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 1.125 by root, Sun Feb 21 15:37:53 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 (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))
84#else 86#else
85# define JSON_SLOW 0
86# define JSON_STASH json_stash 87# define JSON_STASH json_stash
88# define BOOL_STASH bool_stash
89# define GET_BOOL(value) bool_ ## value
87#endif 90#endif
88 91
89// 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
90#define STACK_HES 64 93#define STACK_HES 64
91 94
92static HV *json_stash, *types_boolean_stash; // JSON::XS:: 95static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean::
93static SV *types_true, *types_false, *sv_json; 96static SV *bool_false, *bool_true;
97static SV *sv_json;
94 98
95enum { 99enum {
96 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
97 INCR_M_STR, // inside string 103 INCR_M_STR, // inside string
98 INCR_M_BS, // inside backslash 104 INCR_M_BS, // inside backslash
99 INCR_M_C0, // inside comment in initial whitespace sequence 105 INCR_M_C0, // inside comment in initial whitespace sequence
100 INCR_M_C1, // inside comment in other places 106 INCR_M_C1, // inside comment in other places
101 INCR_M_JSON // outside anything, count nesting 107 INCR_M_JSON // outside anything, count nesting
… …
114 // for the incremental parser 120 // for the incremental parser
115 SV *incr_text; // the source text so far 121 SV *incr_text; // the source text so far
116 STRLEN incr_pos; // the current offset into the text 122 STRLEN incr_pos; // the current offset into the text
117 int incr_nest; // {[]}-nesting level 123 int incr_nest; // {[]}-nesting level
118 unsigned char incr_mode; 124 unsigned char incr_mode;
125
126 SV *v_false, *v_true;
119} JSON; 127} JSON;
120 128
121INLINE void 129INLINE void
122json_init (JSON *json) 130json_init (JSON *json)
123{ 131{
124 Zero (json, 1, JSON); 132 static const JSON init = { F_ALLOW_NONREF, 512 };
125 json->max_depth = 512; 133
134 *json = init;
126} 135}
127 136
128///////////////////////////////////////////////////////////////////////////// 137/////////////////////////////////////////////////////////////////////////////
129// utility functions 138// utility functions
130 139
… …
152 SvPV_renew (sv, SvCUR (sv) + 1); 161 SvPV_renew (sv, SvCUR (sv) + 1);
153#endif 162#endif
154 } 163 }
155} 164}
156 165
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
157// 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
158// case of an error. 192// case of an error.
159// we special-case "safe" characters from U+80 .. U+7FF, 193// we special-case "safe" characters from U+80 .. U+7FF,
160// but use the very good perl function to parse anything else. 194// but use the very good perl function to parse anything else.
161// note that we never call this function for a ascii codepoints 195// note that we never call this function for a ascii codepoints
162INLINE UV 196INLINE UV
… …
299 // a recursion depth of ten gives us >>500 bits 333 // a recursion depth of ten gives us >>500 bits
300 json_atof_scan1 (s, &accum, &expo, 0, 10); 334 json_atof_scan1 (s, &accum, &expo, 0, 10);
301 335
302 return neg ? -accum : accum; 336 return neg ? -accum : accum;
303} 337}
338
339// target of scalar reference is bool? -1 == nope, 0 == false, 1 == true
340static int
341ref_bool_type (SV *sv)
342{
343 svtype svt = SvTYPE (sv);
344
345 if (svt < SVt_PVAV)
346 {
347 STRLEN len = 0;
348 char *pv = svt ? SvPV (sv, len) : 0;
349
350 if (len == 1)
351 if (*pv == '1')
352 return 1;
353 else if (*pv == '0')
354 return 0;
355 }
356
357 return -1;
358}
359
360// returns whether scalar is not a reference in the sense of allow_nonref
361static int
362json_nonref (SV *scalar)
363{
364 if (!SvROK (scalar))
365 return 1;
366
367 scalar = SvRV (scalar);
368
369 if (SvTYPE (scalar) >= SVt_PVMG)
370 {
371 if (SvSTASH (scalar) == bool_stash)
372 return 1;
373
374 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
375 return 1;
376 }
377
378 return 0;
379}
380
304///////////////////////////////////////////////////////////////////////////// 381/////////////////////////////////////////////////////////////////////////////
