414 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; 448 case '\010': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
415 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; 449 case '\011': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
416 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; 450 case '\012': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
417 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; 451 case '\014': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
418 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; 452 case '\015': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
419 453
420 default: 454 default:
421 { 455 {
422 STRLEN clen; 456 STRLEN clen;
423 UV uch; 457 UV uch;
… …
439 if (uch >= 0x10000UL) 473 if (uch >= 0x10000UL)
440 { 474 {
441 if (uch >= 0x110000UL) 475 if (uch >= 0x110000UL)
442 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 476 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
443 477
444 need (enc, len += 11); 478 need (enc, len + 11);
445 sprintf (enc->cur, "\\u%04x\\u%04x", 479 sprintf (enc->cur, "\\u%04x\\u%04x",
446 (int)((uch - 0x10000) / 0x400 + 0xD800), 480 (int)((uch - 0x10000) / 0x400 + 0xD800),
447 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 481 (int)((uch - 0x10000) % 0x400 + 0xDC00));
448 enc->cur += 12; 482 enc->cur += 12;
449 } 483 }
450 else 484 else
451 { 485 {
452 need (enc, len += 5); 486 need (enc, len + 5);
453 *enc->cur++ = '\\'; 487 *enc->cur++ = '\\';
454 *enc->cur++ = 'u'; 488 *enc->cur++ = 'u';
455 *enc->cur++ = PL_hexdigit [ uch >> 12 ]; 489 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
456 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; 490 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
457 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; 491 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
… …
465 *enc->cur++ = uch; 499 *enc->cur++ = uch;
466 str += clen; 500 str += clen;
467 } 501 }
468 else if (is_utf8) 502 else if (is_utf8)
469 { 503 {
470 need (enc, len += clen); 504 need (enc, len + clen);
471 do 505 do
472 { 506 {
473 *enc->cur++ = *str++; 507 *enc->cur++ = *str++;
474 } 508 }
475 while (--clen); 509 while (--clen);
476 } 510 }
477 else 511 else
478 { 512 {
479 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 513 need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed
480 enc->cur = encode_utf8 (enc->cur, uch); 514 enc->cur = encode_utf8 (enc->cur, uch);
481 ++str; 515 ++str;
482 } 516 }
483 } 517 }
484 } 518 }
… …
748 { 782 {
749 HV *stash = SvSTASH (sv); 783 HV *stash = SvSTASH (sv);
750 784
751 if (stash == bool_stash) 785 if (stash == bool_stash)
752 { 786 {
753 if (SvIV (sv))
754 encode_str (enc, "true", 4, 0); 787 if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
755 else
756 encode_str (enc, "false", 5, 0); 788 else encode_str (enc, "false", 5, 0);
757 } 789 }
758 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0))) 790 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
759 { 791 {
760 int count; 792 int count;
761 dSP; 793 dSP;
762 794
763 ENTER; SAVETMPS; 795 ENTER; SAVETMPS;
764 SAVESTACK_POS ();
765 PUSHMARK (SP); 796 PUSHMARK (SP);
766 EXTEND (SP, 2); 797 EXTEND (SP, 2);
