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 1.126 by root, Sun Feb 21 16:18:19 2016 UTC vs.
Revision 1.136 by root, Thu Nov 15 22:35:35 2018 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
… …
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
… …
1524 { 1561 {
1525 sv = newSVsv (POPs); 1562 sv = newSVsv (POPs);
1526 FREETMPS; LEAVE; 1563 FREETMPS; LEAVE;
1527 return sv; 1564 return sv;
1528 } 1565 }
1566 else if (count)
1567 croak ("filter_json_object callbacks must not return more than one scalar");
1529 1568
1530 SvREFCNT_inc (sv); 1569 SvREFCNT_inc (sv);
1531 FREETMPS; LEAVE; 1570 FREETMPS; LEAVE;
1532 } 1571 }
1533 } 1572 }
… …
1638 case '-': 1677 case '-':
1639 case '0': case '1': case '2': case '3': case '4': 1678 case '0': case '1': case '2': case '3': case '4':
1640 case '5': case '6': case '7': case '8': case '9': 1679 case '5': case '6': case '7': case '8': case '9':
1641 return decode_num (dec); 1680 return decode_num (dec);
1642 1681
1682 case 'f':
1683 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1684 {
1685 dec->cur += 5;
1686
1687 if (expect_false (!dec->json.v_false))
1688 dec->json.v_false = GET_BOOL (false);
1689
1690 return newSVsv (dec->json.v_false);
1691 }
1692 else
1693 ERR ("'false' expected");
1694
1695 break;
1696
1643 case 't': 1697 case 't':
1644 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1698 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1645 { 1699 {
1646 dec->cur += 4; 1700 dec->cur += 4;
1647#if JSON_SLOW 1701
1648 bool_true = get_bool ("Types::Serialiser::true"); 1702 if (expect_false (!dec->json.v_true))
1649#endif 1703 dec->json.v_true = GET_BOOL (true);
1704
1650 return newSVsv (bool_true); 1705 return newSVsv (dec->json.v_true);
1651 } 1706 }
1652 else 1707 else
1653 ERR ("'true' expected"); 1708 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 1709
1669 break; 1710 break;
1670 1711
1671 case 'n': 1712 case 'n':
1672 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1713 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
… …
1687fail: 1728fail:
1688 return 0; 1729 return 0;
1689} 1730}
1690 1731
1691static SV * 1732static SV *
1692decode_json (SV *string, JSON *json, char **offset_return) 1733decode_json (SV *string, JSON *json, STRLEN *offset_return)
1693{ 1734{
1694 dec_t dec; 1735 dec_t dec;
1695 SV *sv; 1736 SV *sv;
1696 1737
1697 /* work around bugs in 5.10 where manipulating magic values 1738 /* work around bugs in 5.10 where manipulating magic values
1698 * makes perl ignore the magic in subsequent accesses. 1739 * makes perl ignore the magic in subsequent accesses.
1699 * also make a copy of non-PV values, to get them into a clean 1740 * also make a copy of non-PV values, to get them into a clean
1700 * state (SvPV should do that, but it's buggy, see below). 1741 * state (SvPV should do that, but it's buggy, see below).
1742 *
1743 * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
1744 * as reported by Reini Urban.
1701 */ 1745 */
1702 /*SvGETMAGIC (string);*/ 1746 /*SvGETMAGIC (string);*/
1703 if (SvMAGICAL (string) || !SvPOK (string)) 1747 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1704 string = sv_2mortal (newSVsv (string)); 1748 string = sv_2mortal (newSVsv (string));
1705 1749
1706 SvUPGRADE (string, SVt_PV); 1750 SvUPGRADE (string, SVt_PV);
1707 1751
1708 /* work around a bug in perl 5.10, which causes SvCUR to fail an 1752 /* work around a bug in perl 5.10, which causes SvCUR to fail an
… …
1747 1791
1748 decode_ws (&dec); 1792 decode_ws (&dec);
1749 sv = decode_sv (&dec); 1793 sv = decode_sv (&dec);
1750 1794
1751 if (offset_return) 1795 if (offset_return)
1752 *offset_return = dec.cur; 1796 *offset_return = dec.cur - SvPVX (string);
1753 1797 else if (sv)
1754 if (!(offset_return || !sv))
1755 { 1798 {
1756 // check for trailing garbage 1799 // check for trailing garbage
1757 decode_ws (&dec); 1800 decode_ws (&dec);
1758 1801
1759 if (*dec.cur) 1802 if (dec.cur != dec.end)
1760 { 1803 {
1761 dec.err = "garbage after JSON object"; 1804 dec.err = "garbage after JSON object";
1762 SvREFCNT_dec (sv); 1805 SvREFCNT_dec (sv);
1763 sv = 0; 1806 sv = 0;
1764 } 1807 }
… …
1802 // the state machine here is a bit convoluted and could be simplified a lot 1845 // the state machine here is a bit convoluted and could be simplified a lot
1803 // but this would make it slower, so... 1846 // but this would make it slower, so...
