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 1.54 by root, Tue Jul 10 16:22:31 2007 UTC vs.
Revision 1.126 by root, Sun Feb 21 16:18:19 2016 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include <assert.h>
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>
10#include <float.h>
9 11
10#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 14#endif
13 15
14// some old perls do not have this, try to make it work, no 16// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces. 17// guarantees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
18#endif 20#endif
21
22// compatibility with perl <5.18
23#ifndef HvNAMELEN_get
24# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
25#endif
26#ifndef HvNAMELEN
27# define HvNAMELEN(hv) HvNAMELEN_get (hv)
28#endif
29#ifndef HvNAMEUTF8
30# define HvNAMEUTF8(hv) 0
31#endif
32
33// three extra for rounding, sign, and end of string
34#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
19 35
20#define F_ASCII 0x00000001UL 36#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL 37#define F_LATIN1 0x00000002UL
22#define F_UTF8 0x00000004UL 38#define F_UTF8 0x00000004UL
23#define F_INDENT 0x00000008UL 39#define F_INDENT 0x00000008UL
… …
26#define F_SPACE_AFTER 0x00000040UL 42#define F_SPACE_AFTER 0x00000040UL
27#define F_ALLOW_NONREF 0x00000100UL 43#define F_ALLOW_NONREF 0x00000100UL
28#define F_SHRINK 0x00000200UL 44#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL 45#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL 46#define F_CONV_BLESSED 0x00000800UL
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL 47#define F_RELAXED 0x00001000UL
34#define S_MAXSIZE 20 48#define F_ALLOW_UNKNOWN 0x00002000UL
49#define F_ALLOW_TAGS 0x00004000UL
35#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 50#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
36 51
37#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
38#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
39
40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 52#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
41#define F_DEFAULT (9UL << S_MAXDEPTH)
42 53
43#define INIT_SIZE 32 // initial scalar size to be allocated 54#define INIT_SIZE 32 // initial scalar size to be allocated
44#define INDENT_STEP 3 // spaces per indentation level 55#define INDENT_STEP 3 // spaces per indentation level
45 56
46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 57#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
47 58
59#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
60
48#define SB do { 61#define SB do {
49#define SE } while (0) 62#define SE } while (0)
50 63
51#if __GNUC__ >= 3 64#if __GNUC__ >= 3
52# define expect(expr,value) __builtin_expect ((expr),(value)) 65# define expect(expr,value) __builtin_expect ((expr), (value))
53# define inline inline 66# define INLINE static inline
54#else 67#else
55# define expect(expr,value) (expr) 68# define expect(expr,value) (expr)
56# define inline static 69# define INLINE static
57#endif 70#endif
58 71
59#define expect_false(expr) expect ((expr) != 0, 0) 72#define expect_false(expr) expect ((expr) != 0, 0)
60#define expect_true(expr) expect ((expr) != 0, 1) 73#define expect_true(expr) expect ((expr) != 0, 1)
61 74
62static HV *json_stash, *json_boolean_stash; // JSON::XS:: 75#define IN_RANGE_INC(type,val,beg,end) \
63static SV *json_true, *json_false; 76 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
77 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
78
79#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
80
81#ifdef USE_ITHREADS
82# define JSON_SLOW 1
83# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
84# define BOOL_STASH (bool_stash ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1))
85#else
86# define JSON_SLOW 0
87# define JSON_STASH json_stash
88# define BOOL_STASH bool_stash
89#endif
90
91// the amount of HEs to allocate on the stack, when sorting keys
92#define STACK_HES 64
93
94static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean::
95static SV *bool_true, *bool_false, *sv_json;
96
97enum {
98 INCR_M_WS = 0, // initial whitespace skipping, must be 0
99 INCR_M_STR, // inside string
100 INCR_M_BS, // inside backslash
101 INCR_M_C0, // inside comment in initial whitespace sequence
102 INCR_M_C1, // inside comment in other places
103 INCR_M_JSON // outside anything, count nesting
104};
105
106#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
64 107
65typedef struct { 108typedef struct {
66 U32 flags; 109 U32 flags;
110 U32 max_depth;
111 STRLEN max_size;
112
67 SV *cb_object; 113 SV *cb_object;
68 HV *cb_sk_object; 114 HV *cb_sk_object;
115
116 // for the incremental parser
117 SV *incr_text; // the source text so far
118 STRLEN incr_pos; // the current offset into the text
119 int incr_nest; // {[]}-nesting level
120 unsigned char incr_mode;
69} JSON; 121} JSON;
122
123INLINE void
124json_init (JSON *json)
125{
126 Zero (json, 1, JSON);
127 json->max_depth = 512;
128}
70 129
71///////////////////////////////////////////////////////////////////////////// 130/////////////////////////////////////////////////////////////////////////////
72// utility functions 131// utility functions
73 132
74inline void 133INLINE SV *
134get_bool (const char *name)
135{
136 SV *sv = get_sv (name, 1);
137
138 SvREADONLY_on (sv);
139 SvREADONLY_on (SvRV (sv));
140
141 return sv;
142}
143
144INLINE void
75shrink (SV *sv) 145shrink (SV *sv)
76{ 146{
77 sv_utf8_downgrade (sv, 1); 147 sv_utf8_downgrade (sv, 1);
148
78 if (SvLEN (sv) > SvCUR (sv) + 1) 149 if (SvLEN (sv) > SvCUR (sv) + 1)
79 { 150 {
80#ifdef SvPV_shrink_to_cur 151#ifdef SvPV_shrink_to_cur
81 SvPV_shrink_to_cur (sv); 152 SvPV_shrink_to_cur (sv);
82#elif defined (SvPV_renew) 153#elif defined (SvPV_renew)
… …
88// decode an utf-8 character and return it, or (UV)-1 in 159// decode an utf-8 character and return it, or (UV)-1 in
89// case of an error. 160// case of an error.
90// we special-case "safe" characters from U+80 .. U+7FF, 161// we special-case "safe" characters from U+80 .. U+7FF,
91// but use the very good perl function to parse anything else. 162// but use the very good perl function to parse anything else.
92// note that we never call this function for a ascii codepoints 163// note that we never call this function for a ascii codepoints
93inline UV 164INLINE UV
94decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 165decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
95{ 166{
96 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 167 if (expect_true (len >= 2
97 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 168 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
98 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 169 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
99 { 170 {
100 *clen = 2; 171 *clen = 2;
101 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 172 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
102 } 173 }
103 else 174 else
104 { 175 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
105 *clen = (STRLEN)-1; 176}
106 return (UV)-1; 177
178// likewise for encoding, also never called for ascii codepoints
179// this function takes advantage of this fact, although current gccs
180// seem to optimise the check for >= 0x80 away anyways
181INLINE unsigned char *
182encode_utf8 (unsigned char *s, UV ch)
183{
184 if (expect_false (ch < 0x000080))
185 *s++ = ch;
186 else if (expect_true (ch < 0x000800))
187 *s++ = 0xc0 | ( ch >> 6),
188 *s++ = 0x80 | ( ch & 0x3f);
189 else if ( ch < 0x010000)
190 *s++ = 0xe0 | ( ch >> 12),
191 *s++ = 0x80 | ((ch >> 6) & 0x3f),
192 *s++ = 0x80 | ( ch & 0x3f);
193 else if ( ch < 0x110000)
194 *s++ = 0xf0 | ( ch >> 18),
195 *s++ = 0x80 | ((ch >> 12) & 0x3f),
196 *s++ = 0x80 | ((ch >> 6) & 0x3f),
197 *s++ = 0x80 | ( ch & 0x3f);
198
199 return s;
200}
201
202// convert offset pointer to character index, sv must be string
203static STRLEN
204ptr_to_index (SV *sv, char *offset)
205{
206 return SvUTF8 (sv)
207 ? utf8_distance (offset, SvPVX (sv))
208 : offset - SvPVX (sv);
209}
210
211/////////////////////////////////////////////////////////////////////////////
212// fp hell
213
214// scan a group of digits, and a trailing exponent
215static void
216json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
217{
218 UV uaccum = 0;
219 int eaccum = 0;
220
221 // if we recurse too deep, skip all remaining digits
222 // to avoid a stack overflow attack
223 if (expect_false (--maxdepth <= 0))
224 while (((U8)*s - '0') < 10)
225 ++s;
226
227 for (;;)
107 } 228 {
229 U8 dig = (U8)*s - '0';
230
231 if (expect_false (dig >= 10))
232 {
233 if (dig == (U8)((U8)'.' - (U8)'0'))
234 {
235 ++s;
236 json_atof_scan1 (s, accum, expo, 1, maxdepth);
237 }
238 else if ((dig | ' ') == 'e' - '0')
239 {
240 int exp2 = 0;
241 int neg = 0;
242
243 ++s;
244
245 if (*s == '-')
246 {
247 ++s;
248 neg = 1;
249 }
250 else if (*s == '+')
251 ++s;
252
253 while ((dig = (U8)*s - '0') < 10)
254 exp2 = exp2 * 10 + *s++ - '0';
255
256 *expo += neg ? -exp2 : exp2;
257 }
258
259 break;
260 }
261
262 ++s;
263
264 uaccum = uaccum * 10 + dig;
265 ++eaccum;
266
267 // if we have too many digits, then recurse for more
268 // we actually do this for rather few digits
269 if (uaccum >= (UV_MAX - 9) / 10)
270 {
271 if (postdp) *expo -= eaccum;
272 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
273 if (postdp) *expo += eaccum;
274
275 break;
276 }
277 }
278
279 // this relies greatly on the quality of the pow ()
280 // implementation of the platform, but a good
281 // implementation is hard to beat.