767 // we re-bless the reference to get overload and other niceties right 798 // we re-bless the reference to get overload and other niceties right
768 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 799 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
769 PUSHs (sv_json); 800 PUSHs (sv_json);
… …
781 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash)); 812 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
782 encode_ch (enc, '"'); 813 encode_ch (enc, '"');
783 encode_ch (enc, ')'); 814 encode_ch (enc, ')');
784 encode_ch (enc, '['); 815 encode_ch (enc, '[');
785 816
786 while (count) 817 if (count)
787 { 818 {
819 int i;
820
821 for (i = 0; i < count - 1; ++i)
822 {
788 encode_sv (enc, SP[1 - count--]); 823 encode_sv (enc, SP[i + 1 - count]);
789
790 if (count)
791 encode_ch (enc, ','); 824 encode_ch (enc, ',');
825 }
826
827 encode_sv (enc, TOPs);
828 SP -= count;
792 } 829 }
793 830
794 encode_ch (enc, ']'); 831 encode_ch (enc, ']');
795 832
796 FREETMPS; LEAVE; 833 FREETMPS; LEAVE;
… …
1165 1202
1166 if (sv) 1203 if (sv)
1167 { 1204 {
1168 STRLEN cur = SvCUR (sv); 1205 STRLEN cur = SvCUR (sv);
1169 1206
1170 if (SvLEN (sv) <= cur + len) 1207 if (SvLEN (sv) - cur <= len)
1171 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 1208 json_sv_grow (sv, cur, len);
1172 1209
1173 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1210 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
1174 SvCUR_set (sv, SvCUR (sv) + len); 1211 SvCUR_set (sv, SvCUR (sv) + len);
1175 } 1212 }
1176 else 1213 else
… …
1467 1504
1468 DEC_DEC_DEPTH; 1505 DEC_DEC_DEPTH;
1469 sv = newRV_noinc ((SV *)hv); 1506 sv = newRV_noinc ((SV *)hv);
1470 1507
1471 // check filter callbacks 1508 // check filter callbacks
1472 if (dec->json.flags & F_HOOK) 1509 if (expect_false (dec->json.flags & F_HOOK))
1473 { 1510 {
1474 if (dec->json.cb_sk_object && HvKEYS (hv) == 1) 1511 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1475 { 1512 {
1476 HE *cb, *he; 1513 HE *cb, *he;
1477 1514
… …
1487 { 1524 {
1488 dSP; 1525 dSP;
1489 int count; 1526 int count;
1490 1527
1491 ENTER; SAVETMPS; 1528 ENTER; SAVETMPS;
1492 SAVESTACK_POS ();
1493 PUSHMARK (SP); 1529 PUSHMARK (SP);
1494 XPUSHs (HeVAL (he)); 1530 XPUSHs (HeVAL (he));
1495 sv_2mortal (sv); 1531 sv_2mortal (sv);
1496 1532
1497 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1533 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
… …
1500 { 1536 {
1501 sv = newSVsv (POPs); 1537 sv = newSVsv (POPs);
1502 FREETMPS; LEAVE; 1538 FREETMPS; LEAVE;
1503 return sv; 1539 return sv;
1504 } 1540 }
1541 else if (count)
1542 croak ("filter_json_single_key_object callbacks must not return more than one scalar");
1505 1543
1506 SvREFCNT_inc (sv); 1544 SvREFCNT_inc (sv);
1507 FREETMPS; LEAVE; 1545 FREETMPS; LEAVE;
1508 } 1546 }
1509 } 1547 }
… …
1512 { 1550 {
1513 dSP; 1551 dSP;
1514 int count; 1552 int count;
1515 1553
1516 ENTER; SAVETMPS; 1554 ENTER; SAVETMPS;
1517 SAVESTACK_POS ();
1518 PUSHMARK (SP); 1555 PUSHMARK (SP);
1519 XPUSHs (sv_2mortal (sv)); 1556 XPUSHs (sv_2mortal (sv));
1520 1557
1521 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1558 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1522 1559
1523 if (count == 1) 1560 if (count == 1)
1524 {
1525 sv = newSVsv (POPs); 1561 sv = newSVsv (POPs);
1526 FREETMPS; LEAVE; 1562 else if (count == 0)
1527 return sv;
1528 }
