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 1.129 by root, Wed Nov 16 18:06:34 2016 UTC vs.
Revision 1.133 by root, Tue Aug 28 16:16:17 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
… …
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
99 INCR_M_STR, // inside string 101 INCR_M_STR, // inside string
100 INCR_M_BS, // inside backslash 102 INCR_M_BS, // inside backslash
… …
116 // for the incremental parser 118 // for the incremental parser
117 SV *incr_text; // the source text so far 119 SV *incr_text; // the source text so far
118 STRLEN incr_pos; // the current offset into the text 120 STRLEN incr_pos; // the current offset into the text
119 int incr_nest; // {[]}-nesting level 121 int incr_nest; // {[]}-nesting level
120 unsigned char incr_mode; 122 unsigned char incr_mode;
123
124 SV *v_false, *v_true;
121} JSON; 125} JSON;
122 126
123INLINE void 127INLINE void
124json_init (JSON *json) 128json_init (JSON *json)
125{ 129{
… …
154 SvPV_renew (sv, SvCUR (sv) + 1); 158 SvPV_renew (sv, SvCUR (sv) + 1);
155#endif 159#endif
156 } 160 }
157} 161}
158 162
163/* adds two STRLENs together, slow, and with paranoia */
164STRLEN
165strlen_sum (STRLEN l1, STRLEN l2)
166{
167 size_t sum = l1 + l2;
168
169 if (sum < (size_t)l2 || sum != (size_t)(STRLEN)sum)
170 croak ("JSON::XS: string size overflow");
171
172 return sum;
173}
174
175/* similar to SvGROW, but somewhat safer and guarantees exponential realloc strategy */
176static char *
177json_sv_grow (SV *sv, size_t len1, size_t len2)
178{
179 len1 = strlen_sum (len1, len2);
180 len1 = strlen_sum (len1, len1 >> 1);
181
182 if (len1 > 4096 - 24)
183 len1 = (len1 | 4095) - 24;
184
185 return SvGROW (sv, len1);
186}
187
159// decode an utf-8 character and return it, or (UV)-1 in 188// decode a utf-8 character and return it, or (UV)-1 in
160// case of an error. 189// case of an error.
161// we special-case "safe" characters from U+80 .. U+7FF, 190// we special-case "safe" characters from U+80 .. U+7FF,
162// but use the very good perl function to parse anything else. 191// but use the very good perl function to parse anything else.
163// note that we never call this function for a ascii codepoints 192// note that we never call this function for a ascii codepoints
164INLINE UV 193INLINE UV
… …
361} enc_t; 390} enc_t;
362 391
363INLINE void 392INLINE void
364need (enc_t *enc, STRLEN len) 393need (enc_t *enc, STRLEN len)
365{ 394{
366 if (expect_false (enc->cur + len >= enc->end)) 395 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
367 { 396 {
368 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 397 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
369 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 398 char *buf = json_sv_grow (enc->sv, cur, len);
370 enc->cur = SvPVX (enc->sv) + cur; 399 enc->cur = buf + cur;
371 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 400 enc->end = buf + SvLEN (enc->sv) - 1;
372 } 401 }
373} 402}
374 403
375INLINE void 404INLINE void
376encode_ch (enc_t *enc, char ch) 405encode_ch (enc_t *enc, char ch)
… …
392 421
393 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case 422 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
394 { 423 {
395 if (expect_false (ch == '"')) // but with slow exceptions 424 if (expect_false (ch == '"')) // but with slow exceptions
396 { 425 {
397 need (enc, len += 1); 426 need (enc, len + 1);
398 *enc->cur++ = '\\'; 427 *enc->cur++ = '\\';
399 *enc->cur++ = '"'; 428 *enc->cur++ = '"';
400 } 429 }
401 else if (expect_false (ch == '\\')) 430 else if (expect_false (ch == '\\'))
402 { 431 {
403 need (enc, len += 1); 432 need (enc, len + 1);
404 *enc->cur++ = '\\'; 433 *enc->cur++ = '\\';
405 *enc->cur++ = '\\'; 434 *enc->cur++ = '\\';
406 } 435 }
407 else 436 else
408 *enc->cur++ = ch; 437 *enc->cur++ = ch;
… …
411 } 440 }
412 else 441 else
413 { 442 {
414 switch (ch) 443 switch (ch)
415 { 444 {
416 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; 445 case '\010': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
417 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; 446 case '\011': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
418 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; 447 case '\012': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
419 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; 448 case '\014': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
420 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; 449 case '\015': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
421 450
422 default: 451 default:
423 { 452 {
424 STRLEN clen; 453 STRLEN clen;
425 UV uch; 454 UV uch;
… …
441 if (uch >= 0x10000UL) 470 if (uch >= 0x10000UL)
442 { 471 {
443 if (uch >= 0x110000UL) 472 if (uch >= 0x110000UL)
444 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); 473 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
445 474
446 need (enc, len += 11); 475 need (enc, len + 11);
447 sprintf (enc->cur, "\\u%04x\\u%04x", 476 sprintf (enc->cur, "\\u%04x\\u%04x",