282 // (IEEE 754 conformant ones are required to be exact)
283 if (postdp) *expo -= eaccum;
284 *accum += uaccum * Perl_pow (10., *expo);
285 *expo += eaccum;
286}
287
288static NV
289json_atof (const char *s)
290{
291 NV accum = 0.;
292 int expo = 0;
293 int neg = 0;
294
295 if (*s == '-')
296 {
297 ++s;
298 neg = 1;
299 }
300
301 // a recursion depth of ten gives us >>500 bits
302 json_atof_scan1 (s, &accum, &expo, 0, 10);
303
304 return neg ? -accum : accum;
305}
306
307// target of scalar reference is bool? -1 == nope, 0 == false, 1 == true
308static int
309ref_bool_type (SV *sv)
310{
311 svtype svt = SvTYPE (sv);
312
313 if (svt < SVt_PVAV)
314 {
315 STRLEN len = 0;
316 char *pv = svt ? SvPV (sv, len) : 0;
317
318 if (len == 1)
319 if (*pv == '1')
320 return 1;
321 else if (*pv == '0')
322 return 0;
323
324 }
325
326 return -1;
327}
328
329// returns whether scalar is not a reference in the sense of allow_nonref
330static int
331json_nonref (SV *scalar)
332{
333 if (!SvROK (scalar))
334 return 1;
335
336 scalar = SvRV (scalar);
337
338 if (SvSTASH (scalar) == bool_stash)
339 return 1;
340
341 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
342 return 1;
343
344 return 0;
108} 345}
109 346
110///////////////////////////////////////////////////////////////////////////// 347/////////////////////////////////////////////////////////////////////////////
111// encoder 348// encoder
112 349
… …
116 char *cur; // SvPVX (sv) + current output position 353 char *cur; // SvPVX (sv) + current output position
117 char *end; // SvEND (sv) 354 char *end; // SvEND (sv)
118 SV *sv; // result scalar 355 SV *sv; // result scalar
119 JSON json; 356 JSON json;
120 U32 indent; // indentation level 357 U32 indent; // indentation level
121 U32 maxdepth; // max. indentation/recursion level 358 UV limit; // escape character values >= this value when encoding
122} enc_t; 359} enc_t;
123 360
124inline void 361INLINE void
125need (enc_t *enc, STRLEN len) 362need (enc_t *enc, STRLEN len)
126{ 363{
127 if (expect_false (enc->cur + len >= enc->end)) 364 if (expect_false (enc->cur + len >= enc->end))
128 { 365 {
129 STRLEN cur = enc->cur - SvPVX (enc->sv); 366 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
130 SvGROW (enc->sv, cur + len + 1); 367 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
131 enc->cur = SvPVX (enc->sv) + cur; 368 enc->cur = SvPVX (enc->sv) + cur;
132 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 369 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
133 } 370 }
134} 371}
135 372
136inline void 373INLINE void
137encode_ch (enc_t *enc, char ch) 374encode_ch (enc_t *enc, char ch)
138{ 375{
139 need (enc, 1); 376 need (enc, 1);
140 *enc->cur++ = ch; 377 *enc->cur++ = ch;
141} 378}
… …
195 { 432 {
196 uch = ch; 433 uch = ch;
197 clen = 1; 434 clen = 1;
198 } 435 }
199 436
200 if (uch > 0x10FFFFUL) 437 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
201 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
202
203 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
204 { 438 {
205 if (uch > 0xFFFFUL) 439 if (uch >= 0x10000UL)
206 { 440 {
441 if (uch >= 0x110000UL)
442 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
443
207 need (enc, len += 11); 444 need (enc, len += 11);
208 sprintf (enc->cur, "\\u%04x\\u%04x", 445 sprintf (enc->cur, "\\u%04x\\u%04x",
209 (int)((uch - 0x10000) / 0x400 + 0xD800), 446 (int)((uch - 0x10000) / 0x400 + 0xD800),
210 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 447 (int)((uch - 0x10000) % 0x400 + 0xDC00));
211 enc->cur += 12; 448 enc->cur += 12;
212 } 449 }
213 else 450 else
214 { 451 {
215 static char hexdigit [16] = "0123456789abcdef";
216 need (enc, len += 5); 452 need (enc, len += 5);
217 *enc->cur++ = '\\'; 453 *enc->cur++ = '\\';
218 *enc->cur++ = 'u'; 454 *enc->cur++ = 'u';
219 *enc->cur++ = hexdigit [ uch >> 12 ]; 455 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
220 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 456 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
221 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 457 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
222 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 458 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
223 } 459 }
224 460
225 str += clen; 461 str += clen;
226 } 462 }
227 else if (enc->json.flags & F_LATIN1) 463 else if (enc->json.flags & F_LATIN1)
… …
239 while (--clen); 475 while (--clen);
240 } 476 }
241 else 477 else
242 { 478 {
243 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 479 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
244 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 480 enc->cur = encode_utf8 (enc->cur, uch);
245 ++str; 481 ++str;
246 } 482 }
247 } 483 }
248 } 484 }
249 } 485 }
250 486
251 --len; 487 --len;
252 } 488 }
253} 489}
254 490
255inline void 491INLINE void
256encode_indent (enc_t *enc) 492encode_indent (enc_t *enc)
257{ 493{
258 if (enc->json.flags & F_INDENT) 494 if (enc->json.flags & F_INDENT)
259 { 495 {
260 int spaces = enc->indent * INDENT_STEP; 496 int spaces = enc->indent * INDENT_STEP;
… …
263 memset (enc->cur, ' ', spaces); 499 memset (enc->cur, ' ', spaces);
264 enc->cur += spaces; 500 enc->cur += spaces;
265 } 501 }
266} 502}
267 503
268inline void 504INLINE void
269encode_space (enc_t *enc) 505encode_space (enc_t *enc)
270{ 506{
271 need (enc, 1); 507 need (enc, 1);
272 encode_ch (enc, ' '); 508 encode_ch (enc, ' ');
273} 509}
274 510
275inline void 511INLINE void
276encode_nl (enc_t *enc) 512encode_nl (enc_t *enc)
277{ 513{
278 if (enc->json.flags & F_INDENT) 514 if (enc->json.flags & F_INDENT)
279 { 515 {
280 need (enc, 1); 516 need (enc, 1);
281 encode_ch (enc, '\n'); 517 encode_ch (enc, '\n');
282 } 518 }
283} 519}
284 520
285inline void 521INLINE void
286encode_comma (enc_t *enc) 522encode_comma (enc_t *enc)
287{ 523{
288 encode_ch (enc, ','); 524 encode_ch (enc, ',');
289 525
290 if (enc->json.flags & F_INDENT) 526 if (enc->json.flags & F_INDENT)
… …
298static void 534static void
299encode_av (enc_t *enc, AV *av) 535encode_av (enc_t *enc, AV *av)
300{ 536{
301 int i, len = av_len (av); 537 int i, len = av_len (av);
302 538
303 if (enc->indent >= enc->maxdepth) 539 if (enc->indent >= enc->json.max_depth)
304 croak ("data structure too deep (hit recursion limit)"); 540 croak (ERR_NESTING_EXCEEDED);
305 541
306 encode_ch (enc, '['); encode_nl (enc); 542 encode_ch (enc, '[');
307 ++enc->indent;
308 543
544 if (len >= 0)
545 {
546 encode_nl (enc); ++enc->indent;
547
309 for (i = 0; i <= len; ++i) 548 for (i = 0; i <= len; ++i)
310 { 549 {
550 SV **svp = av_fetch (av, i, 0);
551
311 encode_indent (enc); 552 encode_indent (enc);
312 encode_sv (enc, *av_fetch (av, i, 0));
313 553
554 if (svp)
555 encode_sv (enc, *svp);
556 else
557 encode_str (enc, "null", 4, 0);
558
314 if (i < len) 559 if (i < len)
315 encode_comma (enc); 560 encode_comma (enc);
316 } 561 }
317 562
563 encode_nl (enc); --enc->indent; encode_indent (enc);
564 }
565
318 encode_nl (enc); 566 encode_ch (enc, ']');
319
320 --enc->indent;
321 encode_indent (enc); encode_ch (enc, ']');
322} 567}
323 568
324static void 569static void
325encode_he (enc_t *enc, HE *he) 570encode_hk (enc_t *enc, HE *he)
326{ 571{
327 encode_ch (enc, '"'); 572 encode_ch (enc, '"');
328 573
329 if (HeKLEN (he) == HEf_SVKEY) 574 if (HeKLEN (he) == HEf_SVKEY)
330 { 575 {
331 SV *sv = HeSVKEY (he); 576 SV *sv = HeSVKEY (he);
332 STRLEN len; 577 STRLEN len;
333 char *str; 578 char *str;
334 579
335 SvGETMAGIC (sv); 580 SvGETMAGIC (sv);
336 str = SvPV (sv, len); 581 str = SvPV (sv, len);
337 582
338 encode_str (enc, str, len, SvUTF8 (sv)); 583 encode_str (enc, str, len, SvUTF8 (sv));