305// encoder 382// encoder
306 383
307// structure used for encoding JSON 384// structure used for encoding JSON
308typedef struct 385typedef struct
… …
316} enc_t; 393} enc_t;
317 394
318INLINE void 395INLINE void
319need (enc_t *enc, STRLEN len) 396need (enc_t *enc, STRLEN len)
320{ 397{
321 if (expect_false (enc->cur + len >= enc->end)) 398 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
322 { 399 {
323 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 400 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
324 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 401 char *buf = json_sv_grow (enc->sv, cur, len);
325 enc->cur = SvPVX (enc->sv) + cur; 402 enc->cur = buf + cur;
326 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 403 enc->end = buf + SvLEN (enc->sv) - 1;
327 } 404 }
328} 405}
329 406
330INLINE void 407INLINE void
331encode_ch (enc_t *enc, char ch) 408encode_ch (enc_t *enc, char ch)
… …
347 424
348 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case 425 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
349 { 426 {
350 if (expect_false (ch == '"')) // but with slow exceptions 427 if (expect_false (ch == '"')) // but with slow exceptions
351 { 428 {
352 need (enc, len += 1); 429 need (enc, len + 1);
353 *enc->cur++ = '\\'; 430 *enc->cur++ = '\\';
354 *enc->cur++ = '"'; 431 *enc->cur++ = '"';
355 } 432 }
356 else if (expect_false (ch == '\\')) 433 else if (expect_false (ch == '\\'))
357 { 434 {
358 need (enc, len += 1); 435 need (enc, len + 1);
359 *enc->cur++ = '\\'; 436 *enc->cur++ = '\\';
360 *enc->cur++ = '\\'; 437 *enc->cur++ = '\\';
361 } 438 }
362 else 439 else
363 *enc->cur++ = ch; 440 *enc->cur++ = ch;
… …
366 } 443 }
367 else 444 else
368 { 445 {
369 switch (ch) 446 switch (ch)
370 { 447 {
371 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;
372 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;
373 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;
374 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;
375 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;
376 453
377 default: 454 default:
378 { 455 {
379 STRLEN clen; 456 STRLEN clen;
380 UV uch; 457 UV uch;
… …
396 if (uch >= 0x10000UL) 473 if (uch >= 0x10000UL)
397 { 474 {
398 if (uch >= 0x110000UL) 475 if (uch >= 0x110000UL)
399 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);
400 477
401 need (enc, len += 11); 478 need (enc, len + 11);
402 sprintf (enc->cur, "\\u%04x\\u%04x", 479 sprintf (enc->cur, "\\u%04x\\u%04x",
403 (int)((uch - 0x10000) / 0x400 + 0xD800), 480 (int)((uch - 0x10000) / 0x400 + 0xD800),
404 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 481 (int)((uch - 0x10000) % 0x400 + 0xDC00));
405 enc->cur += 12; 482 enc->cur += 12;
406 } 483 }
407 else 484 else
408 { 485 {
409 need (enc, len += 5); 486 need (enc, len + 5);
410 *enc->cur++ = '\\'; 487 *enc->cur++ = '\\';
411 *enc->cur++ = 'u'; 488 *enc->cur++ = 'u';
412 *enc->cur++ = PL_hexdigit [ uch >> 12 ]; 489 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
413 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; 490 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
414 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; 491 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
… …
422 *enc->cur++ = uch; 499 *enc->cur++ = uch;
423 str += clen; 500 str += clen;
424 } 501 }
425 else if (is_utf8) 502 else if (is_utf8)
426 { 503 {
427 need (enc, len += clen); 504 need (enc, len + clen);
428 do 505 do
429 { 506 {
430 *enc->cur++ = *str++; 507 *enc->cur++ = *str++;
431 } 508 }
432 while (--clen); 509 while (--clen);
433 } 510 }
434 else 511 else
435 { 512 {
436 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
437 enc->cur = encode_utf8 (enc->cur, uch); 514 enc->cur = encode_utf8 (enc->cur, uch);
438 ++str; 515 ++str;
439 } 516 }
440 } 517 }
441 } 518 }
… …
701 SvGETMAGIC (sv); 778 SvGETMAGIC (sv);
702 svt = SvTYPE (sv); 779 svt = SvTYPE (sv);
703 780
704 if (expect_false (SvOBJECT (sv))) 781 if (expect_false (SvOBJECT (sv)))
705 { 782 {