1804 1847
1805 for (;;) 1848 for (;;)
1806 { 1849 {
1807 //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
1808 switch (self->incr_mode) 1850 switch (self->incr_mode)
1809 { 1851 {
1852 // reached end of a scalar, see if we are inside a nested structure or not
1853 end_of_scalar:
1854 self->incr_mode = INCR_M_JSON;
1855
1856 if (self->incr_nest) // end of a scalar inside array, object or tag
1857 goto incr_m_json;
1858 else // end of scalar outside structure, json text ends here
1859 goto interrupt;
1860
1810 // only used for initial whitespace skipping 1861 // only used for initial whitespace skipping
1811 case INCR_M_WS: 1862 case INCR_M_WS:
1812 for (;;) 1863 for (;;)
1813 { 1864 {
1814 if (*p > 0x20) 1865 if (*p > 0x20)
… …
1856 ++p; 1907 ++p;
1857 } 1908 }
1858 1909
1859 break; 1910 break;
1860 1911
1912 // inside true/false/null
1913 case INCR_M_TFN:
1914 incr_m_tfn:
1915 for (;;)
1916 switch (*p++)
1917 {
1918 case 'r': case 'u': case 'e': // tRUE, falsE, nUll
1919 case 'a': case 'l': case 's': // fALSe, nuLL
1920 // allowed
1921 break;
1922
1923 default:
1924 --p;
1925 goto end_of_scalar;
1926 }
1927
1928 // inside a number
1929 case INCR_M_NUM:
1930 incr_m_num:
1931 for (;;)
1932 switch (*p++)
1933 {
1934 case 'e': case 'E': case '.': case '+':
1935 case '-':
1936 case '0': case '1': case '2': case '3': case '4':
1937 case '5': case '6': case '7': case '8': case '9':
1938 // allowed
1939 break;
1940
1941 default:
1942 --p;
1943 goto end_of_scalar;
1944 }
1945
1861 // inside a string 1946 // inside a string
1862 case INCR_M_STR: 1947 case INCR_M_STR:
1863 incr_m_str: 1948 incr_m_str:
1864 for (;;) 1949 for (;;)
1865 { 1950 {
1866 if (*p == '"') 1951 if (*p == '"')
1867 { 1952 {
1868 ++p; 1953 ++p;
1869 self->incr_mode = INCR_M_JSON;
1870
1871 if (!self->incr_nest)
1872 goto interrupt;
1873
1874 goto incr_m_json; 1954 goto end_of_scalar;
1875 } 1955 }
1876 else if (*p == '\\') 1956 else if (*p == '\\')
1877 { 1957 {
1878 ++p; // "virtually" consumes character after \ 1958 ++p; // "virtually" consumes character after \
1879 1959
… …
1909 --p; // do not eat the whitespace, let the next round do it 1989 --p; // do not eat the whitespace, let the next round do it
1910 goto interrupt; 1990 goto interrupt;
1911 } 1991 }
1912 break; 1992 break;
1913 1993
1994 // the following three blocks handle scalars. this makes the parser
1995 // more strict than required inside arrays or objects, and could
1996 // be moved to a special case on the toplevel (except strings)
1997 case 't':
1998 case 'f':
1999 case 'n':
2000 self->incr_mode = INCR_M_TFN;
2001 goto incr_m_tfn;
2002
2003 case '-':
2004 case '0': case '1': case '2': case '3': case '4':
2005 case '5': case '6': case '7': case '8': case '9':
2006 self->incr_mode = INCR_M_NUM;
2007 goto incr_m_num;
2008
1914 case '"': 2009 case '"':
1915 self->incr_mode = INCR_M_STR; 2010 self->incr_mode = INCR_M_STR;
1916 goto incr_m_str; 2011 goto incr_m_str;
1917 2012
1918 case '[': 2013 case '[':
… …
1965 : i >= 'A' && i <= 'F' ? i - 'A' + 10 2060 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1966 : -1; 2061 : -1;
1967 2062
1968 json_stash = gv_stashpv ("JSON::XS" , 1); 2063 json_stash = gv_stashpv ("JSON::XS" , 1);
1969 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); 2064 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
2065 bool_false = get_bool ("Types::Serialiser::false");
1970 bool_true = get_bool ("Types::Serialiser::true"); 2066 bool_true = get_bool ("Types::Serialiser::true");
1971 bool_false = get_bool ("Types::Serialiser::false");
1972 2067
1973 sv_json = newSVpv ("JSON", 0); 2068 sv_json = newSVpv ("JSON", 0);
1974 SvREADONLY_on (sv_json); 2069 SvREADONLY_on (sv_json);
1975 2070
1976 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ 2071 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
… …
1978 2073
1979PROTOTYPES: DISABLE 2074PROTOTYPES: DISABLE
1980 2075
1981void CLONE (...) 2076void CLONE (...)
1982 CODE: 2077 CODE:
2078 // as long as these writes are atomic, the race should not matter
2079 // as existing threads either already use 0, or use the old value,
2080 // which is sitll correct for the initial thread.
