[ViewVC] Diff of: cvs/JSON-XS/XS.xs

Comparing JSON-XS/XS.xs (file contents):
Revision 1.129 by root, Wed Nov 16 18:06:34 2016 UTC vs.
Revision 1.143 by root, Mon Jun 21 01:04:30 2021 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
94	static HV json_stash, bool_stash; // JSON::XS::, Types::Serialiser::Boolean::	95	static HV json_stash, bool_stash; // JSON::XS::, Types::Serialiser::Boolean::
95	static SV bool_true, bool_false, *sv_json;	96	static SV bool_false, bool_true;
		97	static SV *sv_json;
96		98
97	enum {	99	enum {
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
123	INLINE void	129	INLINE void
124	json_init (JSON *json)	130	json_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 */
		167	static STRLEN
		168	strlen_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 */
		179	static char *
		180	json_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
164	INLINE UV	196	INLINE UV
…		…
361	} enc_t;	393	} enc_t;
362		394
363	INLINE void	395	INLINE void
364	need (enc_t *enc, STRLEN len)	396	need (enc_t *enc, STRLEN len)
365	{	397	{
366	if (expect_false (enc->cur + len >= enc->end))	398	if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
367	{	399	{
368	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);	400	STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
369	SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	401	char *buf = json_sv_grow (enc->sv, cur, len);
370	enc->cur = SvPVX (enc->sv) + cur;	402	enc->cur = buf + cur;
371	enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;	403	enc->end = buf + SvLEN (enc->sv) - 1;
372	}	404	}
373	}	405	}
374		406
375	INLINE void	407	INLINE void
376	encode_ch (enc_t *enc, char ch)	408	encode_ch (enc_t *enc, char ch)
…		…
392		424
393	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case	425	if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
394	{	426	{
395	if (expect_false (ch == '"')) // but with slow exceptions	427	if (expect_false (ch == '"')) // but with slow exceptions
396	{	428	{
397	need (enc, len += 1);	429	need (enc, len + 1);
398	*enc->cur++ = '\\';	430	*enc->cur++ = '\\';
399	*enc->cur++ = '"';	431	*enc->cur++ = '"';
400	}	432	}
401	else if (expect_false (ch == '\\'))	433	else if (expect_false (ch == '\\'))
402	{	434	{
403	need (enc, len += 1);	435	need (enc, len + 1);
404	*enc->cur++ = '\\';	436	*enc->cur++ = '\\';
405	*enc->cur++ = '\\';	437	*enc->cur++ = '\\';
406	}	438	}
407	else	439	else
408	*enc->cur++ = ch;	440	*enc->cur++ = ch;
…		…
411	}	443	}
412	else	444	else
413	{	445	{
414	switch (ch)	446	switch (ch)
415	{	447	{
416	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;
417	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;
418	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;
419	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;
420	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;
421		453
422	default:	454	default:
423	{	455	{
424	STRLEN clen;	456	STRLEN clen;
425	UV uch;	457	UV uch;
…		…
441	if (uch >= 0x10000UL)	473	if (uch >= 0x10000UL)
442	{	474	{
443	if (uch >= 0x110000UL)	475	if (uch >= 0x110000UL)