339 } 584 }
… …
343 encode_ch (enc, '"'); 588 encode_ch (enc, '"');
344 589
345 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); 590 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
346 encode_ch (enc, ':'); 591 encode_ch (enc, ':');
347 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); 592 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
348 encode_sv (enc, HeVAL (he));
349} 593}
350 594
351// compare hash entries, used when all keys are bytestrings 595// compare hash entries, used when all keys are bytestrings
352static int 596static int
353he_cmp_fast (const void *a_, const void *b_) 597he_cmp_fast (const void *a_, const void *b_)
… …
358 HE *b = *(HE **)b_; 602 HE *b = *(HE **)b_;
359 603
360 STRLEN la = HeKLEN (a); 604 STRLEN la = HeKLEN (a);
361 STRLEN lb = HeKLEN (b); 605 STRLEN lb = HeKLEN (b);
362 606
363 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 607 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
364 cmp = la - lb; 608 cmp = lb - la;
365 609
366 return cmp; 610 return cmp;
367} 611}
368 612
369// compare hash entries, used when some keys are sv's or utf-x 613// compare hash entries, used when some keys are sv's or utf-x
370static int 614static int
371he_cmp_slow (const void *a, const void *b) 615he_cmp_slow (const void *a, const void *b)
372{ 616{
373 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 617 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
374} 618}
375 619
376static void 620static void
377encode_hv (enc_t *enc, HV *hv) 621encode_hv (enc_t *enc, HV *hv)
378{ 622{
379 int count, i; 623 HE *he;
380 624
381 if (enc->indent >= enc->maxdepth) 625 if (enc->indent >= enc->json.max_depth)
382 croak ("data structure too deep (hit recursion limit)"); 626 croak (ERR_NESTING_EXCEEDED);
383 627
384 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 628 encode_ch (enc, '{');
385 629
386 if ((count = hv_iterinit (hv)))
387 {
388 // for canonical output we have to sort by keys first 630 // for canonical output we have to sort by keys first
389 // actually, this is mostly due to the stupid so-called 631 // actually, this is mostly due to the stupid so-called
390 // security workaround added somewhere in 5.8.x. 632 // security workaround added somewhere in 5.8.x
391 // that randomises hash orderings 633 // that randomises hash orderings
392 if (enc->json.flags & F_CANONICAL) 634 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
635 {
636 int count = hv_iterinit (hv);
637
638 if (SvMAGICAL (hv))
393 { 639 {
640 // need to count by iterating. could improve by dynamically building the vector below
641 // but I don't care for the speed of this special case.
642 // note also that we will run into undefined behaviour when the two iterations
643 // do not result in the same count, something I might care for in some later release.
644
645 count = 0;
646 while (hv_iternext (hv))
647 ++count;
648
649 hv_iterinit (hv);
650 }
651
652 if (count)
653 {
394 int fast = 1; 654 int i, fast = 1;
395 HE *he; 655 HE *hes_stack [STACK_HES];
396#if defined(__BORLANDC__) || defined(_MSC_VER) 656 HE **hes = hes_stack;
397 HE **hes = _alloca (count * sizeof (HE)); 657
398#else 658 // allocate larger arrays on the heap
399 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 659 if (count > STACK_HES)
400#endif 660 {
661 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
662 hes = (HE **)SvPVX (sv);
663 }
401 664
402 i = 0; 665 i = 0;
403 while ((he = hv_iternext (hv))) 666 while ((he = hv_iternext (hv)))
404 { 667 {
405 hes [i++] = he; 668 hes [i++] = he;
… …
427 690
428 FREETMPS; 691 FREETMPS;
429 LEAVE; 692 LEAVE;
430 } 693 }
431 694
432 for (i = 0; i < count; ++i) 695 encode_nl (enc); ++enc->indent;
696
697 while (count--)
433 { 698 {
434 encode_indent (enc); 699 encode_indent (enc);
700 he = hes [count];
435 encode_he (enc, hes [i]); 701 encode_hk (enc, he);
702 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
436 703
437 if (i < count - 1) 704 if (count)
438 encode_comma (enc); 705 encode_comma (enc);
439 } 706 }
440 707
441 encode_nl (enc); 708 encode_nl (enc); --enc->indent; encode_indent (enc);
442 } 709 }
710 }
443 else 711 else
712 {
713 if (hv_iterinit (hv) || SvMAGICAL (hv))
714 if ((he = hv_iternext (hv)))
444 { 715 {
445 HE *he = hv_iternext (hv); 716 encode_nl (enc); ++enc->indent;
446 717
447 for (;;) 718 for (;;)
448 { 719 {
449 encode_indent (enc); 720 encode_indent (enc);
450 encode_he (enc, he); 721 encode_hk (enc, he);
722 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
451 723
452 if (!(he = hv_iternext (hv))) 724 if (!(he = hv_iternext (hv)))
453 break; 725 break;
454 726
455 encode_comma (enc); 727 encode_comma (enc);
456 } 728 }
457 729
458 encode_nl (enc); 730 encode_nl (enc); --enc->indent; encode_indent (enc);
459 } 731 }
460 } 732 }
461 733
462 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 734 encode_ch (enc, '}');
463} 735}
464 736
465// encode objects, arrays and special \0=false and \1=true values. 737// encode objects, arrays and special \0=false and \1=true values.
466static void 738static void
467encode_rv (enc_t *enc, SV *sv) 739encode_rv (enc_t *enc, SV *sv)
468{ 740{
469 svtype svt; 741 svtype svt;
742 GV *method;
470 743
471 SvGETMAGIC (sv); 744 SvGETMAGIC (sv);
472 svt = SvTYPE (sv); 745 svt = SvTYPE (sv);
473 746
474 if (expect_false (SvOBJECT (sv))) 747 if (expect_false (SvOBJECT (sv)))
475 { 748 {
476 if (SvSTASH (sv) == json_boolean_stash) 749 HV *stash = SvSTASH (sv);
750
751 if (stash == bool_stash)
477 { 752 {
478 if (SvIV (sv)) 753 if (SvIV (sv))
479 encode_str (enc, "true", 4, 0); 754 encode_str (enc, "true", 4, 0);
480 else 755 else
481 encode_str (enc, "false", 5, 0); 756 encode_str (enc, "false", 5, 0);
482 } 757 }
758 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
759 {
760 int count;
761 dSP;
762
763 ENTER; SAVETMPS;
764 SAVESTACK_POS ();
765 PUSHMARK (SP);
766 EXTEND (SP, 2);
767 // we re-bless the reference to get overload and other niceties right
768 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
769 PUSHs (sv_json);
770
771 PUTBACK;
772 count = call_sv ((SV *)GvCV (method), G_ARRAY);
773 SPAGAIN;
774
775 // catch this surprisingly common error
776 if (SvROK (TOPs) && SvRV (TOPs) == sv)
777 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
778
779 encode_ch (enc, '(');
780 encode_ch (enc, '"');
781 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
782 encode_ch (enc, '"');
783 encode_ch (enc, ')');
784 encode_ch (enc, '[');
785
786 while (count)
787 {
788 encode_sv (enc, SP[1 - count--]);
789
790 if (count)
791 encode_ch (enc, ',');
792 }
793
794 encode_ch (enc, ']');
795
796 FREETMPS; LEAVE;
797 }
798 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
799 {
800 dSP;
801
802 ENTER; SAVETMPS;
803 PUSHMARK (SP);
804 // we re-bless the reference to get overload and other niceties right
805 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
806
807 // calling with G_SCALAR ensures that we always get a 1 return value
808 PUTBACK;
809 call_sv ((SV *)GvCV (method), G_SCALAR);
810 SPAGAIN;
811
812 // catch this surprisingly common error
813 if (SvROK (TOPs) && SvRV (TOPs) == sv)
814 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
815
816 sv = POPs;
817 PUTBACK;
818
819 encode_sv (enc, sv);
820
821 FREETMPS; LEAVE;
822 }
823 else if (enc->json.flags & F_ALLOW_BLESSED)
824 encode_str (enc, "null", 4, 0);
483 else 825 else
484 {
485#if 0
486 if (0 && sv_derived_from (rv, "JSON::Literal"))
487 {
488 // not yet
489 }
490#endif
491 if (enc->json.flags & F_CONV_BLESSED)
492 {
493 // we re-bless the reference to get overload and other niceties right