1529
1530 SvREFCNT_inc (sv); 1563 SvREFCNT_inc (sv);
1564 else
1565 croak ("filter_json_object callbacks must not return more than one scalar");
1566
1531 FREETMPS; LEAVE; 1567 FREETMPS; LEAVE;
1532 } 1568 }
1533 } 1569 }
1534 1570
1535 return sv; 1571 return sv;
… …
1638 case '-': 1674 case '-':
1639 case '0': case '1': case '2': case '3': case '4': 1675 case '0': case '1': case '2': case '3': case '4':
1640 case '5': case '6': case '7': case '8': case '9': 1676 case '5': case '6': case '7': case '8': case '9':
1641 return decode_num (dec); 1677 return decode_num (dec);
1642 1678
1679 case 'f':
1680 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1681 {
1682 dec->cur += 5;
1683
1684 if (expect_false (!dec->json.v_false))
1685 dec->json.v_false = GET_BOOL (false);
1686
1687 return newSVsv (dec->json.v_false);
1688 }
1689 else
1690 ERR ("'false' expected");
1691
1692 break;
1693
1643 case 't': 1694 case 't':
1644 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1695 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1645 { 1696 {
1646 dec->cur += 4; 1697 dec->cur += 4;
1647#if JSON_SLOW 1698
1648 bool_true = get_bool ("Types::Serialiser::true"); 1699 if (expect_false (!dec->json.v_true))
1649#endif 1700 dec->json.v_true = GET_BOOL (true);
1701
1650 return newSVsv (bool_true); 1702 return newSVsv (dec->json.v_true);
1651 } 1703 }
1652 else 1704 else
1653 ERR ("'true' expected"); 1705 ERR ("'true' expected");
1654
1655 break;
1656
1657 case 'f':
1658 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1659 {
1660 dec->cur += 5;
1661#if JSON_SLOW
1662 bool_false = get_bool ("Types::Serialiser::false");
1663#endif
1664 return newSVsv (bool_false);
1665 }
1666 else
1667 ERR ("'false' expected");
1668 1706
1669 break; 1707 break;
1670 1708
1671 case 'n': 1709 case 'n':
1672 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1710 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
… …
1687fail: 1725fail:
1688 return 0; 1726 return 0;
1689} 1727}
1690 1728
1691static SV * 1729static SV *
1692decode_json (SV *string, JSON *json, char **offset_return) 1730decode_json (SV *string, JSON *json, STRLEN *offset_return)
1693{ 1731{
1694 dec_t dec; 1732 dec_t dec;
1695 SV *sv; 1733 SV *sv;
1696 1734
1697 /* work around bugs in 5.10 where manipulating magic values 1735 /* work around bugs in 5.10 where manipulating magic values
… …
1750 1788
1751 decode_ws (&dec); 1789 decode_ws (&dec);
1752 sv = decode_sv (&dec); 1790 sv = decode_sv (&dec);
1753 1791
1754 if (offset_return) 1792 if (offset_return)
1755 *offset_return = dec.cur; 1793 *offset_return = dec.cur - SvPVX (string);
1756 1794 else if (sv)
1757 if (!(offset_return || !sv))
1758 { 1795 {
1759 // check for trailing garbage 1796 // check for trailing garbage
1760 decode_ws (&dec); 1797 decode_ws (&dec);
1761 1798
1762 if (*dec.cur) 1799 if (dec.cur != dec.end)
1763 { 1800 {
1764 dec.err = "garbage after JSON object"; 1801 dec.err = "garbage after JSON object";
1765 SvREFCNT_dec (sv); 1802 SvREFCNT_dec (sv);
1766 sv = 0; 1803 sv = 0;
1767 } 1804 }
… …
1805 // the state machine here is a bit convoluted and could be simplified a lot 1842 // the state machine here is a bit convoluted and could be simplified a lot
1806 // but this would make it slower, so... 1843 // but this would make it slower, so...