444	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);
445		477
446	need (enc, len += 11);	478	need (enc, len + 11);
447	sprintf (enc->cur, "\\u%04x\\u%04x",	479	sprintf (enc->cur, "\\u%04x\\u%04x",
448	(int)((uch - 0x10000) / 0x400 + 0xD800),	480	(int)((uch - 0x10000) / 0x400 + 0xD800),
449	(int)((uch - 0x10000) % 0x400 + 0xDC00));	481	(int)((uch - 0x10000) % 0x400 + 0xDC00));
450	enc->cur += 12;	482	enc->cur += 12;
451	}	483	}
452	else	484	else
453	{	485	{
454	need (enc, len += 5);	486	need (enc, len + 5);
455	*enc->cur++ = '\\';	487	*enc->cur++ = '\\';
456	*enc->cur++ = 'u';	488	*enc->cur++ = 'u';
457	*enc->cur++ = PL_hexdigit [ uch >> 12 ];	489	*enc->cur++ = PL_hexdigit [ uch >> 12 ];
458	*enc->cur++ = PL_hexdigit [(uch >> 8) & 15];	490	*enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
459	*enc->cur++ = PL_hexdigit [(uch >> 4) & 15];	491	*enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
…		…
467	*enc->cur++ = uch;	499	*enc->cur++ = uch;
468	str += clen;	500	str += clen;
469	}	501	}
470	else if (is_utf8)	502	else if (is_utf8)
471	{	503	{
472	need (enc, len += clen);	504	need (enc, len + clen);
473	do	505	do
474	{	506	{
475	enc->cur++ = str++;	507	enc->cur++ = str++;
476	}	508	}
477	while (--clen);	509	while (--clen);
478	}	510	}
479	else	511	else
480	{	512	{
481	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
482	enc->cur = encode_utf8 (enc->cur, uch);	514	enc->cur = encode_utf8 (enc->cur, uch);
483	++str;	515	++str;
484	}	516	}
485	}	517	}
486	}	518	}
…		…
750	{	782	{
751	HV *stash = SvSTASH (sv);	783	HV *stash = SvSTASH (sv);
752		784
753	if (stash == bool_stash)	785	if (stash == bool_stash)
754	{	786	{
755	if (SvIV (sv))
756	encode_str (enc, "true", 4, 0);	787	if (SvIV (sv)) encode_str (enc, "true" , 4, 0);
757	else
758	encode_str (enc, "false", 5, 0);	788	else encode_str (enc, "false", 5, 0);
759	}	789	}
760	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)))
761	{	791	{
762	int count;	792	int count;
763	dSP;	793	dSP;
764		794
765	ENTER; SAVETMPS;	795	ENTER; SAVETMPS;
766	SAVESTACK_POS ();
767	PUSHMARK (SP);	796	PUSHMARK (SP);
768	EXTEND (SP, 2);	797	EXTEND (SP, 2);
769	// 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
770	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));	799	PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
771	PUSHs (sv_json);	800	PUSHs (sv_json);
…		…
783	encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));	812	encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
784	encode_ch (enc, '"');	813	encode_ch (enc, '"');
785	encode_ch (enc, ')');	814	encode_ch (enc, ')');
786	encode_ch (enc, '[');	815	encode_ch (enc, '[');
787		816
788	while (count)	817	if (count)
789	{	818	{
		819	int i;
		820
		821	for (i = 0; i < count - 1; ++i)
		822	{
790	encode_sv (enc, SP[1 - count--]);	823	encode_sv (enc, SP[i + 1 - count]);
791
792	if (count)
793	encode_ch (enc, ',');	824	encode_ch (enc, ',');
		825	}
		826
		827	encode_sv (enc, TOPs);
		828	SP -= count;
794	}	829	}
		830
		831	PUTBACK;
795		832
796	encode_ch (enc, ']');	833	encode_ch (enc, ']');
797		834
798	FREETMPS; LEAVE;	835	FREETMPS; LEAVE;
799	}	836	}
…		…