448 (int)((uch - 0x10000) / 0x400 + 0xD800), 477 (int)((uch - 0x10000) / 0x400 + 0xD800),
449 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 478 (int)((uch - 0x10000) % 0x400 + 0xDC00));
450 enc->cur += 12; 479 enc->cur += 12;
451 } 480 }
452 else 481 else
453 { 482 {
454 need (enc, len += 5); 483 need (enc, len + 5);
455 *enc->cur++ = '\\'; 484 *enc->cur++ = '\\';
456 *enc->cur++ = 'u'; 485 *enc->cur++ = 'u';
457 *enc->cur++ = PL_hexdigit [ uch >> 12 ]; 486 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
458 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; 487 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
459 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; 488 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
… …
467 *enc->cur++ = uch; 496 *enc->cur++ = uch;
468 str += clen; 497 str += clen;
469 } 498 }
470 else if (is_utf8) 499 else if (is_utf8)
471 { 500 {
472 need (enc, len += clen); 501 need (enc, len + clen);
473 do 502 do
474 { 503 {
475 *enc->cur++ = *str++; 504 *enc->cur++ = *str++;
476 } 505 }
477 while (--clen); 506 while (--clen);
478 } 507 }
479 else 508 else
480 { 509 {
481 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 510 need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed
482 enc->cur = encode_utf8 (enc->cur, uch); 511 enc->cur = encode_utf8 (enc->cur, uch);
483 ++str; 512 ++str;
484 } 513 }
485 } 514 }
486 } 515 }
… …
750 { 779 {
751 HV *stash = SvSTASH (sv); 780 HV *stash = SvSTASH (sv);
752 781
753 if (stash == bool_stash) 782 if (stash == bool_stash)
754 { 783 {
755 if (SvIV (sv))
756 encode_str (enc, "true", 4, 0); 784 if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
757 else
758 encode_str (enc, "false", 5, 0); 785 else encode_str (enc, "false", 5, 0);
759 } 786 }
760 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0))) 787 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
761 { 788 {
762 int count; 789 int count;
763 dSP; 790 dSP;
… …
1167 1194
1168 if (sv) 1195 if (sv)
1169 { 1196 {
1170 STRLEN cur = SvCUR (sv); 1197 STRLEN cur = SvCUR (sv);
1171 1198
1172 if (SvLEN (sv) <= cur + len) 1199 if (SvLEN (sv) - cur <= len)
1173 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 1200 json_sv_grow (sv, cur, len);
1174 1201
1175 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1202 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
1176 SvCUR_set (sv, SvCUR (sv) + len); 1203 SvCUR_set (sv, SvCUR (sv) + len);
1177 } 1204 }
1178 else 1205 else
… …
1640 case '-': 1667 case '-':
1641 case '0': case '1': case '2': case '3': case '4': 1668 case '0': case '1': case '2': case '3': case '4':
1642 case '5': case '6': case '7': case '8': case '9': 1669 case '5': case '6': case '7': case '8': case '9':
1643 return decode_num (dec); 1670 return decode_num (dec);
1644 1671
1672 case 'f':
1673 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1674 {
1675 dec->cur += 5;
1676
1677 if (expect_false (!dec->json.v_false))
1678 dec->json.v_false = GET_BOOL (false);
1679
1680 return newSVsv (dec->json.v_false);
1681 }
1682 else
1683 ERR ("'false' expected");
1684
1685 break;
1686
1645 case 't': 1687 case 't':
1646 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1688 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1647 { 1689 {
1648 dec->cur += 4; 1690 dec->cur += 4;
1649#if JSON_SLOW 1691
1650 bool_true = get_bool ("Types::Serialiser::true"); 1692 if (expect_false (!dec->json.v_true))
1651#endif 1693 dec->json.v_true = GET_BOOL (true);
1694
1652 return newSVsv (bool_true); 1695 return newSVsv (dec->json.v_true);
1653 } 1696 }
1654 else 1697 else
1655 ERR ("'true' expected"); 1698 ERR ("'true' expected");
1656
1657 break;
1658
1659 case 'f':
1660 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1661 {
1662 dec->cur += 5;
1663#if JSON_SLOW
1664 bool_false = get_bool ("Types::Serialiser::false");
1665#endif
1666 return newSVsv (bool_false);
1667 }
1668 else
1669 ERR ("'false' expected");
1670 1699
1671 break; 1700 break;
1672 1701
1673 case 'n': 1702 case 'n':
1674 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1703 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
… …
1969 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1998 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1970 : -1; 1999 : -1;
1971 2000
1972 json_stash = gv_stashpv ("JSON::XS" , 1); 2001 json_stash = gv_stashpv ("JSON::XS" , 1);
1973 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); 2002 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
2003 bool_false = get_bool ("Types::Serialiser::false");
1974 bool_true = get_bool ("Types::Serialiser::true"); 2004 bool_true = get_bool ("Types::Serialiser::true");
1975 bool_false = get_bool ("Types::Serialiser::false");
1976 2005
1977 sv_json = newSVpv ("JSON", 0); 2006 sv_json = newSVpv ("JSON", 0);
1978 SvREADONLY_on (sv_json); 2007 SvREADONLY_on (sv_json);
1979 2008
1980 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ 2009 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
… …
1982 2011
1983PROTOTYPES: DISABLE 2012PROTOTYPES: DISABLE
1984 2013
1985void CLONE (...) 2014void CLONE (...)
1986 CODE: 2015 CODE:
2016 // as long as these writes are atomic, the race should not matter
2017 // as existing threads either already use 0, or use the old value,
2018 // which is sitll correct for the initial thread.
1987 json_stash = 0; 2019 json_stash = 0;
1988 bool_stash = 0; 2020 bool_stash = 0;
2021 bool_false = 0;
2022 bool_true = 0;
1989 2023
1990void new (char *klass) 2024void new (char *klass)
1991 PPCODE: 2025 PPCODE:
1992{ 2026{
1993 SV *pv = NEWSV (0, sizeof (JSON)); 2027 SV *pv = NEWSV (0, sizeof (JSON));
… …
1996 XPUSHs (sv_2mortal (sv_bless ( 2030 XPUSHs (sv_2mortal (sv_bless (
1997 newRV_noinc (pv), 2031 newRV_noinc (pv),
1998 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 2032 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1999 ))); 2033 )));
2000} 2034}
2035
2036void boolean_values (JSON *self, SV *v_false = 0, SV *v_true = 0)
2037 PPCODE:
2038 self->v_false = newSVsv (v_false);
2039 self->v_true = newSVsv (v_true);
2040 XPUSHs (ST (0));
2041
2042void get_boolean_values (JSON *self)
2043 PPCODE:
2044 if (self->v_false && self->v_true)
2045 {
2046 EXTEND (SP, 2);
2047 PUSHs (self->v_false);
2048 PUSHs (self->v_true);
2049 }
2001 2050
2002void ascii (JSON *self, int enable = 1) 2051void ascii (JSON *self, int enable = 1)
2003 ALIAS: 2052 ALIAS:
2004 ascii = F_ASCII 2053 ascii = F_ASCII
2005 latin1 = F_LATIN1 2054 latin1 = F_LATIN1
… …
2158 { 2207 {
2159 STRLEN len; 2208 STRLEN len;
2160 const char *str = SvPV (jsonstr, len); 2209 const char *str = SvPV (jsonstr, len);
2161 STRLEN cur = SvCUR (self->incr_text); 2210 STRLEN cur = SvCUR (self->incr_text);
2162 2211
2163 if (SvLEN (self->incr_text) <= cur + len) 2212 if (SvLEN (self->incr_text) - cur <= len)
2164 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 2213 json_sv_grow (self->incr_text, cur, len);
2165 2214
2166 Move (str, SvEND (self->incr_text), len, char); 2215 Move (str, SvEND (self->incr_text), len, char);
2167 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 2216 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2168 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there 2217 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2169 } 2218 }
… …
2242 self->incr_mode = 0; 2291 self->incr_mode = 0;
2243} 2292}
2244 2293
2245void DESTROY (JSON *self) 2294void DESTROY (JSON *self)
2246 CODE: 2295 CODE:
2296 SvREFCNT_dec (self->v_false);
2297 SvREFCNT_dec (self->v_true);
2247 SvREFCNT_dec (self->cb_sk_object); 2298 SvREFCNT_dec (self->cb_sk_object);
2248 SvREFCNT_dec (self->cb_object); 2299 SvREFCNT_dec (self->cb_object);
2249 SvREFCNT_dec (self->incr_text); 2300 SvREFCNT_dec (self->incr_text);
2250 2301
2251PROTOTYPES: ENABLE 2302PROTOTYPES: ENABLE

Diff Legend

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