1807 1844
1808 for (;;) 1845 for (;;)
1809 { 1846 {
1810 //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D
1811 switch (self->incr_mode) 1847 switch (self->incr_mode)
1812 { 1848 {
1849 // reached end of a scalar, see if we are inside a nested structure or not
1850 end_of_scalar:
1851 self->incr_mode = INCR_M_JSON;
1852
1853 if (self->incr_nest) // end of a scalar inside array, object or tag
1854 goto incr_m_json;
1855 else // end of scalar outside structure, json text ends here
1856 goto interrupt;
1857
1813 // only used for initial whitespace skipping 1858 // only used for initial whitespace skipping
1814 case INCR_M_WS: 1859 case INCR_M_WS:
1815 for (;;) 1860 for (;;)
1816 { 1861 {
1817 if (*p > 0x20) 1862 if (*p > 0x20)
… …
1859 ++p; 1904 ++p;
1860 } 1905 }
1861 1906
1862 break; 1907 break;
1863 1908
1909 // inside true/false/null
1910 case INCR_M_TFN:
1911 incr_m_tfn:
1912 for (;;)
1913 switch (*p++)
1914 {
1915 case 'r': case 'u': case 'e': // tRUE, falsE, nUll
1916 case 'a': case 'l': case 's': // fALSe, nuLL
1917 // allowed
1918 break;
1919
1920 default:
1921 --p;
1922 goto end_of_scalar;
1923 }
1924
1925 // inside a number
1926 case INCR_M_NUM:
1927 incr_m_num:
1928 for (;;)
1929 switch (*p++)
1930 {
1931 case 'e': case 'E': case '.': case '+':
1932 case '-':
1933 case '0': case '1': case '2': case '3': case '4':
1934 case '5': case '6': case '7': case '8': case '9':
1935 // allowed
1936 break;
1937
1938 default:
1939 --p;
1940 goto end_of_scalar;
1941 }
1942
1864 // inside a string 1943 // inside a string
1865 case INCR_M_STR: 1944 case INCR_M_STR:
1866 incr_m_str: 1945 incr_m_str:
1867 for (;;) 1946 for (;;)
1868 { 1947 {
1869 if (*p == '"') 1948 if (*p == '"')
1870 { 1949 {
1871 ++p; 1950 ++p;
1872 self->incr_mode = INCR_M_JSON;
1873
1874 if (!self->incr_nest)
1875 goto interrupt;
1876
1877 goto incr_m_json; 1951 goto end_of_scalar;
1878 } 1952 }
1879 else if (*p == '\\') 1953 else if (*p == '\\')
1880 { 1954 {
1881 ++p; // "virtually" consumes character after \ 1955 ++p; // "virtually" consumes character after \
1882 1956
… …
1912 --p; // do not eat the whitespace, let the next round do it 1986 --p; // do not eat the whitespace, let the next round do it
1913 goto interrupt; 1987 goto interrupt;
1914 } 1988 }
1915 break; 1989 break;
1916 1990
1991 // the following three blocks handle scalars. this makes the parser
1992 // more strict than required inside arrays or objects, and could
1993 // be moved to a special case on the toplevel (except strings)
1994 case 't':
1995 case 'f':
1996 case 'n':
1997 self->incr_mode = INCR_M_TFN;
1998 goto incr_m_tfn;
1999
2000 case '-':
2001 case '0': case '1': case '2': case '3': case '4':
2002 case '5': case '6': case '7': case '8': case '9':
2003 self->incr_mode = INCR_M_NUM;
2004 goto incr_m_num;
2005
1917 case '"': 2006 case '"':
1918 self->incr_mode = INCR_M_STR; 2007 self->incr_mode = INCR_M_STR;
1919 goto incr_m_str; 2008 goto incr_m_str;
1920 2009
1921 case '[': 2010 case '[':
… …
1968 : i >= 'A' && i <= 'F' ? i - 'A' + 10 2057 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1969 : -1; 2058 : -1;
1970 2059
1971 json_stash = gv_stashpv ("JSON::XS" , 1); 2060 json_stash = gv_stashpv ("JSON::XS" , 1);
1972 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); 2061 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
2062 bool_false = get_bool ("Types::Serialiser::false");
1973 bool_true = get_bool ("Types::Serialiser::true"); 2063 bool_true = get_bool ("Types::Serialiser::true");
1974 bool_false = get_bool ("Types::Serialiser::false");
1975 2064
1976 sv_json = newSVpv ("JSON", 0); 2065 sv_json = newSVpv ("JSON", 0);
1977 SvREADONLY_on (sv_json); 2066 SvREADONLY_on (sv_json);
1978 2067
1979 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ 2068 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
… …
1981 2070
1982PROTOTYPES: DISABLE 2071PROTOTYPES: DISABLE
1983 2072
1984void CLONE (...) 2073void CLONE (...)