881	if (SvIsUV (sv) ? SvUVX (sv) <= 59000	918	if (SvIsUV (sv) ? SvUVX (sv) <= 59000
882	: SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)	919	: SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
883	{	920	{
884	// optimise the "small number case"	921	// optimise the "small number case"
885	// code will likely be branchless and use only a single multiplication	922	// code will likely be branchless and use only a single multiplication
886	// works for numbers up to 59074	923	// 4.28 works for numbers up to 59074
887	I32 i = SvIVX (sv);	924	I32 i = SvIVX (sv);
888	U32 u;	925	U32 u;
889	char digit, nz = 0;	926	char digit, nz = 0;
890		927
891	need (enc, 6);	928	need (enc, 6);
…		…
1001	else	1038	else
1002	break;	1039	break;
1003	}	1040	}
1004	else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)	1041	else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
1005	break; // parse error, but let higher level handle it, gives better error messages	1042	break; // parse error, but let higher level handle it, gives better error messages
1006		1043	else
1007	++dec->cur;	1044	++dec->cur;
1008	}	1045	}
1009	}	1046	}
1010		1047
1011	#define ERR(reason) SB dec->err = reason; goto fail; SE	1048	#define ERR(reason) SB dec->err = reason; goto fail; SE
1012		1049
…		…
1167		1204
1168	if (sv)	1205	if (sv)
1169	{	1206	{
1170	STRLEN cur = SvCUR (sv);	1207	STRLEN cur = SvCUR (sv);
1171		1208
1172	if (SvLEN (sv) <= cur + len)	1209	if (SvLEN (sv) - cur <= len)
1173	SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	1210	json_sv_grow (sv, cur, len);
1174		1211
1175	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);	1212	memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
1176	SvCUR_set (sv, SvCUR (sv) + len);	1213	SvCUR_set (sv, SvCUR (sv) + len);
1177	}	1214	}
1178	else	1215	else
…		…
1469		1506
1470	DEC_DEC_DEPTH;	1507	DEC_DEC_DEPTH;
1471	sv = newRV_noinc ((SV *)hv);	1508	sv = newRV_noinc ((SV *)hv);
1472		1509
1473	// check filter callbacks	1510	// check filter callbacks
1474	if (dec->json.flags & F_HOOK)	1511	if (expect_false (dec->json.flags & F_HOOK))
1475	{	1512	{
1476	if (dec->json.cb_sk_object && HvKEYS (hv) == 1)	1513	if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1477	{	1514	{
1478	HE cb, he;	1515	HE cb, he;
1479		1516
…		…
1489	{	1526	{
1490	dSP;	1527	dSP;
1491	int count;	1528	int count;
1492		1529
1493	ENTER; SAVETMPS;	1530	ENTER; SAVETMPS;
1494	SAVESTACK_POS ();
1495	PUSHMARK (SP);	1531	PUSHMARK (SP);
1496	XPUSHs (HeVAL (he));	1532	XPUSHs (HeVAL (he));
1497	sv_2mortal (sv);	1533	sv_2mortal (sv);
1498		1534
1499	PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;	1535	PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1500		1536
1501	if (count == 1)	1537	if (count == 1)
1502	{	1538	{
1503	sv = newSVsv (POPs);	1539	sv = newSVsv (POPs);
		1540	PUTBACK;
1504	FREETMPS; LEAVE;	1541	FREETMPS; LEAVE;
1505	return sv;	1542	return sv;
1506	}	1543	}
		1544	else if (count)
		1545	croak ("filter_json_single_key_object callbacks must not return more than one scalar");
		1546
		1547	PUTBACK;
1507		1548
1508	SvREFCNT_inc (sv);	1549	SvREFCNT_inc (sv);
		1550