494 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
495
496 if (to_json)
497 {
498 int count;
499 dSP;
500
501 ENTER; SAVETMPS; PUSHMARK (SP);
502 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
503
504 // calling with G_SCALAR ensures that we always get a 1 return value
505 PUTBACK;
506 call_sv ((SV *)GvCV (to_json), G_SCALAR);
507 SPAGAIN;
508
509 // catch this surprisingly common error
510 if (SvROK (TOPs) && SvRV (TOPs) == sv)
511 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
512
513 sv = POPs;
514 PUTBACK;
515
516 encode_sv (enc, sv);
517
518 FREETMPS; LEAVE;
519 }
520 else if (enc->json.flags & F_ALLOW_BLESSED)
521 encode_str (enc, "null", 4, 0);
522 else
523 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
524 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
525 }
526 else if (enc->json.flags & F_ALLOW_BLESSED)
527 encode_str (enc, "null", 4, 0);
528 else
529 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 826 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
530 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 827 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
531 }
532 } 828 }
533 else if (svt == SVt_PVHV) 829 else if (svt == SVt_PVHV)
534 encode_hv (enc, (HV *)sv); 830 encode_hv (enc, (HV *)sv);
535 else if (svt == SVt_PVAV) 831 else if (svt == SVt_PVAV)
536 encode_av (enc, (AV *)sv); 832 encode_av (enc, (AV *)sv);
537 else if (svt < SVt_PVAV) 833 else if (svt < SVt_PVAV)
538 { 834 {
539 STRLEN len = 0; 835 int bool_type = ref_bool_type (sv);
540 char *pv = svt ? SvPV (sv, len) : 0;
541 836
542 if (len == 1 && *pv == '1') 837 if (bool_type == 1)
543 encode_str (enc, "true", 4, 0); 838 encode_str (enc, "true", 4, 0);
544 else if (len == 1 && *pv == '0') 839 else if (bool_type == 0)
545 encode_str (enc, "false", 5, 0); 840 encode_str (enc, "false", 5, 0);
841 else if (enc->json.flags & F_ALLOW_UNKNOWN)
842 encode_str (enc, "null", 4, 0);
546 else 843 else
547 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 844 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
548 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 845 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
549 } 846 }
847 else if (enc->json.flags & F_ALLOW_UNKNOWN)
848 encode_str (enc, "null", 4, 0);
550 else 849 else
551 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 850 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
552 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 851 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
553} 852}
554 853
… …
572 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 871 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
573 enc->cur += strlen (enc->cur); 872 enc->cur += strlen (enc->cur);
574 } 873 }
575 else if (SvIOKp (sv)) 874 else if (SvIOKp (sv))
576 { 875 {
577 // we assume we can always read an IV as a UV 876 // we assume we can always read an IV as a UV and vice versa
578 if (SvUV (sv) & ~(UV)0x7fff) 877 // we assume two's complement
579 { 878 // we assume no aliasing issues in the union
580 // large integer, use the (rather slow) snprintf way. 879 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
581 need (enc, sizeof (UV) * 3); 880 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
582 enc->cur +=
583 SvIsUV(sv)
584 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
585 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
586 }
587 else
588 { 881 {
589 // optimise the "small number case" 882 // optimise the "small number case"
590 // code will likely be branchless and use only a single multiplication 883 // code will likely be branchless and use only a single multiplication
884 // works for numbers up to 59074
591 I32 i = SvIV (sv); 885 I32 i = SvIVX (sv);
592 U32 u; 886 U32 u;
593 char digit, nz = 0; 887 char digit, nz = 0;
594 888
595 need (enc, 6); 889 need (enc, 6);
596 890
… …
602 896
603 // now output digit by digit, each time masking out the integer part 897 // now output digit by digit, each time masking out the integer part
604 // and multiplying by 5 while moving the decimal point one to the right, 898 // and multiplying by 5 while moving the decimal point one to the right,
605 // resulting in a net multiplication by 10. 899 // resulting in a net multiplication by 10.
606 // we always write the digit to memory but conditionally increment 900 // we always write the digit to memory but conditionally increment
607 // the pointer, to ease the usage of conditional move instructions. 901 // the pointer, to enable the use of conditional move instructions.
608 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 902 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
609 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 903 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
610 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 904 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
611 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 905 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
612 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 906 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
613 } 907 }
908 else
909 {
910 // large integer, use the (rather slow) snprintf way.
911 need (enc, IVUV_MAXCHARS);
912 enc->cur +=
913 SvIsUV(sv)
914 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
915 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
916 }
614 } 917 }
615 else if (SvROK (sv)) 918 else if (SvROK (sv))
616 encode_rv (enc, SvRV (sv)); 919 encode_rv (enc, SvRV (sv));
617 else if (!SvOK (sv)) 920 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
618 encode_str (enc, "null", 4, 0); 921 encode_str (enc, "null", 4, 0);
619 else 922 else
620 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 923 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
621 SvPV_nolen (sv), SvFLAGS (sv)); 924 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
622} 925}
623 926
624static SV * 927static SV *
625encode_json (SV *scalar, JSON *json) 928encode_json (SV *scalar, JSON *json)
626{ 929{
627 enc_t enc; 930 enc_t enc;
628 931
629 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 932 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
630 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 933 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
631 934
632 enc.json = *json; 935 enc.json = *json;
633 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 936 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
634 enc.cur = SvPVX (enc.sv); 937 enc.cur = SvPVX (enc.sv);
635 enc.end = SvEND (enc.sv); 938 enc.end = SvEND (enc.sv);
636 enc.indent = 0; 939 enc.indent = 0;
637 enc.maxdepth = DEC_DEPTH (enc.json.flags); 940 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
941 : enc.json.flags & F_LATIN1 ? 0x000100UL
942 : 0x110000UL;
638 943
639 SvPOK_only (enc.sv); 944 SvPOK_only (enc.sv);
640 encode_sv (&enc, scalar); 945 encode_sv (&enc, scalar);
946 encode_nl (&enc);
641 947
642 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 948 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
643 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 949 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
644 950
645 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 951 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
663 JSON json; 969 JSON json;
664 U32 depth; // recursion depth 970 U32 depth; // recursion depth
665 U32 maxdepth; // recursion depth limit 971 U32 maxdepth; // recursion depth limit
666} dec_t; 972} dec_t;
667 973
668inline void 974INLINE void
975decode_comment (dec_t *dec)
976{
977 // only '#'-style comments allowed a.t.m.