706 HV *boolean_stash = !JSON_SLOW || types_boolean_stash
707 ? types_boolean_stash
708 : gv_stashpv ("Types::Serialiser::Boolean", 1);
709 HV *stash = SvSTASH (sv); 783 HV *stash = SvSTASH (sv);
710 784
711 if (stash == boolean_stash) 785 if (stash == bool_stash)
712 { 786 {
713 if (SvIV (sv))
714 encode_str (enc, "true", 4, 0); 787 if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
715 else
716 encode_str (enc, "false", 5, 0); 788 else encode_str (enc, "false", 5, 0);
717 } 789 }
718 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)))
719 { 791 {
720 int count; 792 int count;
721 dSP; 793 dSP;
722 794
723 ENTER; SAVETMPS; 795 ENTER; SAVETMPS;
724 SAVESTACK_POS ();
725 PUSHMARK (SP); 796 PUSHMARK (SP);
726 EXTEND (SP, 2); 797 EXTEND (SP, 2);
727 // 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
728 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 799 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
729 PUSHs (sv_json); 800 PUSHs (sv_json);
… …
741 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash)); 812 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
742 encode_ch (enc, '"'); 813 encode_ch (enc, '"');
743 encode_ch (enc, ')'); 814 encode_ch (enc, ')');
744 encode_ch (enc, '['); 815 encode_ch (enc, '[');
745 816
746 while (count) 817 if (count)
747 { 818 {
819 int i;
820
821 for (i = 0; i < count - 1; ++i)
822 {
748 encode_sv (enc, SP[1 - count--]); 823 encode_sv (enc, SP[i + 1 - count]);
749
750 if (count)
751 encode_ch (enc, ','); 824 encode_ch (enc, ',');
825 }
826
827 encode_sv (enc, TOPs);
828 SP -= count;
752 } 829 }
753 830
754 encode_ch (enc, ']'); 831 encode_ch (enc, ']');
755 832
756 FREETMPS; LEAVE; 833 FREETMPS; LEAVE;
… …
790 encode_hv (enc, (HV *)sv); 867 encode_hv (enc, (HV *)sv);
791 else if (svt == SVt_PVAV) 868 else if (svt == SVt_PVAV)
792 encode_av (enc, (AV *)sv); 869 encode_av (enc, (AV *)sv);
793 else if (svt < SVt_PVAV) 870 else if (svt < SVt_PVAV)
794 { 871 {
795 STRLEN len = 0; 872 int bool_type = ref_bool_type (sv);
796 char *pv = svt ? SvPV (sv, len) : 0;
797 873
798 if (len == 1 && *pv == '1') 874 if (bool_type == 1)
799 encode_str (enc, "true", 4, 0); 875 encode_str (enc, "true", 4, 0);
800 else if (len == 1 && *pv == '0') 876 else if (bool_type == 0)
801 encode_str (enc, "false", 5, 0); 877 encode_str (enc, "false", 5, 0);
802 else if (enc->json.flags & F_ALLOW_UNKNOWN) 878 else if (enc->json.flags & F_ALLOW_UNKNOWN)
803 encode_str (enc, "null", 4, 0); 879 encode_str (enc, "null", 4, 0);
804 else 880 else
805 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 881 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
… …
883 else 959 else
884 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data", 960 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
885 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); 961 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
886} 962}
887 963
888static int
889json_scalar (SV *scalar)
890{
891 return 0;//D
892 if (!SvROK (scalar))
893 return 1;
894}
895
896static SV * 964static SV *
897encode_json (SV *scalar, JSON *json) 965encode_json (SV *scalar, JSON *json)
898{ 966{
899 enc_t enc; 967 enc_t enc;
900 968
901 if (!(json->flags & F_ALLOW_NONREF) && json_scalar (scalar)) 969 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
902 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 970 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
903 971
904 enc.json = *json; 972 enc.json = *json;
905 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 973 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
906 enc.cur = SvPVX (enc.sv); 974 enc.cur = SvPVX (enc.sv);
… …
1134 1202
1135 if (sv) 1203 if (sv)
1136 { 1204 {
1137 STRLEN cur = SvCUR (sv); 1205 STRLEN cur = SvCUR (sv);
1138 1206
1139 if (SvLEN (sv) <= cur + len) 1207 if (SvLEN (sv) - cur <= len)
1140 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 1208 json_sv_grow (sv, cur, len);
1141 1209
1142 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1210 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
1143 SvCUR_set (sv, SvCUR (sv) + len); 1211 SvCUR_set (sv, SvCUR (sv) + len);