1509	FREETMPS; LEAVE;	1551	FREETMPS; LEAVE;
1510	}	1552	}
1511	}	1553	}
1512		1554
1513	if (dec->json.cb_object)	1555	if (dec->json.cb_object)
1514	{	1556	{
1515	dSP;	1557	dSP;
1516	int count;	1558	int count;
1517		1559
1518	ENTER; SAVETMPS;	1560	ENTER; SAVETMPS;
1519	SAVESTACK_POS ();
1520	PUSHMARK (SP);	1561	PUSHMARK (SP);
1521	XPUSHs (sv_2mortal (sv));	1562	XPUSHs (sv_2mortal (sv));
1522		1563
1523	PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;	1564	PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1524		1565
1525	if (count == 1)	1566	if (count == 1)
1526	{
1527	sv = newSVsv (POPs);	1567	sv = newSVsv (POPs);
1528	FREETMPS; LEAVE;	1568	else if (count == 0)
1529	return sv;
1530	}
1531
1532	SvREFCNT_inc (sv);	1569	SvREFCNT_inc (sv);
		1570	else
		1571	croak ("filter_json_object callbacks must not return more than one scalar");
		1572
		1573	PUTBACK;
		1574
1533	FREETMPS; LEAVE;	1575	FREETMPS; LEAVE;
1534	}	1576	}
1535	}	1577	}
1536		1578
1537	return sv;	1579	return sv;
…		…
1640	case '-':	1682	case '-':
1641	case '0': case '1': case '2': case '3': case '4':	1683	case '0': case '1': case '2': case '3': case '4':
1642	case '5': case '6': case '7': case '8': case '9':	1684	case '5': case '6': case '7': case '8': case '9':
1643	return decode_num (dec);	1685	return decode_num (dec);
1644		1686
		1687	case 'f':
		1688	if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
		1689	{
		1690	dec->cur += 5;
		1691
		1692	if (expect_false (!dec->json.v_false))
		1693	dec->json.v_false = GET_BOOL (false);
		1694
		1695	return newSVsv (dec->json.v_false);
		1696	}
		1697	else
		1698	ERR ("'false' expected");
		1699
		1700	break;
		1701
1645	case 't':	1702	case 't':
1646	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))	1703	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1647	{	1704	{
1648	dec->cur += 4;	1705	dec->cur += 4;
1649	#if JSON_SLOW	1706
1650	bool_true = get_bool ("Types::Serialiser::true");	1707	if (expect_false (!dec->json.v_true))
1651	#endif	1708	dec->json.v_true = GET_BOOL (true);
		1709
1652	return newSVsv (bool_true);	1710	return newSVsv (dec->json.v_true);
1653	}	1711	}
1654	else	1712	else
1655	ERR ("'true' expected");	1713	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		1714
1671	break;	1715	break;
1672		1716
1673	case 'n':	1717	case 'n':
1674	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))	1718	if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4))
…		…
1758	else if (sv)	1802	else if (sv)
1759	{	1803	{
1760	// check for trailing garbage	1804	// check for trailing garbage
1761	decode_ws (&dec);	1805	decode_ws (&dec);
1762		1806
1763	if (*dec.cur)	1807	if (dec.cur != dec.end)
1764	{	1808	{
1765	dec.err = "garbage after JSON object";	1809	dec.err = "garbage after JSON object";
1766	SvREFCNT_dec (sv);	1810	SvREFCNT_dec (sv);
1767	sv = 0;	1811	sv = 0;
1768	}	1812	}
…		…
1806	// the state machine here is a bit convoluted and could be simplified a lot	1850	// the state machine here is a bit convoluted and could be simplified a lot
1807	// but this would make it slower, so...	1851	// but this would make it slower, so...
1808		1852
1809	for (;;)	1853	for (;;)
1810	{	1854	{
1811	//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
1812	switch (self->incr_mode)	1855	switch (self->incr_mode)
1813	{	1856	{
		1857	// reached end of a scalar, see if we are inside a nested structure or not
		1858	end_of_scalar:
		1859	self->incr_mode = INCR_M_JSON;
		1860
		1861	if (self->incr_nest) // end of a scalar inside array, object or tag
		1862	goto incr_m_json;
		1863	else // end of scalar outside structure, json text ends here
		1864	goto interrupt;
		1865
1814	// only used for initial whitespace skipping	1866	// only used for initial whitespace skipping
1815	case INCR_M_WS:	1867	case INCR_M_WS:
1816	for (;;)	1868	for (;;)
1817	{	1869	{
1818	if (*p > 0x20)	1870	if (*p > 0x20)
…		…
1860	++p;	1912	++p;
1861	}	1913	}
1862		1914
1863	break;	1915	break;
1864		1916
		1917	// inside true/false/null
		1918	case INCR_M_TFN:
		1919	incr_m_tfn:
		1920	for (;;)
		1921	switch (*p++)
		1922	{
		1923	case 'r': case 'u': case 'e': // tRUE, falsE, nUll
		1924	case 'a': case 'l': case 's': // fALSe, nuLL
		1925	// allowed
		1926	break;
		1927
		1928	default:
		1929	--p;
		1930	goto end_of_scalar;
		1931	}
		1932
		1933	// inside a number
		1934	case INCR_M_NUM:
		1935	incr_m_num:
		1936	for (;;)
		1937	switch (*p++)
		1938	{
		1939	case 'e': case 'E': case '.': case '+':
		1940	case '-':
		1941	case '0': case '1': case '2': case '3': case '4':
		1942	case '5': case '6': case '7': case '8': case '9':
		1943	// allowed
		1944	break;
		1945
		1946	default:
		1947	--p;
		1948	goto end_of_scalar;
		1949	}
		1950
1865	// inside a string	1951	// inside a string
1866	case INCR_M_STR:	1952	case INCR_M_STR:
1867	incr_m_str:	1953	incr_m_str:
1868	for (;;)	1954	for (;;)
1869	{	1955	{
1870	if (*p == '"')	1956	if (*p == '"')
1871	{	1957	{
1872	++p;	1958	++p;
1873	self->incr_mode = INCR_M_JSON;
1874
1875	if (!self->incr_nest)
1876	goto interrupt;
1877
1878	goto incr_m_json;	1959	goto end_of_scalar;
1879	}	1960	}
1880	else if (*p == '\\')	1961	else if (*p == '\\')
1881	{	1962	{
1882	++p; // "virtually" consumes character after \	1963	++p; // "virtually" consumes character after \
1883		1964
…		…
1913	--p; // do not eat the whitespace, let the next round do it	1994	--p; // do not eat the whitespace, let the next round do it
1914	goto interrupt;	1995	goto interrupt;
1915	}	1996	}
1916	break;	1997	break;
1917		1998
		1999	// the following three blocks handle scalars. this makes the parser
		2000	// more strict than required inside arrays or objects, and could
		2001	// be moved to a special case on the toplevel (except strings)
		2002	case 't':
		2003	case 'f':
		2004	case 'n':
		2005	self->incr_mode = INCR_M_TFN;
		2006	goto incr_m_tfn;
		2007
		2008	case '-':
		2009	case '0': case '1': case '2': case '3': case '4':
		2010	case '5': case '6': case '7': case '8': case '9':
		2011	self->incr_mode = INCR_M_NUM;
		2012	goto incr_m_num;
		2013
1918	case '"':	2014	case '"':
1919	self->incr_mode = INCR_M_STR;	2015	self->incr_mode = INCR_M_STR;
1920	goto incr_m_str;	2016	goto incr_m_str;
1921		2017
1922	case '[':	2018	case '[':
…		…
1969	: i >= 'A' && i <= 'F' ? i - 'A' + 10	2065	: i >= 'A' && i <= 'F' ? i - 'A' + 10
1970	: -1;	2066	: -1;
1971		2067
1972	json_stash = gv_stashpv ("JSON::XS" , 1);	2068	json_stash = gv_stashpv ("JSON::XS" , 1);
1973	bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);	2069	bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
		2070	bool_false = get_bool ("Types::Serialiser::false");
1974	bool_true = get_bool ("Types::Serialiser::true");	2071	bool_true = get_bool ("Types::Serialiser::true");
1975	bool_false = get_bool ("Types::Serialiser::false");
1976		2072
1977	sv_json = newSVpv ("JSON", 0);	2073	sv_json = newSVpv ("JSON", 0);
1978	SvREADONLY_on (sv_json);	2074	SvREADONLY_on (sv_json);
1979		2075
1980	CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */	2076	CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
…		…
1982		2078
1983	PROTOTYPES: DISABLE	2079	PROTOTYPES: DISABLE
1984		2080
1985	void CLONE (...)	2081	void CLONE (...)
1986	CODE:	2082	CODE:
		2083	// as long as these writes are atomic, the race should not matter
		2084	// as existing threads either already use 0, or use the old value,
		2085	// which is sitll correct for the initial thread.
1987	json_stash = 0;	2086	json_stash = 0;
1988	bool_stash = 0;	2087	bool_stash = 0;
		2088	bool_false = 0;
		2089	bool_true = 0;
1989		2090
1990	void new (char *klass)	2091	void new (char *klass)
1991	PPCODE:	2092	PPCODE:
1992	{	2093	{
1993	SV *pv = NEWSV (0, sizeof (JSON));	2094	SV *pv = NEWSV (0, sizeof (JSON));
…		…
1996	XPUSHs (sv_2mortal (sv_bless (	2097	XPUSHs (sv_2mortal (sv_bless (
1997	newRV_noinc (pv),	2098	newRV_noinc (pv),
1998	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)	2099	strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1999	)));	2100	)));
2000	}	2101	}
		2102
		2103	void boolean_values (JSON self, SV v_false = 0, SV *v_true = 0)
		2104	PPCODE:
		2105	self->v_false = newSVsv (v_false);
		2106	self->v_true = newSVsv (v_true);
		2107	XPUSHs (ST (0));
		2108
		2109	void get_boolean_values (JSON *self)
		2110	PPCODE:
		2111	if (self->v_false && self->v_true)
		2112	{
		2113	EXTEND (SP, 2);
		2114	PUSHs (self->v_false);
		2115	PUSHs (self->v_true);
		2116	}
2001		2117
2002	void ascii (JSON *self, int enable = 1)	2118	void ascii (JSON *self, int enable = 1)
2003	ALIAS:	2119	ALIAS:
2004	ascii = F_ASCII	2120	ascii = F_ASCII
2005	latin1 = F_LATIN1	2121	latin1 = F_LATIN1
…		…
2158	{	2274	{
2159	STRLEN len;	2275	STRLEN len;
2160	const char *str = SvPV (jsonstr, len);	2276	const char *str = SvPV (jsonstr, len);
2161	STRLEN cur = SvCUR (self->incr_text);	2277	STRLEN cur = SvCUR (self->incr_text);
2162		2278
2163	if (SvLEN (self->incr_text) <= cur + len)	2279	if (SvLEN (self->incr_text) - cur <= len)
2164	SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);	2280	json_sv_grow (self->incr_text, cur, len);
2165		2281
2166	Move (str, SvEND (self->incr_text), len, char);	2282	Move (str, SvEND (self->incr_text), len, char);
2167	SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);	2283	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	2284	*SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2169	}	2285	}
…		…
2242	self->incr_mode = 0;	2358	self->incr_mode = 0;
2243	}	2359	}
2244		2360
2245	void DESTROY (JSON *self)	2361	void DESTROY (JSON *self)
2246	CODE:	2362	CODE:
		2363	SvREFCNT_dec (self->v_false);
		2364	SvREFCNT_dec (self->v_true);
2247	SvREFCNT_dec (self->cb_sk_object);	2365	SvREFCNT_dec (self->cb_sk_object);
2248	SvREFCNT_dec (self->cb_object);	2366	SvREFCNT_dec (self->cb_object);
2249	SvREFCNT_dec (self->incr_text);	2367	SvREFCNT_dec (self->incr_text);
2250		2368
2251	PROTOTYPES: ENABLE	2369	PROTOTYPES: ENABLE

Diff Legend

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

Comparing JSON-XS/XS.xs (file contents): Revision 1.129 by root, Wed Nov 16 18:06:34 2016 UTC vs. Revision 1.143 by root, Mon Jun 21 01:04:30 2021 UTC

Diff Legend

Comparing JSON-XS/XS.xs (file contents):
Revision 1.129 by root, Wed Nov 16 18:06:34 2016 UTC vs.
Revision 1.143 by root, Mon Jun 21 01:04:30 2021 UTC