1985 CODE: 2074 CODE:
2075 // as long as these writes are atomic, the race should not matter
2076 // as existing threads either already use 0, or use the old value,
2077 // which is sitll correct for the initial thread.
1986 json_stash = 0; 2078 json_stash = 0;
1987 bool_stash = 0; 2079 bool_stash = 0;
2080 bool_false = 0;
2081 bool_true = 0;
1988 2082
1989void new (char *klass) 2083void new (char *klass)
1990 PPCODE: 2084 PPCODE:
1991{ 2085{
1992 SV *pv = NEWSV (0, sizeof (JSON)); 2086 SV *pv = NEWSV (0, sizeof (JSON));
… …
1995 XPUSHs (sv_2mortal (sv_bless ( 2089 XPUSHs (sv_2mortal (sv_bless (
1996 newRV_noinc (pv), 2090 newRV_noinc (pv),
1997 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 2091 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1998 ))); 2092 )));
1999} 2093}
2094
2095void boolean_values (JSON *self, SV *v_false = 0, SV *v_true = 0)
2096 PPCODE:
2097 self->v_false = newSVsv (v_false);
2098 self->v_true = newSVsv (v_true);
2099 XPUSHs (ST (0));
2100
2101void get_boolean_values (JSON *self)
2102 PPCODE:
2103 if (self->v_false && self->v_true)
2104 {
2105 EXTEND (SP, 2);
2106 PUSHs (self->v_false);
2107 PUSHs (self->v_true);
2108 }
2000 2109
2001void ascii (JSON *self, int enable = 1) 2110void ascii (JSON *self, int enable = 1)
2002 ALIAS: 2111 ALIAS:
2003 ascii = F_ASCII 2112 ascii = F_ASCII
2004 latin1 = F_LATIN1 2113 latin1 = F_LATIN1
… …
2097 XPUSHs (ST (0)); 2206 XPUSHs (ST (0));
2098} 2207}
2099 2208
2100void encode (JSON *self, SV *scalar) 2209void encode (JSON *self, SV *scalar)
2101 PPCODE: 2210 PPCODE:
2102 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN; 2211 PUTBACK; XPUSHs (encode_json (scalar, self));
2103 XPUSHs (scalar);
2104 2212
2105void decode (JSON *self, SV *jsonstr) 2213void decode (JSON *self, SV *jsonstr)
2106 PPCODE: 2214 PPCODE:
2107 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN; 2215 PUTBACK; XPUSHs (decode_json (jsonstr, self, 0));
2108 XPUSHs (jsonstr);
2109 2216
2110void decode_prefix (JSON *self, SV *jsonstr) 2217void decode_prefix (JSON *self, SV *jsonstr)
2111 PPCODE: 2218 PPCODE:
2112{ 2219{
2113 SV *sv; 2220 SV *sv;
2114 char *offset; 2221 STRLEN offset;
2115 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; 2222 PUTBACK; sv = decode_json (jsonstr, self, &offset);
2116 EXTEND (SP, 2); 2223 EXTEND (SP, 2);
2117 PUSHs (sv); 2224 PUSHs (sv);
2118 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); 2225 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
2119} 2226}
2120 2227
2121void incr_parse (JSON *self, SV *jsonstr = 0) 2228void incr_parse (JSON *self, SV *jsonstr = 0)
2122 PPCODE: 2229 PPCODE:
2123{ 2230{
… …
2157 { 2264 {
2158 STRLEN len; 2265 STRLEN len;
2159 const char *str = SvPV (jsonstr, len); 2266 const char *str = SvPV (jsonstr, len);
2160 STRLEN cur = SvCUR (self->incr_text); 2267 STRLEN cur = SvCUR (self->incr_text);
2161 2268