978
979 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
980 ++dec->cur;
981}
982
983INLINE void
669decode_ws (dec_t *dec) 984decode_ws (dec_t *dec)
670{ 985{
671 for (;;) 986 for (;;)
672 { 987 {
673 char ch = *dec->cur; 988 char ch = *dec->cur;
674 989
675 if (ch > 0x20 990 if (ch > 0x20)
991 {
992 if (expect_false (ch == '#'))
993 {
994 if (dec->json.flags & F_RELAXED)
995 decode_comment (dec);
996 else
997 break;
998 }
999 else
1000 break;
1001 }
676 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 1002 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
677 break; 1003 break; // parse error, but let higher level handle it, gives better error messages
678 1004
679 ++dec->cur; 1005 ++dec->cur;
680 } 1006 }
681} 1007}
682 1008
… …
686 if (*dec->cur != ch) \ 1012 if (*dec->cur != ch) \
687 ERR (# ch " expected"); \ 1013 ERR (# ch " expected"); \
688 ++dec->cur; \ 1014 ++dec->cur; \
689 SE 1015 SE
690 1016
691#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 1017#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
692#define DEC_DEC_DEPTH --dec->depth 1018#define DEC_DEC_DEPTH --dec->depth
693 1019
694static SV *decode_sv (dec_t *dec); 1020static SV *decode_sv (dec_t *dec);
695 1021
696static signed char decode_hexdigit[256]; 1022static signed char decode_hexdigit[256];
… …
788 1114
789 if (hi >= 0x80) 1115 if (hi >= 0x80)
790 { 1116 {
791 utf8 = 1; 1117 utf8 = 1;
792 1118
793 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1119 cur = encode_utf8 (cur, hi);
794 } 1120 }
795 else 1121 else
796 *cur++ = hi; 1122 *cur++ = hi;
797 } 1123 }
798 break; 1124 break;
… …
800 default: 1126 default:
801 --dec_cur; 1127 --dec_cur;
802 ERR ("illegal backslash escape sequence in string"); 1128 ERR ("illegal backslash escape sequence in string");
803 } 1129 }
804 } 1130 }
805 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1131 else if (expect_true (ch >= 0x20 && ch < 0x80))
806 *cur++ = ch; 1132 *cur++ = ch;
807 else if (ch >= 0x80) 1133 else if (ch >= 0x80)
808 { 1134 {
809 STRLEN clen; 1135 STRLEN clen;
810 UV uch;
811 1136
812 --dec_cur; 1137 --dec_cur;
813 1138
814 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1139 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
815 if (clen == (STRLEN)-1) 1140 if (clen == (STRLEN)-1)
816 ERR ("malformed UTF-8 character in JSON string"); 1141 ERR ("malformed UTF-8 character in JSON string");
817 1142
818 do 1143 do
819 *cur++ = *dec_cur++; 1144 *cur++ = *dec_cur++;
820 while (--clen); 1145 while (--clen);
821 1146
822 utf8 = 1; 1147 utf8 = 1;
823 } 1148 }
1149 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1150 *cur++ = ch;
824 else 1151 else
825 { 1152 {
826 --dec_cur; 1153 --dec_cur;
827 1154
828 if (!ch) 1155 if (!ch)
… …
836 { 1163 {
837 STRLEN len = cur - buf; 1164 STRLEN len = cur - buf;
838 1165
839 if (sv) 1166 if (sv)
840 { 1167 {
841 SvGROW (sv, SvCUR (sv) + len + 1); 1168 STRLEN cur = SvCUR (sv);
1169
1170 if (SvLEN (sv) <= cur + len)
1171 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1172
842 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1173 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
843 SvCUR_set (sv, SvCUR (sv) + len); 1174 SvCUR_set (sv, SvCUR (sv) + len);
844 } 1175 }
845 else 1176 else
846 sv = newSVpvn (buf, len); 1177 sv = newSVpvn (buf, len);
… …
931 is_nv = 1; 1262 is_nv = 1;
932 } 1263 }
933 1264
934 if (!is_nv) 1265 if (!is_nv)
935 { 1266 {
1267 int len = dec->cur - start;
1268
936 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1269 // special case the rather common 1..5-digit-int case
937 if (*start == '-') 1270 if (*start == '-')
938 switch (dec->cur - start) 1271 switch (len)
939 { 1272 {
940 case 2: return newSViv (-( start [1] - '0' * 1)); 1273 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
941 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1274 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
942 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1275 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
943 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1276 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1277 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
944 } 1278 }
945 else 1279 else
946 switch (dec->cur - start) 1280 switch (len)
947 { 1281 {
948 case 1: return newSViv ( start [0] - '0' * 1); 1282 case 1: return newSViv ( start [0] - '0' * 1);
949 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1283 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
950 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1284 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
951 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1285 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1286 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
952 } 1287 }
953 1288
954 { 1289 {
955 UV uv; 1290 UV uv;
956 int numtype = grok_number (start, dec->cur - start, &uv); 1291 int numtype = grok_number (start, len, &uv);
957 if (numtype & IS_NUMBER_IN_UV) 1292 if (numtype & IS_NUMBER_IN_UV)
958 if (numtype & IS_NUMBER_NEG) 1293 if (numtype & IS_NUMBER_NEG)
959 { 1294 {
960 if (uv < (UV)IV_MIN) 1295 if (uv < (UV)IV_MIN)
961 return newSViv (-(IV)uv); 1296 return newSViv (-(IV)uv);
962 } 1297 }
963 else 1298 else
964 return newSVuv (uv); 1299 return newSVuv (uv);
965
966 // here would likely be the place for bigint support
967 } 1300 }
968 }
969 1301
970 // if we ever support bigint or bigfloat, this is the place for bigfloat 1302 len -= *start == '-' ? 1 : 0;
1303
1304 // does not fit into IV or UV, try NV
1305 if (len <= NV_DIG)
1306 // fits into NV without loss of precision
1307 return newSVnv (json_atof (start));
1308
1309 // everything else fails, convert it to a string
1310 return newSVpvn (start, dec->cur - start);
1311 }
1312
1313 // loss of precision here
971 return newSVnv (Atof (start)); 1314 return newSVnv (json_atof (start));
972 1315
973fail: 1316fail:
974 return 0; 1317 return 0;
975} 1318}
976 1319
… …
1000 if (*dec->cur == ']') 1343 if (*dec->cur == ']')
1001 { 1344 {
1002 ++dec->cur; 1345 ++dec->cur;
1003 break; 1346 break;
1004 } 1347 }
1005 1348
1006 if (*dec->cur != ',') 1349 if (*dec->cur != ',')
1007 ERR (", or ] expected while parsing array"); 1350 ERR (", or ] expected while parsing array");
1008 1351
1009 ++dec->cur; 1352 ++dec->cur;
1353
1354 decode_ws (dec);
1355
1356 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1357 {
1358 ++dec->cur;
1359 break;
1360 }
1010 } 1361 }
1011 1362
1012 DEC_DEC_DEPTH; 1363 DEC_DEC_DEPTH;
1013 return newRV_noinc ((SV *)av); 1364 return newRV_noinc ((SV *)av);
1014 1365
… …
1030 if (*dec->cur == '}') 1381 if (*dec->cur == '}')
1031 ++dec->cur; 1382 ++dec->cur;
1032 else 1383 else
1033 for (;;) 1384 for (;;)
1034 { 1385 {
1035 decode_ws (dec); EXPECT_CH ('"'); 1386 EXPECT_CH ('"');
1036 1387
1037 // heuristic: assume that 1388 // heuristic: assume that
1038 // a) decode_str + hv_store_ent are abysmally slow. 1389 // a) decode_str + hv_store_ent are abysmally slow.
1039 // b) most hash keys are short, simple ascii text. 1390 // b) most hash keys are short, simple ascii text.
1040 // => try to "fast-match" such strings to avoid 1391 // => try to "fast-match" such strings to avoid
… …
1044 char *p = dec->cur; 1395 char *p = dec->cur;
1045 char *e = p + 24; // only try up to 24 bytes 1396 char *e = p + 24; // only try up to 24 bytes
1046 1397
1047 for (;;) 1398 for (;;)
1048 { 1399 {
1049 // the >= 0x80 is true on most architectures 1400 // the >= 0x80 is false on most architectures
1050 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1401 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1051 { 1402 {
1052 // slow path, back up and use decode_str 1403 // slow path, back up and use decode_str
1053 SV *key = decode_str (dec); 1404 SV *key = decode_str (dec);
1054 if (!key) 1405 if (!key)
1055 goto fail; 1406 goto fail;
1056 1407
1057 decode_ws (dec); EXPECT_CH (':'); 1408 decode_ws (dec); EXPECT_CH (':');
1058 1409
1410 decode_ws (dec);
1059 value = decode_sv (dec); 1411 value = decode_sv (dec);
1060 if (!value) 1412 if (!value)
1061 { 1413 {
1062 SvREFCNT_dec (key); 1414 SvREFCNT_dec (key);
1063 goto fail; 1415 goto fail;
… …
1075 int len = p - key; 1427 int len = p - key;
1076 dec->cur = p + 1; 1428 dec->cur = p + 1;
1077 1429
1078 decode_ws (dec); EXPECT_CH (':'); 1430 decode_ws (dec); EXPECT_CH (':');
1079 1431
1432 decode_ws (dec);
1080 value = decode_sv (dec); 1433 value = decode_sv (dec);
1081 if (!value) 1434 if (!value)
1082 goto fail; 1435 goto fail;
1083 1436
1084 hv_store (hv, key, len, value, 0); 1437 hv_store (hv, key, len, value, 0);
… …
1100 1453
1101 if (*dec->cur != ',') 1454 if (*dec->cur != ',')
1102 ERR (", or } expected while parsing object/hash"); 1455 ERR (", or } expected while parsing object/hash");
1103 1456
1104 ++dec->cur; 1457 ++dec->cur;
1458
1459 decode_ws (dec);
1460
1461 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1462 {
1463 ++dec->cur;
1464 break;
1465 }
1105 } 1466 }
1106 1467
1107 DEC_DEC_DEPTH; 1468 DEC_DEC_DEPTH;
1108 sv = newRV_noinc ((SV *)hv); 1469 sv = newRV_noinc ((SV *)hv);
1109 1470
… …
1116 1477
1117 hv_iterinit (hv); 1478 hv_iterinit (hv);
1118 he = hv_iternext (hv); 1479 he = hv_iternext (hv);
1119 hv_iterinit (hv); 1480 hv_iterinit (hv);
1120 1481
1121 // the next line creates a mortal sv each time its called. 1482 // the next line creates a mortal sv each time it's called.
1122 // might want to optimise this for common cases. 1483 // might want to optimise this for common cases.
1123 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); 1484 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1124 1485
1125 if (cb) 1486 if (cb)
1126 { 1487 {
1127 dSP; 1488 dSP;
1128 int count; 1489 int count;
1129 1490
1130 ENTER; SAVETMPS; PUSHMARK (SP); 1491 ENTER; SAVETMPS;
1492 SAVESTACK_POS ();
1493 PUSHMARK (SP);
1131 XPUSHs (HeVAL (he)); 1494 XPUSHs (HeVAL (he));
1495 sv_2mortal (sv);
1132 1496
1133 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1497 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1134 1498
1135 if (count == 1) 1499 if (count == 1)
1136 { 1500 {
1137 sv = newSVsv (POPs); 1501 sv = newSVsv (POPs);
1138 FREETMPS; LEAVE; 1502 FREETMPS; LEAVE;
1139 return sv; 1503 return sv;
1140 } 1504 }
1141 1505
1506 SvREFCNT_inc (sv);
1142 FREETMPS; LEAVE; 1507 FREETMPS; LEAVE;
1143 } 1508 }
1144 } 1509 }
1145 1510
1146 if (dec->json.cb_object) 1511 if (dec->json.cb_object)
1147 { 1512 {
1148 dSP; 1513 dSP;
1149 int count; 1514 int count;
1150 1515
1151 ENTER; SAVETMPS; PUSHMARK (SP); 1516 ENTER; SAVETMPS;
1517 SAVESTACK_POS ();
1518 PUSHMARK (SP);
1152 XPUSHs (sv_2mortal (sv)); 1519 XPUSHs (sv_2mortal (sv));
1153 1520
1154 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1521 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1155 1522
1156 if (count == 1) 1523 if (count == 1)
… …
1172 DEC_DEC_DEPTH; 1539 DEC_DEC_DEPTH;
1173 return 0; 1540 return 0;
1174} 1541}
1175 1542
1176static SV * 1543static SV *
1544decode_tag (dec_t *dec)
1545{
1546 SV *tag = 0;
1547 SV *val = 0;
1548
1549 if (!(dec->json.flags & F_ALLOW_TAGS))
1550 ERR ("malformed JSON string, neither array, object, number, string or atom");
1551
1552 ++dec->cur;
1553
1554 decode_ws (dec);
1555
1556 tag = decode_sv (dec);
1557 if (!tag)
1558 goto fail;
1559
1560 if (!SvPOK (tag))
1561 ERR ("malformed JSON string, (tag) must be a string");
1562
1563 decode_ws (dec);
1564
1565 if (*dec->cur != ')')
1566 ERR (") expected after tag");
1567
1568 ++dec->cur;
1569
1570 decode_ws (dec);
1571
1572 val = decode_sv (dec);
1573 if (!val)
1574 goto fail;
1575
1576 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1577 ERR ("malformed JSON string, tag value must be an array");
1578
1579 {
1580 AV *av = (AV *)SvRV (val);
1581 int i, len = av_len (av) + 1;
1582 HV *stash = gv_stashsv (tag, 0);
1583 SV *sv;
1584
1585 if (!stash)
1586 ERR ("cannot decode perl-object (package does not exist)");
1587
1588 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1589
1590 if (!method)
1591 ERR ("cannot decode perl-object (package does not have a THAW method)");
1592
1593 dSP;
1594
1595 ENTER; SAVETMPS;
1596 PUSHMARK (SP);
1597 EXTEND (SP, len + 2);
1598 // we re-bless the reference to get overload and other niceties right
1599 PUSHs (tag);
1600 PUSHs (sv_json);
1601
1602 for (i = 0; i < len; ++i)
1603 PUSHs (*av_fetch (av, i, 1));
1604
1605 PUTBACK;
1606 call_sv ((SV *)GvCV (method), G_SCALAR);
1607 SPAGAIN;
1608
1609 SvREFCNT_dec (tag);
1610 SvREFCNT_dec (val);
1611 sv = SvREFCNT_inc (POPs);
1612
1613 PUTBACK;
1614
1615 FREETMPS; LEAVE;
1616
1617 return sv;
1618 }
1619
1620fail:
1621 SvREFCNT_dec (tag);
1622 SvREFCNT_dec (val);
1623 return 0;
1624}
1625
1626static SV *
1177decode_sv (dec_t *dec) 1627decode_sv (dec_t *dec)
1178{ 1628{
1179 decode_ws (dec);
1180
1181 // the beauty of JSON: you need exactly one character lookahead 1629 // the beauty of JSON: you need exactly one character lookahead
1182 // to parse anything. 1630 // to parse everything.
1183 switch (*dec->cur) 1631 switch (*dec->cur)
1184 { 1632 {
1185 case '"': ++dec->cur; return decode_str (dec); 1633 case '"': ++dec->cur; return decode_str (dec);
1186 case '[': ++dec->cur; return decode_av (dec); 1634 case '[': ++dec->cur; return decode_av (dec);
1187 case '{': ++dec->cur; return decode_hv (dec); 1635 case '{': ++dec->cur; return decode_hv (dec);
1636 case '(': return decode_tag (dec);
1188 1637
1189 case '-': 1638 case '-':
1190 case '0': case '1': case '2': case '3': case '4': 1639 case '0': case '1': case '2': case '3': case '4':
1191 case '5': case '6': case '7': case '8': case '9': 1640 case '5': case '6': case '7': case '8': case '9':
1192 return decode_num (dec); 1641 return decode_num (dec);
1193 1642
1194 case 't': 1643 case 't':
1195 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1644 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1196 { 1645 {
1197 dec->cur += 4; 1646 dec->cur += 4;
1198 return SvREFCNT_inc (json_true); 1647#if JSON_SLOW
1648 bool_true = get_bool ("Types::Serialiser::true");
1649#endif
1650 return newSVsv (bool_true);
1199 } 1651 }
1200 else 1652 else
1201 ERR ("'true' expected"); 1653 ERR ("'true' expected");
1202 1654
1203 break; 1655 break;
1204 1656
1205 case 'f': 1657 case 'f':
1206 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1658 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1207 { 1659 {
1208 dec->cur += 5; 1660 dec->cur += 5;
1661#if JSON_SLOW
1662 bool_false = get_bool ("Types::Serialiser::false");
1663#endif
1209 return SvREFCNT_inc (json_false); 1664 return newSVsv (bool_false);
1210 } 1665 }
1211 else 1666 else
1212 ERR ("'false' expected"); 1667 ERR ("'false' expected");
1213 1668
1214 break; 1669 break;
… …
1223 ERR ("'null' expected"); 1678 ERR ("'null' expected");
1224 1679
1225 break; 1680 break;
1226 1681
1227 default: 1682 default:
1228 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1683 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1229 break; 1684 break;
1230 } 1685 }
1231 1686
1232fail: 1687fail:
1233 return 0; 1688 return 0;
1234} 1689}
1235 1690
1236static SV * 1691static SV *
1237decode_json (SV *string, JSON *json, UV *offset_return) 1692decode_json (SV *string, JSON *json, char **offset_return)
1238{ 1693{
1239 dec_t dec; 1694 dec_t dec;
1240 UV offset;
1241 SV *sv; 1695 SV *sv;
1242 1696
1697 /* work around bugs in 5.10 where manipulating magic values
1698 * makes perl ignore the magic in subsequent accesses.
1699 * 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).
1701 */
1243 SvGETMAGIC (string); 1702 /*SvGETMAGIC (string);*/
1703 if (SvMAGICAL (string) || !SvPOK (string))
1704 string = sv_2mortal (newSVsv (string));
1705
1244 SvUPGRADE (string, SVt_PV); 1706 SvUPGRADE (string, SVt_PV);
1245 1707
1246 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1708 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1709 * assertion with -DDEBUGGING, although SvCUR is documented to
1710 * return the xpv_cur field which certainly exists after upgrading.
1711 * according to nicholas clark, calling SvPOK fixes this.
1712 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1713 * and hope for the best.
1714 * Damnit, SvPV_nolen still trips over yet another assertion. This
1715 * assertion business is seriously broken, try yet another workaround
1716 * for the broken -DDEBUGGING.