1144 } 1212 }
1145 else 1213 else
… …
1436 1504
1437 DEC_DEC_DEPTH; 1505 DEC_DEC_DEPTH;
1438 sv = newRV_noinc ((SV *)hv); 1506 sv = newRV_noinc ((SV *)hv);
1439 1507
1440 // check filter callbacks 1508 // check filter callbacks
1441 if (dec->json.flags & F_HOOK) 1509 if (expect_false (dec->json.flags & F_HOOK))
1442 { 1510 {
1443 if (dec->json.cb_sk_object && HvKEYS (hv) == 1) 1511 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1444 { 1512 {
1445 HE *cb, *he; 1513 HE *cb, *he;
1446 1514
… …
1456 { 1524 {
1457 dSP; 1525 dSP;
1458 int count; 1526 int count;
1459 1527
1460 ENTER; SAVETMPS; 1528 ENTER; SAVETMPS;
1461 SAVESTACK_POS ();
1462 PUSHMARK (SP); 1529 PUSHMARK (SP);
1463 XPUSHs (HeVAL (he)); 1530 XPUSHs (HeVAL (he));
1464 sv_2mortal (sv); 1531 sv_2mortal (sv);
1465 1532
1466 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1533 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
… …
1469 { 1536 {
1470 sv = newSVsv (POPs); 1537 sv = newSVsv (POPs);
1471 FREETMPS; LEAVE; 1538 FREETMPS; LEAVE;
1472 return sv; 1539 return sv;
1473 } 1540 }
1541 else if (count)
1542 croak ("filter_json_single_key_object callbacks must not return more than one scalar");
1474 1543
1475 SvREFCNT_inc (sv); 1544 SvREFCNT_inc (sv);
1476 FREETMPS; LEAVE; 1545 FREETMPS; LEAVE;
1477 } 1546 }
1478 } 1547 }
… …
1481 { 1550 {
1482 dSP; 1551 dSP;
1483 int count; 1552 int count;
1484 1553
1485 ENTER; SAVETMPS; 1554 ENTER; SAVETMPS;
1486 SAVESTACK_POS ();
1487 PUSHMARK (SP); 1555 PUSHMARK (SP);
1488 XPUSHs (sv_2mortal (sv)); 1556 XPUSHs (sv_2mortal (sv));
1489 1557
1490 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1558 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1491 1559
1492 if (count == 1) 1560 if (count == 1)
1493 {
1494 sv = newSVsv (POPs); 1561 sv = newSVsv (POPs);
1495 FREETMPS; LEAVE; 1562 else if (count == 0)
1496 return sv;
1497 }
1498
1499 SvREFCNT_inc (sv); 1563 SvREFCNT_inc (sv);
1564 else
1565 croak ("filter_json_object callbacks must not return more than one scalar");
1566
1500 FREETMPS; LEAVE; 1567 FREETMPS; LEAVE;
1501 } 1568 }
1502 } 1569 }
1503 1570
1504 return sv; 1571 return sv;
… …
1607 case '-': 1674 case '-':
1608 case '0': case '1': case '2': case '3': case '4': 1675 case '0': case '1': case '2': case '3': case '4':
1609 case '5': case '6': case '7': case '8': case '9': 1676 case '5': case '6': case '7': case '8': case '9':
1610 return decode_num (dec); 1677 return decode_num (dec);
1611 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
1612 case 't': 1694 case 't':
1613 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1695 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1614 { 1696 {
1615 dec->cur += 4; 1697 dec->cur += 4;
1616#if JSON_SLOW 1698
1617 types_true = get_bool ("Types::Serialiser::true"); 1699 if (expect_false (!dec->json.v_true))
1618#endif 1700 dec->json.v_true = GET_BOOL (true);
1701
1619 return newSVsv (types_true); 1702 return newSVsv (dec->json.v_true);
1620 } 1703 }
1621 else 1704 else
1622 ERR ("'true' expected"); 1705 ERR ("'true' expected");
1623
1624 break;
1625
1626 case 'f':
1627 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1628 {
1629 dec->cur += 5;
1630#if JSON_SLOW
1631 types_false = get_bool ("Types::Serialiser::false");
1632#endif
1633 return newSVsv (types_false);
1634 }
1635 else
1636 ERR ("'false' expected");
1637 1706
1638 break; 1707 break;
1639 1708
1640 case 'n': 1709 case 'n':
1641 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1710 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
… …
1656fail: 1725fail:
1657 return 0; 1726 return 0;
1658} 1727}
1659 1728
1660static SV * 1729static SV *
1661decode_json (SV *string, JSON *json, char **offset_return) 1730decode_json (SV *string, JSON *json, STRLEN *offset_return)
1662{ 1731{
1663 dec_t dec; 1732 dec_t dec;
1664 SV *sv; 1733 SV *sv;
1665 1734
1666 /* work around bugs in 5.10 where manipulating magic values 1735 /* work around bugs in 5.10 where manipulating magic values
1667 * makes perl ignore the magic in subsequent accesses. 1736 * makes perl ignore the magic in subsequent accesses.