2162 if (SvLEN (self->incr_text) <= cur + len) 2269 if (SvLEN (self->incr_text) - cur <= len)
2163 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 2270 json_sv_grow (self->incr_text, cur, len);
2164 2271
2165 Move (str, SvEND (self->incr_text), len, char); 2272 Move (str, SvEND (self->incr_text), len, char);
2166 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 2273 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2167 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there 2274 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2168 } 2275 }
… …
2170 2277
2171 if (GIMME_V != G_VOID) 2278 if (GIMME_V != G_VOID)
2172 do 2279 do
2173 { 2280 {
2174 SV *sv; 2281 SV *sv;
2175 char *offset; 2282 STRLEN offset;
2176 2283
2177 if (!INCR_DONE (self)) 2284 if (!INCR_DONE (self))
2178 { 2285 {
2179 incr_parse (self); 2286 incr_parse (self);
2180 2287
… …
2193 2300
2194 break; 2301 break;
2195 } 2302 }
2196 } 2303 }
2197 2304
2198 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; 2305 PUTBACK; sv = decode_json (self->incr_text, self, &offset);
2199 XPUSHs (sv); 2306 XPUSHs (sv);
2200 2307
2201 self->incr_pos -= offset - SvPVX (self->incr_text); 2308 self->incr_pos -= offset;
2202 self->incr_nest = 0; 2309 self->incr_nest = 0;
2203 self->incr_mode = 0; 2310 self->incr_mode = 0;
2204 2311
2205 sv_chop (self->incr_text, offset); 2312 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
2206 } 2313 }
2207 while (GIMME_V == G_ARRAY); 2314 while (GIMME_V == G_ARRAY);
2208} 2315}
2209 2316
2210SV *incr_text (JSON *self) 2317SV *incr_text (JSON *self)
… …
2241 self->incr_mode = 0; 2348 self->incr_mode = 0;
2242} 2349}
2243 2350
2244void DESTROY (JSON *self) 2351void DESTROY (JSON *self)
2245 CODE: 2352 CODE:
2353 SvREFCNT_dec (self->v_false);
2354 SvREFCNT_dec (self->v_true);
2246 SvREFCNT_dec (self->cb_sk_object); 2355 SvREFCNT_dec (self->cb_sk_object);
2247 SvREFCNT_dec (self->cb_object); 2356 SvREFCNT_dec (self->cb_object);
2248 SvREFCNT_dec (self->incr_text); 2357 SvREFCNT_dec (self->incr_text);
2249 2358
2250PROTOTYPES: ENABLE 2359PROTOTYPES: ENABLE
… …
2253 PPCODE: 2362 PPCODE:
2254{ 2363{
2255 JSON json; 2364 JSON json;
2256 json_init (&json); 2365 json_init (&json);
2257 json.flags |= F_UTF8; 2366 json.flags |= F_UTF8;
2258 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN; 2367 PUTBACK; XPUSHs (encode_json (scalar, &json));
2259 XPUSHs (scalar);
2260} 2368}
2261 2369
2262void decode_json (SV *jsonstr) 2370void decode_json (SV *jsonstr)
2263 PPCODE: 2371 PPCODE:
2264{ 2372{
2265 JSON json; 2373 JSON json;
2266 json_init (&json); 2374 json_init (&json);
2267 json.flags |= F_UTF8; 2375 json.flags |= F_UTF8;
2268 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN; 2376 PUTBACK; XPUSHs (decode_json (jsonstr, &json, 0));
2269 XPUSHs (jsonstr);
2270} 2377}
2271 2378

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines