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 1.130 by root, Sat Nov 26 06:09:29 2016 UTC vs.
Revision 1.133 by root, Tue Aug 28 16:16:17 2018 UTC

… …
78 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) 78 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
79 79
80#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?)"
81 81
82#ifdef USE_ITHREADS 82#ifdef USE_ITHREADS
83# define JSON_SLOW 1
84# 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))
85# 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))
86#else 86#else
87# define JSON_SLOW 0
88# define JSON_STASH json_stash 87# define JSON_STASH json_stash
89# define BOOL_STASH bool_stash 88# define BOOL_STASH bool_stash
89# define GET_BOOL(value) bool_ ## value
90#endif 90#endif
91 91
92// 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
93#define STACK_HES 64 93#define STACK_HES 64
94 94
95static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean:: 95static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean::
96static SV *bool_true, *bool_false, *sv_json; 96static SV *bool_false, *bool_true;
97static SV *sv_json;
97 98
98enum { 99enum {
99 INCR_M_WS = 0, // initial whitespace skipping, must be 0 100 INCR_M_WS = 0, // initial whitespace skipping, must be 0
100 INCR_M_STR, // inside string 101 INCR_M_STR, // inside string
101 INCR_M_BS, // inside backslash 102 INCR_M_BS, // inside backslash
… …
117 // for the incremental parser 118 // for the incremental parser
118 SV *incr_text; // the source text so far 119 SV *incr_text; // the source text so far
119 STRLEN incr_pos; // the current offset into the text 120 STRLEN incr_pos; // the current offset into the text
120 int incr_nest; // {[]}-nesting level 121 int incr_nest; // {[]}-nesting level
121 unsigned char incr_mode; 122 unsigned char incr_mode;
123
124 SV *v_false, *v_true;
122} JSON; 125} JSON;
123 126
124INLINE void 127INLINE void
125json_init (JSON *json) 128json_init (JSON *json)
126{ 129{
… …
155 SvPV_renew (sv, SvCUR (sv) + 1); 158 SvPV_renew (sv, SvCUR (sv) + 1);
156#endif 159#endif
157 } 160 }
158} 161}
159 162
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
160// 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
161// case of an error. 189// case of an error.
162// we special-case "safe" characters from U+80 .. U+7FF, 190// we special-case "safe" characters from U+80 .. U+7FF,
163// but use the very good perl function to parse anything else. 191// but use the very good perl function to parse anything else.
164// note that we never call this function for a ascii codepoints 192// note that we never call this function for a ascii codepoints
165INLINE UV 193INLINE UV
… …
365need (enc_t *enc, STRLEN len) 393need (enc_t *enc, STRLEN len)
366{ 394{
367 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len)) 395 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
368 { 396 {
369 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 397 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
370 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 398 char *buf = json_sv_grow (enc->sv, cur, len);
371 enc->cur = SvPVX (enc->sv) + cur; 399 enc->cur = buf + cur;
372 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 400 enc->end = buf + SvLEN (enc->sv) - 1;
373 } 401 }
374} 402}
375 403
376INLINE void 404INLINE void
377encode_ch (enc_t *enc, char ch) 405encode_ch (enc_t *enc, char ch)
… …
393 421
394 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case 422 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
395 { 423 {
396 if (expect_false (ch == '"')) // but with slow exceptions 424 if (expect_false (ch == '"')) // but with slow exceptions
397 { 425 {
398 need (enc, len += 1); 426 need (enc, len + 1);
399 *enc->cur++ = '\\'; 427 *enc->cur++ = '\\';
400 *enc->cur++ = '"'; 428 *enc->cur++ = '"';
401 } 429 }
402 else if (expect_false (ch == '\\')) 430 else if (expect_false (ch == '\\'))
403 { 431 {
404 need (enc, len += 1); 432 need (enc, len + 1);
405 *enc->cur++ = '\\'; 433 *enc->cur++ = '\\';
406 *enc->cur++ = '\\'; 434 *enc->cur++ = '\\';
407 } 435 }
408 else 436 else
409 *enc->cur++ = ch; 437 *enc->cur++ = ch;
… …
412 } 440 }
413 else 441 else
414 { 442 {
415 switch (ch) 443 switch (ch)
416 { 444 {
417 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;
418 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;
419 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;
420 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;
421 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;
422 450
423 default: 451 default:
424 { 452 {
425 STRLEN clen; 453 STRLEN clen;
426 UV uch; 454 UV uch;
… …
442 if (uch >= 0x10000UL) 470 if (uch >= 0x10000UL)
443 { 471 {
444 if (uch >= 0x110000UL) 472 if (uch >= 0x110000UL)
445 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);
446 474
447 need (enc, len += 11); 475 need (enc, len + 11);
448 sprintf (enc->cur, "\\u%04x\\u%04x", 476 sprintf (enc->cur, "\\u%04x\\u%04x",
449 (int)((uch - 0x10000) / 0x400 + 0xD800), 477 (int)((uch - 0x10000) / 0x400 + 0xD800),
450 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 478 (int)((uch - 0x10000) % 0x400 + 0xDC00));
451 enc->cur += 12; 479 enc->cur += 12;
452 } 480 }
453 else 481 else
454 { 482 {
455 need (enc, len += 5); 483 need (enc, len + 5);
456 *enc->cur++ = '\\'; 484 *enc->cur++ = '\\';
457 *enc->cur++ = 'u'; 485 *enc->cur++ = 'u';
458 *enc->cur++ = PL_hexdigit [ uch >> 12 ]; 486 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
459 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; 487 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
460 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; 488 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
… …
468 *enc->cur++ = uch; 496 *enc->cur++ = uch;
469 str += clen; 497 str += clen;
470 } 498 }
471 else if (is_utf8) 499 else if (is_utf8)
472 { 500 {
473 need (enc, len += clen); 501 need (enc, len + clen);
474 do 502 do
475 { 503 {
476 *enc->cur++ = *str++; 504 *enc->cur++ = *str++;
477 } 505 }
478 while (--clen); 506 while (--clen);
479 } 507 }
480 else 508 else
481 { 509 {
482 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
483 enc->cur = encode_utf8 (enc->cur, uch); 511 enc->cur = encode_utf8 (enc->cur, uch);
484 ++str; 512 ++str;
485 } 513 }
486 } 514 }
487 } 515 }
… …
751 { 779 {
752 HV *stash = SvSTASH (sv); 780 HV *stash = SvSTASH (sv);
753 781
754 if (stash == bool_stash) 782 if (stash == bool_stash)
755 { 783 {
756 if (SvIV (sv))
757 encode_str (enc, "true", 4, 0); 784 if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
758 else
759 encode_str (enc, "false", 5, 0); 785 else encode_str (enc, "false", 5, 0);
760 } 786 }
761 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)))
762 { 788 {
763 int count; 789 int count;
764 dSP; 790 dSP;
… …
1168 1194
1169 if (sv) 1195 if (sv)
1170 { 1196 {
1171 STRLEN cur = SvCUR (sv); 1197 STRLEN cur = SvCUR (sv);
1172 1198
1173 if (SvLEN (sv) <= cur + len) 1199 if (SvLEN (sv) - cur <= len)
1174 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 1200 json_sv_grow (sv, cur, len);
1175 1201
1176 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1202 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
1177 SvCUR_set (sv, SvCUR (sv) + len); 1203 SvCUR_set (sv, SvCUR (sv) + len);
1178 } 1204 }
1179 else 1205 else
… …
1641 case '-': 1667 case '-':
1642 case '0': case '1': case '2': case '3': case '4': 1668 case '0': case '1': case '2': case '3': case '4':
1643 case '5': case '6': case '7': case '8': case '9': 1669 case '5': case '6': case '7': case '8': case '9':
1644 return decode_num (dec); 1670 return decode_num (dec);
1645 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
1646 case 't': 1687 case 't':
1647 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1688 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1648 { 1689 {
1649 dec->cur += 4; 1690 dec->cur += 4;
1650#if JSON_SLOW 1691
1651 bool_true = get_bool ("Types::Serialiser::true"); 1692 if (expect_false (!dec->json.v_true))
1652#endif 1693 dec->json.v_true = GET_BOOL (true);
1694
1653 return newSVsv (bool_true); 1695 return newSVsv (dec->json.v_true);
1654 } 1696 }
1655 else 1697 else
1656 ERR ("'true' expected"); 1698 ERR ("'true' expected");
1657
1658 break;
1659
1660 case 'f':
1661 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1662 {
1663 dec->cur += 5;
1664#if JSON_SLOW
1665 bool_false = get_bool ("Types::Serialiser::false");
1666#endif
1667 return newSVsv (bool_false);
1668 }
1669 else
1670 ERR ("'false' expected");
1671 1699
1672 break; 1700 break;
1673 1701
1674 case 'n': 1702 case 'n':
1675 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) 1703 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
… …
1970 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1998 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1971 : -1; 1999 : -1;
1972 2000
1973 json_stash = gv_stashpv ("JSON::XS" , 1); 2001 json_stash = gv_stashpv ("JSON::XS" , 1);
1974 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");
1975 bool_true = get_bool ("Types::Serialiser::true"); 2004 bool_true = get_bool ("Types::Serialiser::true");
1976 bool_false = get_bool ("Types::Serialiser::false");
1977 2005
1978 sv_json = newSVpv ("JSON", 0); 2006 sv_json = newSVpv ("JSON", 0);
1979 SvREADONLY_on (sv_json); 2007 SvREADONLY_on (sv_json);
1980 2008
1981 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 */
… …
1983 2011
1984PROTOTYPES: DISABLE 2012PROTOTYPES: DISABLE
1985 2013
1986void CLONE (...) 2014void CLONE (...)
1987 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.
1988 json_stash = 0; 2019 json_stash = 0;
1989 bool_stash = 0; 2020 bool_stash = 0;
2021 bool_false = 0;
2022 bool_true = 0;
1990 2023
1991void new (char *klass) 2024void new (char *klass)
1992 PPCODE: 2025 PPCODE:
1993{ 2026{
1994 SV *pv = NEWSV (0, sizeof (JSON)); 2027 SV *pv = NEWSV (0, sizeof (JSON));
… …
1997 XPUSHs (sv_2mortal (sv_bless ( 2030 XPUSHs (sv_2mortal (sv_bless (
1998 newRV_noinc (pv), 2031 newRV_noinc (pv),
1999 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) 2032 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
2000 ))); 2033 )));
2001} 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 }
2002 2050
2003void ascii (JSON *self, int enable = 1) 2051void ascii (JSON *self, int enable = 1)
2004 ALIAS: 2052 ALIAS:
2005 ascii = F_ASCII 2053 ascii = F_ASCII
2006 latin1 = F_LATIN1 2054 latin1 = F_LATIN1
… …
2159 { 2207 {
2160 STRLEN len; 2208 STRLEN len;
2161 const char *str = SvPV (jsonstr, len); 2209 const char *str = SvPV (jsonstr, len);
2162 STRLEN cur = SvCUR (self->incr_text); 2210 STRLEN cur = SvCUR (self->incr_text);
2163 2211
2164 if (SvLEN (self->incr_text) <= cur + len) 2212 if (SvLEN (self->incr_text) - cur <= len)
2165 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 2213 json_sv_grow (self->incr_text, cur, len);
2166 2214
2167 Move (str, SvEND (self->incr_text), len, char); 2215 Move (str, SvEND (self->incr_text), len, char);
2168 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 2216 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2169 *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
2170 } 2218 }
… …
2243 self->incr_mode = 0; 2291 self->incr_mode = 0;
2244} 2292}
2245 2293
2246void DESTROY (JSON *self) 2294void DESTROY (JSON *self)
2247 CODE: 2295 CODE:
2296 SvREFCNT_dec (self->v_false);
2297 SvREFCNT_dec (self->v_true);
2248 SvREFCNT_dec (self->cb_sk_object); 2298 SvREFCNT_dec (self->cb_sk_object);
2249 SvREFCNT_dec (self->cb_object); 2299 SvREFCNT_dec (self->cb_object);
2250 SvREFCNT_dec (self->incr_text); 2300 SvREFCNT_dec (self->incr_text);
2251 2301
2252PROTOTYPES: ENABLE 2302PROTOTYPES: ENABLE

Diff Legend

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