1983 json_stash = 0; 2081 json_stash = 0;
1984 bool_stash = 0; 2082 bool_stash = 0;
2083 bool_false = 0;
2084 bool_true = 0;
1985 2085
1986void new (char *klass) 2086void new (char *klass)
1987 PPCODE: 2087 PPCODE:
1988{ 2088{
1989 SV *pv = NEWSV (0, sizeof (JSON)); 2089 SV *pv = NEWSV (0, sizeof (JSON));
… …
1992 XPUSHs (sv_2mortal (sv_bless ( 2092 XPUSHs (sv_2mortal (sv_bless (
1993 newRV_noinc (pv), 2093 newRV_noinc (pv),
1994 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 2094 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1995 ))); 2095 )));
1996} 2096}
2097
2098void boolean_values (JSON *self, SV *v_false = 0, SV *v_true = 0)
2099 PPCODE:
2100 self->v_false = newSVsv (v_false);
2101 self->v_true = newSVsv (v_true);
2102 XPUSHs (ST (0));
2103
2104void get_boolean_values (JSON *self)
2105 PPCODE:
2106 if (self->v_false && self->v_true)
2107 {
2108 EXTEND (SP, 2);
2109 PUSHs (self->v_false);
2110 PUSHs (self->v_true);
2111 }
1997 2112
1998void ascii (JSON *self, int enable = 1) 2113void ascii (JSON *self, int enable = 1)
1999 ALIAS: 2114 ALIAS:
2000 ascii = F_ASCII 2115 ascii = F_ASCII
2001 latin1 = F_LATIN1 2116 latin1 = F_LATIN1
… …
2106 2221
2107void decode_prefix (JSON *self, SV *jsonstr) 2222void decode_prefix (JSON *self, SV *jsonstr)
2108 PPCODE: 2223 PPCODE:
2109{ 2224{
2110 SV *sv; 2225 SV *sv;
2111 char *offset; 2226 STRLEN offset;
2112 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; 2227 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
2113 EXTEND (SP, 2); 2228 EXTEND (SP, 2);
2114 PUSHs (sv); 2229 PUSHs (sv);
2115 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); 2230 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
2116} 2231}
2117 2232
2118void incr_parse (JSON *self, SV *jsonstr = 0) 2233void incr_parse (JSON *self, SV *jsonstr = 0)
2119 PPCODE: 2234 PPCODE:
2120{ 2235{
… …
2154 { 2269 {
2155 STRLEN len; 2270 STRLEN len;
2156 const char *str = SvPV (jsonstr, len); 2271 const char *str = SvPV (jsonstr, len);
2157 STRLEN cur = SvCUR (self->incr_text); 2272 STRLEN cur = SvCUR (self->incr_text);
2158 2273
2159 if (SvLEN (self->incr_text) <= cur + len) 2274 if (SvLEN (self->incr_text) - cur <= len)
2160 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 2275 json_sv_grow (self->incr_text, cur, len);
2161 2276
2162 Move (str, SvEND (self->incr_text), len, char); 2277 Move (str, SvEND (self->incr_text), len, char);
2163 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 2278 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2164 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there 2279 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2165 } 2280 }
… …
2167 2282
2168 if (GIMME_V != G_VOID) 2283 if (GIMME_V != G_VOID)
2169 do 2284 do
2170 { 2285 {
2171 SV *sv; 2286 SV *sv;
2172 char *offset; 2287 STRLEN offset;
2173 2288
2174 if (!INCR_DONE (self)) 2289 if (!INCR_DONE (self))
2175 { 2290 {
2176 incr_parse (self); 2291 incr_parse (self);
2177 2292
… …
2193 } 2308 }
2194 2309
2195 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; 2310 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2196 XPUSHs (sv); 2311 XPUSHs (sv);
2197 2312
2198 self->incr_pos -= offset - SvPVX (self->incr_text); 2313 self->incr_pos -= offset;
2199 self->incr_nest = 0; 2314 self->incr_nest = 0;
2200 self->incr_mode = 0; 2315 self->incr_mode = 0;
2201 2316
2202 sv_chop (self->incr_text, offset); 2317 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
2203 } 2318 }
2204 while (GIMME_V == G_ARRAY); 2319 while (GIMME_V == G_ARRAY);
2205} 2320}
2206 2321
2207SV *incr_text (JSON *self) 2322SV *incr_text (JSON *self)
… …
2238 self->incr_mode = 0; 2353 self->incr_mode = 0;
2239} 2354}
2240 2355
2241void DESTROY (JSON *self) 2356void DESTROY (JSON *self)
2242 CODE: 2357 CODE:
2358 SvREFCNT_dec (self->v_false);
2359 SvREFCNT_dec (self->v_true);
2243 SvREFCNT_dec (self->cb_sk_object); 2360 SvREFCNT_dec (self->cb_sk_object);
2244 SvREFCNT_dec (self->cb_object); 2361 SvREFCNT_dec (self->cb_object);
2245 SvREFCNT_dec (self->incr_text); 2362 SvREFCNT_dec (self->incr_text);
2246 2363
2247PROTOTYPES: ENABLE 2364PROTOTYPES: ENABLE

Diff Legend

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