1668 * also make a copy of non-PV values, to get them into a clean 1737 * also make a copy of non-PV values, to get them into a clean
1669 * state (SvPV should do that, but it's buggy, see below). 1738 * state (SvPV should do that, but it's buggy, see below).
1739 *
1740 * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
1741 * as reported by Reini Urban.
1670 */ 1742 */
1671 /*SvGETMAGIC (string);*/ 1743 /*SvGETMAGIC (string);*/
1672 if (SvMAGICAL (string) || !SvPOK (string)) 1744 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1673 string = sv_2mortal (newSVsv (string)); 1745 string = sv_2mortal (newSVsv (string));
1674 1746
1675 SvUPGRADE (string, SVt_PV); 1747 SvUPGRADE (string, SVt_PV);
1676 1748
1677 /* work around a bug in perl 5.10, which causes SvCUR to fail an 1749 /* work around a bug in perl 5.10, which causes SvCUR to fail an
… …
1716 1788
1717 decode_ws (&dec); 1789 decode_ws (&dec);
1718 sv = decode_sv (&dec); 1790 sv = decode_sv (&dec);
1719 1791
1720 if (offset_return) 1792 if (offset_return)
1721 *offset_return = dec.cur; 1793 *offset_return = dec.cur - SvPVX (string);
1722 1794 else if (sv)
1723 if (!(offset_return || !sv))
1724 { 1795 {
1725 // check for trailing garbage 1796 // check for trailing garbage
1726 decode_ws (&dec); 1797 decode_ws (&dec);
1727 1798
1728 if (*dec.cur) 1799 if (dec.cur != dec.end)
1729 { 1800 {
1730 dec.err = "garbage after JSON object"; 1801 dec.err = "garbage after JSON object";
1731 SvREFCNT_dec (sv); 1802 SvREFCNT_dec (sv);
1732 sv = 0; 1803 sv = 0;
1733 } 1804 }
… …
1752 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1823 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1753 } 1824 }
1754 1825
1755 sv = sv_2mortal (sv); 1826 sv = sv_2mortal (sv);
1756 1827
1757 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1828 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1758 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1829 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1759 1830
1760 return sv; 1831 return sv;
1761} 1832}
1762 1833
… …
1771 // 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
1772 // but this would make it slower, so... 1843 // but this would make it slower, so...
1773 1844
1774 for (;;) 1845 for (;;)
1775 { 1846 {
1776 //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
1777 switch (self->incr_mode) 1847 switch (self->incr_mode)
1778 { 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
1779 // only used for initial whitespace skipping 1858 // only used for initial whitespace skipping
1780 case INCR_M_WS: 1859 case INCR_M_WS:
1781 for (;;) 1860 for (;;)
1782 { 1861 {
1783 if (*p > 0x20) 1862 if (*p > 0x20)
… …
1825 ++p; 1904 ++p;
1826 } 1905 }
1827 1906
1828 break; 1907 break;
1829 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
1830 // inside a string 1943 // inside a string
1831 case INCR_M_STR: 1944 case INCR_M_STR:
1832 incr_m_str: 1945 incr_m_str:
1833 for (;;) 1946 for (;;)
1834 { 1947 {
1835 if (*p == '"') 1948 if (*p == '"')
1836 { 1949 {
1837 ++p; 1950 ++p;
1838 self->incr_mode = INCR_M_JSON;
1839
1840 if (!self->incr_nest)
1841 goto interrupt;
1842
1843 goto incr_m_json; 1951 goto end_of_scalar;
1844 } 1952 }
1845 else if (*p == '\\') 1953 else if (*p == '\\')
1846 { 1954 {
1847 ++p; // "virtually" consumes character after \ 1955 ++p; // "virtually" consumes character after \
1848 1956
… …
1878 --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
1879 goto interrupt; 1987 goto interrupt;
1880 } 1988 }
1881 break; 1989 break;
1882 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
1883 case '"': 2006 case '"':
1884 self->incr_mode = INCR_M_STR; 2007 self->incr_mode = INCR_M_STR;
1885 goto incr_m_str; 2008 goto incr_m_str;
1886 2009
1887 case '[': 2010 case '[':
… …
1932 i >= '0' && i <= '9' ? i - '0' 2055 i >= '0' && i <= '9' ? i - '0'
1933 : i >= 'a' && i <= 'f' ? i - 'a' + 10 2056 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1934 : i >= 'A' && i <= 'F' ? i - 'A' + 10 2057 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1935 : -1; 2058 : -1;
1936 2059
1937 json_stash = gv_stashpv ("JSON::XS" , 1); 2060 json_stash = gv_stashpv ("JSON::XS" , 1);
1938 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); 2061 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1939
1940 types_true = get_bool ("Types::Serialiser::true");
1941 types_false = get_bool ("Types::Serialiser::false"); 2062 bool_false = get_bool ("Types::Serialiser::false");
2063 bool_true = get_bool ("Types::Serialiser::true");
1942 2064
1943 sv_json = newSVpv ("JSON", 0); 2065 sv_json = newSVpv ("JSON", 0);
1944 SvREADONLY_on (sv_json); 2066 SvREADONLY_on (sv_json);
1945 2067
1946 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 */
… …
1948 2070
1949PROTOTYPES: DISABLE 2071PROTOTYPES: DISABLE
1950 2072
1951void CLONE (...) 2073void CLONE (...)
1952 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.
1953 json_stash = 0; 2078 json_stash = 0;
1954 types_boolean_stash = 0; 2079 bool_stash = 0;
2080 bool_false = 0;
2081 bool_true = 0;
1955 2082
1956void new (char *klass) 2083void new (char *klass)
1957 PPCODE: 2084 PPCODE:
1958{ 2085{
1959 SV *pv = NEWSV (0, sizeof (JSON)); 2086 SV *pv = NEWSV (0, sizeof (JSON));
… …
1962 XPUSHs (sv_2mortal (sv_bless ( 2089 XPUSHs (sv_2mortal (sv_bless (
1963 newRV_noinc (pv), 2090 newRV_noinc (pv),
1964 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 2091 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1965 ))); 2092 )));
1966} 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 }
1967 2109
1968void ascii (JSON *self, int enable = 1) 2110void ascii (JSON *self, int enable = 1)
1969 ALIAS: 2111 ALIAS:
1970 ascii = F_ASCII 2112 ascii = F_ASCII
1971 latin1 = F_LATIN1 2113 latin1 = F_LATIN1
… …
2064 XPUSHs (ST (0)); 2206 XPUSHs (ST (0));
2065} 2207}
2066 2208
2067void encode (JSON *self, SV *scalar) 2209void encode (JSON *self, SV *scalar)
2068 PPCODE: 2210 PPCODE:
2069 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN; 2211 PUTBACK; XPUSHs (encode_json (scalar, self));
2070 XPUSHs (scalar);
2071 2212
2072void decode (JSON *self, SV *jsonstr) 2213void decode (JSON *self, SV *jsonstr)
2073 PPCODE: 2214 PPCODE:
2074 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN; 2215 PUTBACK; XPUSHs (decode_json (jsonstr, self, 0));
2075 XPUSHs (jsonstr);
2076 2216
2077void decode_prefix (JSON *self, SV *jsonstr) 2217void decode_prefix (JSON *self, SV *jsonstr)
2078 PPCODE: 2218 PPCODE:
2079{ 2219{
2080 SV *sv; 2220 SV *sv;
2081 char *offset; 2221 STRLEN offset;
2082 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; 2222 PUTBACK; sv = decode_json (jsonstr, self, &offset);
2083 EXTEND (SP, 2); 2223 EXTEND (SP, 2);
2084 PUSHs (sv); 2224 PUSHs (sv);
2085 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); 2225 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
2086} 2226}
2087 2227
2088void incr_parse (JSON *self, SV *jsonstr = 0) 2228void incr_parse (JSON *self, SV *jsonstr = 0)
2089 PPCODE: 2229 PPCODE:
2090{ 2230{
… …
2124 { 2264 {
2125 STRLEN len; 2265 STRLEN len;
2126 const char *str = SvPV (jsonstr, len); 2266 const char *str = SvPV (jsonstr, len);
2127 STRLEN cur = SvCUR (self->incr_text); 2267 STRLEN cur = SvCUR (self->incr_text);
2128 2268
2129 if (SvLEN (self->incr_text) <= cur + len) 2269 if (SvLEN (self->incr_text) - cur <= len)
2130 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 2270 json_sv_grow (self->incr_text, cur, len);
2131 2271
2132 Move (str, SvEND (self->incr_text), len, char); 2272 Move (str, SvEND (self->incr_text), len, char);
2133 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 2273 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2134 *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
2135 } 2275 }
… …
2137 2277
2138 if (GIMME_V != G_VOID) 2278 if (GIMME_V != G_VOID)
2139 do 2279 do
2140 { 2280 {
2141 SV *sv; 2281 SV *sv;
2142 char *offset; 2282 STRLEN offset;
2143 2283
2144 if (!INCR_DONE (self)) 2284 if (!INCR_DONE (self))
2145 { 2285 {
2146 incr_parse (self); 2286 incr_parse (self);
2147 2287
… …
2160 2300
2161 break; 2301 break;
2162 } 2302 }
2163 } 2303 }
2164 2304
2165 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; 2305 PUTBACK; sv = decode_json (self->incr_text, self, &offset);
2166 XPUSHs (sv); 2306 XPUSHs (sv);
2167 2307
2168 self->incr_pos -= offset - SvPVX (self->incr_text); 2308 self->incr_pos -= offset;
2169 self->incr_nest = 0; 2309 self->incr_nest = 0;
2170 self->incr_mode = 0; 2310 self->incr_mode = 0;
2171 2311
2172 sv_chop (self->incr_text, offset); 2312 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
2173 } 2313 }
2174 while (GIMME_V == G_ARRAY); 2314 while (GIMME_V == G_ARRAY);
2175} 2315}
2176 2316
2177SV *incr_text (JSON *self) 2317SV *incr_text (JSON *self)
… …
2208 self->incr_mode = 0; 2348 self->incr_mode = 0;
2209} 2349}
2210 2350
2211void DESTROY (JSON *self) 2351void DESTROY (JSON *self)
2212 CODE: 2352 CODE:
2353 SvREFCNT_dec (self->v_false);
2354 SvREFCNT_dec (self->v_true);
2213 SvREFCNT_dec (self->cb_sk_object); 2355 SvREFCNT_dec (self->cb_sk_object);
2214 SvREFCNT_dec (self->cb_object); 2356 SvREFCNT_dec (self->cb_object);
2215 SvREFCNT_dec (self->incr_text); 2357 SvREFCNT_dec (self->incr_text);
2216 2358
2217PROTOTYPES: ENABLE 2359PROTOTYPES: ENABLE
… …
2220 PPCODE: 2362 PPCODE:
2221{ 2363{
2222 JSON json; 2364 JSON json;
2223 json_init (&json); 2365 json_init (&json);
2224 json.flags |= F_UTF8; 2366 json.flags |= F_UTF8;
2225 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN; 2367 PUTBACK; XPUSHs (encode_json (scalar, &json));
2226 XPUSHs (scalar);
2227} 2368}
2228 2369
2229void decode_json (SV *jsonstr) 2370void decode_json (SV *jsonstr)
2230 PPCODE: 2371 PPCODE:
2231{ 2372{
2232 JSON json; 2373 JSON json;
2233 json_init (&json); 2374 json_init (&json);
2234 json.flags |= F_UTF8; 2375 json.flags |= F_UTF8;
2235 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN; 2376 PUTBACK; XPUSHs (decode_json (jsonstr, &json, 0));
2236 XPUSHs (jsonstr);
2237} 2377}
2238 2378

Diff Legend

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