1717 */
1718 {
1719#ifdef DEBUGGING
1720 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1721#else
1722 STRLEN offset = SvCUR (string);
1723#endif
1724
1725 if (offset > json->max_size && json->max_size)
1247 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1726 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1248 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1727 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1728 }
1249 1729
1250 if (json->flags & F_UTF8) 1730 if (DECODE_WANTS_OCTETS (json))
1251 sv_utf8_downgrade (string, 0); 1731 sv_utf8_downgrade (string, 0);
1252 else 1732 else
1253 sv_utf8_upgrade (string); 1733 sv_utf8_upgrade (string);
1254 1734
1255 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1735 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1256 1736
1257 dec.json = *json; 1737 dec.json = *json;
1258 dec.cur = SvPVX (string); 1738 dec.cur = SvPVX (string);
1259 dec.end = SvEND (string); 1739 dec.end = SvEND (string);
1260 dec.err = 0; 1740 dec.err = 0;
1261 dec.depth = 0; 1741 dec.depth = 0;
1262 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1263 1742
1264 if (dec.json.cb_object || dec.json.cb_sk_object) 1743 if (dec.json.cb_object || dec.json.cb_sk_object)
1265 dec.json.flags |= F_HOOK; 1744 dec.json.flags |= F_HOOK;
1266 1745
1267 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1746 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1747
1748 decode_ws (&dec);
1268 sv = decode_sv (&dec); 1749 sv = decode_sv (&dec);
1750
1751 if (offset_return)
1752 *offset_return = dec.cur;
1269 1753
1270 if (!(offset_return || !sv)) 1754 if (!(offset_return || !sv))
1271 { 1755 {
1272 // check for trailing garbage 1756 // check for trailing garbage
1273 decode_ws (&dec); 1757 decode_ws (&dec);
… …
1276 { 1760 {
1277 dec.err = "garbage after JSON object"; 1761 dec.err = "garbage after JSON object";
1278 SvREFCNT_dec (sv); 1762 SvREFCNT_dec (sv);
1279 sv = 0; 1763 sv = 0;
1280 } 1764 }
1281 }
1282
1283 if (offset_return || !sv)
1284 {
1285 offset = dec.json.flags & F_UTF8
1286 ? dec.cur - SvPVX (string)
1287 : utf8_distance (dec.cur, SvPVX (string));
1288
1289 if (offset_return)
1290 *offset_return = offset;
1291 } 1765 }
1292 1766
1293 if (!sv) 1767 if (!sv)
1294 { 1768 {
1295 SV *uni = sv_newmortal (); 1769 SV *uni = sv_newmortal ();
… …
1301 SAVEVPTR (PL_curcop); 1775 SAVEVPTR (PL_curcop);
1302 PL_curcop = &cop; 1776 PL_curcop = &cop;
1303 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1777 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1304 LEAVE; 1778 LEAVE;
1305 1779
1306 croak ("%s, at character offset %d [\"%s\"]", 1780 croak ("%s, at character offset %d (before \"%s\")",
1307 dec.err, 1781 dec.err,
1308 (int)offset, 1782 (int)ptr_to_index (string, dec.cur),
1309 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1783 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1310 } 1784 }
1311 1785
1312 sv = sv_2mortal (sv); 1786 sv = sv_2mortal (sv);
1313 1787
1314 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1788 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1315 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1789 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1316 1790
1317 return sv; 1791 return sv;
1792}
1793
1794/////////////////////////////////////////////////////////////////////////////
1795// incremental parser
1796
1797static void
1798incr_parse (JSON *self)
1799{
1800 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1801
1802 // the state machine here is a bit convoluted and could be simplified a lot
1803 // but this would make it slower, so...
1804
1805 for (;;)
1806 {
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)
1809 {
1810 // only used for initial whitespace skipping
1811 case INCR_M_WS:
1812 for (;;)
1813 {
1814 if (*p > 0x20)
1815 {
1816 if (*p == '#')
1817 {
1818 self->incr_mode = INCR_M_C0;
1819 goto incr_m_c;
1820 }
1821 else
1822 {
1823 self->incr_mode = INCR_M_JSON;
1824 goto incr_m_json;
1825 }
1826 }
1827 else if (!*p)
1828 goto interrupt;
1829
1830 ++p;
1831 }
1832
1833 // skip a single char inside a string (for \\-processing)
1834 case INCR_M_BS:
1835 if (!*p)
1836 goto interrupt;
1837
1838 ++p;
1839 self->incr_mode = INCR_M_STR;
1840 goto incr_m_str;
1841
1842 // inside #-style comments
1843 case INCR_M_C0:
1844 case INCR_M_C1:
1845 incr_m_c:
1846 for (;;)
1847 {
1848 if (*p == '\n')
1849 {
1850 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1851 break;
1852 }
1853 else if (!*p)
1854 goto interrupt;
1855
1856 ++p;
1857 }
1858
1859 break;
1860
1861 // inside a string
1862 case INCR_M_STR:
1863 incr_m_str:
1864 for (;;)
1865 {
1866 if (*p == '"')
1867 {
1868 ++p;
1869 self->incr_mode = INCR_M_JSON;
1870
1871 if (!self->incr_nest)
1872 goto interrupt;
1873
1874 goto incr_m_json;
1875 }
1876 else if (*p == '\\')
1877 {
1878 ++p; // "virtually" consumes character after \
1879
1880 if (!*p) // if at end of string we have to switch modes
1881 {
1882 self->incr_mode = INCR_M_BS;
1883 goto interrupt;
1884 }
1885 }
1886 else if (!*p)
1887 goto interrupt;
1888
1889 ++p;
1890 }
1891
1892 // after initial ws, outside string
1893 case INCR_M_JSON:
1894 incr_m_json:
1895 for (;;)
1896 {
1897 switch (*p++)
1898 {
1899 case 0:
1900 --p;
1901 goto interrupt;
1902
1903 case 0x09:
1904 case 0x0a:
1905 case 0x0d:
1906 case 0x20:
1907 if (!self->incr_nest)
1908 {
1909 --p; // do not eat the whitespace, let the next round do it
1910 goto interrupt;
1911 }
1912 break;
1913
1914 case '"':
1915 self->incr_mode = INCR_M_STR;
1916 goto incr_m_str;
1917
1918 case '[':
1919 case '{':
1920 case '(':
1921 if (++self->incr_nest > self->max_depth)
1922 croak (ERR_NESTING_EXCEEDED);
1923 break;
1924
1925 case ']':
1926 case '}':
1927 if (--self->incr_nest <= 0)
1928 goto interrupt;
1929 break;
1930
1931 case ')':
1932 --self->incr_nest;
1933 break;
1934
1935 case '#':
1936 self->incr_mode = INCR_M_C1;
1937 goto incr_m_c;
1938 }
1939 }
1940 }
1941
1942 modechange:
1943 ;
1944 }
1945
1946interrupt:
1947 self->incr_pos = p - SvPVX (self->incr_text);
1948 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1949 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1318} 1950}
1319 1951
1320///////////////////////////////////////////////////////////////////////////// 1952/////////////////////////////////////////////////////////////////////////////
1321// XS interface functions 1953// XS interface functions
1322 1954
… …
1331 i >= '0' && i <= '9' ? i - '0' 1963 i >= '0' && i <= '9' ? i - '0'
1332 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1964 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1333 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1965 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1334 : -1; 1966 : -1;
1335 1967
1336 json_stash = gv_stashpv ("JSON::XS" , 1); 1968 json_stash = gv_stashpv ("JSON::XS" , 1);
1337 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1969 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1970 bool_true = get_bool ("Types::Serialiser::true");
1971 bool_false = get_bool ("Types::Serialiser::false");
1338 1972
1339 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1973 sv_json = newSVpv ("JSON", 0);
1340 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1974 SvREADONLY_on (sv_json);
1975
1976 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1341} 1977}
1342 1978
1343PROTOTYPES: DISABLE 1979PROTOTYPES: DISABLE
1980
1981void CLONE (...)
1982 CODE:
1983 json_stash = 0;
1984 bool_stash = 0;
1344 1985
1345void new (char *klass) 1986void new (char *klass)
1346 PPCODE: 1987 PPCODE:
1347{ 1988{
1348 SV *pv = NEWSV (0, sizeof (JSON)); 1989 SV *pv = NEWSV (0, sizeof (JSON));
1349 SvPOK_only (pv); 1990 SvPOK_only (pv);
1350 Zero (SvPVX (pv), 1, JSON); 1991 json_init ((JSON *)SvPVX (pv));
1351 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1992 XPUSHs (sv_2mortal (sv_bless (
1352 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash))); 1993 newRV_noinc (pv),
1994 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1995 )));
1353} 1996}
1354 1997
1355void ascii (JSON *self, int enable = 1) 1998void ascii (JSON *self, int enable = 1)
1356 ALIAS: 1999 ALIAS:
1357 ascii = F_ASCII 2000 ascii = F_ASCII
… …
1364 pretty = F_PRETTY 2007 pretty = F_PRETTY
1365 allow_nonref = F_ALLOW_NONREF 2008 allow_nonref = F_ALLOW_NONREF
1366 shrink = F_SHRINK 2009 shrink = F_SHRINK
1367 allow_blessed = F_ALLOW_BLESSED 2010 allow_blessed = F_ALLOW_BLESSED
1368 convert_blessed = F_CONV_BLESSED 2011 convert_blessed = F_CONV_BLESSED
2012 relaxed = F_RELAXED
2013 allow_unknown = F_ALLOW_UNKNOWN
2014 allow_tags = F_ALLOW_TAGS
1369 PPCODE: 2015 PPCODE:
1370{ 2016{
1371 if (enable) 2017 if (enable)
1372 self->flags |= ix; 2018 self->flags |= ix;
1373 else 2019 else
1374 self->flags &= ~ix; 2020 self->flags &= ~ix;
1375 2021
1376 XPUSHs (ST (0)); 2022 XPUSHs (ST (0));
1377} 2023}
1378 2024
1379void max_depth (JSON *self, UV max_depth = 0x80000000UL) 2025void get_ascii (JSON *self)
2026 ALIAS:
2027 get_ascii = F_ASCII
2028 get_latin1 = F_LATIN1
2029 get_utf8 = F_UTF8
2030 get_indent = F_INDENT
2031 get_canonical = F_CANONICAL
2032 get_space_before = F_SPACE_BEFORE
2033 get_space_after = F_SPACE_AFTER
2034 get_allow_nonref = F_ALLOW_NONREF
2035 get_shrink = F_SHRINK
2036 get_allow_blessed = F_ALLOW_BLESSED
2037 get_convert_blessed = F_CONV_BLESSED
2038 get_relaxed = F_RELAXED
2039 get_allow_unknown = F_ALLOW_UNKNOWN
2040 get_allow_tags = F_ALLOW_TAGS
1380 PPCODE: 2041 PPCODE:
1381{ 2042 XPUSHs (boolSV (self->flags & ix));
1382 UV log2 = 0;
1383 2043
1384 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 2044void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1385 2045 PPCODE:
1386 while ((1UL << log2) < max_depth) 2046 self->max_depth = max_depth;
1387 ++log2;
1388
1389 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1390
1391 XPUSHs (ST (0)); 2047 XPUSHs (ST (0));
1392}
1393 2048
2049U32 get_max_depth (JSON *self)
2050 CODE:
2051 RETVAL = self->max_depth;
2052 OUTPUT:
2053 RETVAL
2054
1394void max_size (JSON *self, UV max_size = 0) 2055void max_size (JSON *self, U32 max_size = 0)
1395 PPCODE: 2056 PPCODE:
1396{ 2057 self->max_size = max_size;
1397 UV log2 = 0;
1398
1399 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1400 if (max_size == 1) max_size = 2;
1401
1402 while ((1UL << log2) < max_size)
1403 ++log2;
1404
1405 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1406
1407 XPUSHs (ST (0)); 2058 XPUSHs (ST (0));
1408} 2059
2060int get_max_size (JSON *self)
2061 CODE:
2062 RETVAL = self->max_size;
2063 OUTPUT:
2064 RETVAL
1409 2065
1410void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 2066void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1411 PPCODE: 2067 PPCODE:
1412{ 2068{
1413 SvREFCNT_dec (self->cb_object); 2069 SvREFCNT_dec (self->cb_object);
… …
1417} 2073}
1418 2074
1419void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 2075void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1420 PPCODE: 2076 PPCODE:
1421{ 2077{
1422 if (!self->cb_sk_object) 2078 if (!self->cb_sk_object)
1423 self->cb_sk_object = newHV (); 2079 self->cb_sk_object = newHV ();
1424 2080
1425 if (SvOK (cb)) 2081 if (SvOK (cb))
1426 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2082 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1427 else 2083 else
… …
1438 XPUSHs (ST (0)); 2094 XPUSHs (ST (0));
1439} 2095}
1440 2096
1441void encode (JSON *self, SV *scalar) 2097void encode (JSON *self, SV *scalar)
1442 PPCODE: 2098 PPCODE:
1443 XPUSHs (encode_json (scalar, self)); 2099 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2100 XPUSHs (scalar);
1444 2101
1445void decode (JSON *self, SV *jsonstr) 2102void decode (JSON *self, SV *jsonstr)
1446 PPCODE: 2103 PPCODE:
1447 XPUSHs (decode_json (jsonstr, self, 0)); 2104 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2105 XPUSHs (jsonstr);
1448 2106
1449void decode_prefix (JSON *self, SV *jsonstr) 2107void decode_prefix (JSON *self, SV *jsonstr)
1450 PPCODE: 2108 PPCODE:
1451{ 2109{
1452 UV offset; 2110 SV *sv;
2111 char *offset;
2112 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1453 EXTEND (SP, 2); 2113 EXTEND (SP, 2);
1454 PUSHs (decode_json (jsonstr, self, &offset)); 2114 PUSHs (sv);
1455 PUSHs (sv_2mortal (newSVuv (offset))); 2115 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
2116}
2117
2118void incr_parse (JSON *self, SV *jsonstr = 0)
2119 PPCODE:
2120{
2121 if (!self->incr_text)
2122 self->incr_text = newSVpvn ("", 0);
2123
2124 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2125 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2126 if (DECODE_WANTS_OCTETS (self))
2127 {
2128 if (self->incr_pos)
2129 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2130 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2131
2132 sv_utf8_downgrade (self->incr_text, 0);
2133 }
2134 else
2135 {
2136 sv_utf8_upgrade (self->incr_text);
2137
2138 if (self->incr_pos)
2139 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2140 - (U8 *)SvPVX (self->incr_text);
2141 }
2142
2143 // append data, if any
2144 if (jsonstr)
2145 {
2146 /* make sure both strings have same encoding */
2147 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2148 if (SvUTF8 (jsonstr))
2149 sv_utf8_downgrade (jsonstr, 0);
2150 else
2151 sv_utf8_upgrade (jsonstr);
2152
2153 /* and then just blindly append */
2154 {
2155 STRLEN len;
2156 const char *str = SvPV (jsonstr, len);
2157 STRLEN cur = SvCUR (self->incr_text);
2158
2159 if (SvLEN (self->incr_text) <= cur + len)
2160 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
2161
2162 Move (str, SvEND (self->incr_text), len, char);
2163 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
2165 }
2166 }
2167
2168 if (GIMME_V != G_VOID)
2169 do
2170 {
2171 SV *sv;
2172 char *offset;
2173
2174 if (!INCR_DONE (self))
2175 {
2176 incr_parse (self);
2177
2178 if (self->incr_pos > self->max_size && self->max_size)
2179 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2180 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2181
2182 if (!INCR_DONE (self))
2183 {
2184 // as an optimisation, do not accumulate white space in the incr buffer
2185 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2186 {
2187 self->incr_pos = 0;
2188 SvCUR_set (self->incr_text, 0);
2189 }
2190
2191 break;
2192 }
2193 }
2194
2195 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2196 XPUSHs (sv);
2197
2198 self->incr_pos -= offset - SvPVX (self->incr_text);
2199 self->incr_nest = 0;
2200 self->incr_mode = 0;
2201
2202 sv_chop (self->incr_text, offset);
2203 }
2204 while (GIMME_V == G_ARRAY);
2205}
2206
2207SV *incr_text (JSON *self)
2208 ATTRS: lvalue
2209 CODE:
2210{
2211 if (self->incr_pos)
2212 croak ("incr_text can not be called when the incremental parser already started parsing");
2213
2214 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
2215}
2216 OUTPUT:
2217 RETVAL
2218
2219void incr_skip (JSON *self)
2220 CODE:
2221{
2222 if (self->incr_pos)
2223 {
2224 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2225 self->incr_pos = 0;
2226 self->incr_nest = 0;
2227 self->incr_mode = 0;
2228 }
2229}
2230
2231void incr_reset (JSON *self)
2232 CODE:
2233{
2234 SvREFCNT_dec (self->incr_text);
2235 self->incr_text = 0;
2236 self->incr_pos = 0;
2237 self->incr_nest = 0;
2238 self->incr_mode = 0;
1456} 2239}
1457 2240
1458void DESTROY (JSON *self) 2241void DESTROY (JSON *self)
1459 CODE: 2242 CODE:
1460 SvREFCNT_dec (self->cb_sk_object); 2243 SvREFCNT_dec (self->cb_sk_object);
1461 SvREFCNT_dec (self->cb_object); 2244 SvREFCNT_dec (self->cb_object);
2245 SvREFCNT_dec (self->incr_text);
1462 2246
1463PROTOTYPES: ENABLE 2247PROTOTYPES: ENABLE
1464 2248
1465void to_json (SV *scalar) 2249void encode_json (SV *scalar)
1466 PPCODE: 2250 PPCODE:
1467{ 2251{
1468 JSON json = { F_DEFAULT | F_UTF8 }; 2252 JSON json;
1469 XPUSHs (encode_json (scalar, &json)); 2253 json_init (&json);
2254 json.flags |= F_UTF8;
2255 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2256 XPUSHs (scalar);
1470} 2257}
1471 2258
1472void from_json (SV *jsonstr) 2259void decode_json (SV *jsonstr)
1473 PPCODE: 2260 PPCODE:
1474{ 2261{
1475 JSON json = { F_DEFAULT | F_UTF8 }; 2262 JSON json;
1476 XPUSHs (decode_json (jsonstr, &json, 0)); 2263 json_init (&json);
2264 json.flags |= F_UTF8;
2265 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2266 XPUSHs (jsonstr);
1477} 2267}
1478 2268

Diff Legend

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