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 1.39 by root, Mon Jun 11 03:42:57 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
16// some old perls do not have this, try to make it work, no
17// guarantees, though. if it breaks, you get to keep the pieces.
18#ifndef UTF8_MAXBYTES
19# define UTF8_MAXBYTES 13
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)
35
14#define F_ASCII 0x00000001UL 36#define F_ASCII 0x00000001UL
15#define F_LATIN1 0x00000002UL 37#define F_LATIN1 0x00000002UL
16#define F_UTF8 0x00000004UL 38#define F_UTF8 0x00000004UL
17#define F_INDENT 0x00000008UL 39#define F_INDENT 0x00000008UL
18#define F_CANONICAL 0x00000010UL 40#define F_CANONICAL 0x00000010UL
19#define F_SPACE_BEFORE 0x00000020UL 41#define F_SPACE_BEFORE 0x00000020UL
20#define F_SPACE_AFTER 0x00000040UL 42#define F_SPACE_AFTER 0x00000040UL
21#define F_ALLOW_NONREF 0x00000100UL 43#define F_ALLOW_NONREF 0x00000100UL
22#define F_SHRINK 0x00000200UL 44#define F_SHRINK 0x00000200UL
45#define F_ALLOW_BLESSED 0x00000400UL
46#define F_CONV_BLESSED 0x00000800UL
23#define F_MAXDEPTH 0xf8000000UL 47#define F_RELAXED 0x00001000UL
24#define S_MAXDEPTH 27 48#define F_ALLOW_UNKNOWN 0x00002000UL
25 49#define F_ALLOW_TAGS 0x00004000UL
26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 50#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
27
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
30 51
31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 52#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
32#define F_DEFAULT (9UL << S_MAXDEPTH)
33 53
34#define INIT_SIZE 32 // initial scalar size to be allocated 54#define INIT_SIZE 32 // initial scalar size to be allocated
35#define INDENT_STEP 3 // spaces per indentation level 55#define INDENT_STEP 3 // spaces per indentation level
36 56
37#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
38 58
59#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
60
39#define SB do { 61#define SB do {
40#define SE } while (0) 62#define SE } while (0)
41 63
42#if __GNUC__ >= 3 64#if __GNUC__ >= 3
43# define expect(expr,value) __builtin_expect ((expr),(value)) 65# define expect(expr,value) __builtin_expect ((expr), (value))
44# define inline inline 66# define INLINE static inline
45#else 67#else
46# define expect(expr,value) (expr) 68# define expect(expr,value) (expr)
47# define inline static 69# define INLINE static
48#endif 70#endif
49 71
50#define expect_false(expr) expect ((expr) != 0, 0) 72#define expect_false(expr) expect ((expr) != 0, 0)
51#define expect_true(expr) expect ((expr) != 0, 1) 73#define expect_true(expr) expect ((expr) != 0, 1)
52 74
53static HV *json_stash; // JSON::XS:: 75#define IN_RANGE_INC(type,val,beg,end) \
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)
107
108typedef struct {
109 U32 flags;
110 U32 max_depth;
111 STRLEN max_size;
112
113 SV *cb_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;
121} JSON;
122
123INLINE void
124json_init (JSON *json)
125{
126 Zero (json, 1, JSON);
127 json->max_depth = 512;
128}
54 129
55///////////////////////////////////////////////////////////////////////////// 130/////////////////////////////////////////////////////////////////////////////
56// utility functions 131// utility functions
57 132
58static UV * 133INLINE SV *
59SvJSON (SV *sv) 134get_bool (const char *name)
60{ 135{
61 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 136 SV *sv = get_sv (name, 1);
62 croak ("object is not of type JSON::XS");
63 137
64 return &SvUVX (SvRV (sv)); 138 SvREADONLY_on (sv);
65} 139 SvREADONLY_on (SvRV (sv));
66 140
67static void 141 return sv;
142}
143
144INLINE void
68shrink (SV *sv) 145shrink (SV *sv)
69{ 146{
70 sv_utf8_downgrade (sv, 1); 147 sv_utf8_downgrade (sv, 1);
148
71 if (SvLEN (sv) > SvCUR (sv) + 1) 149 if (SvLEN (sv) > SvCUR (sv) + 1)
72 { 150 {
73#ifdef SvPV_shrink_to_cur 151#ifdef SvPV_shrink_to_cur
74 SvPV_shrink_to_cur (sv); 152 SvPV_shrink_to_cur (sv);
75#elif defined (SvPV_renew) 153#elif defined (SvPV_renew)
… …
81// 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
82// case of an error. 160// case of an error.
83// we special-case "safe" characters from U+80 .. U+7FF, 161// we special-case "safe" characters from U+80 .. U+7FF,
84// but use the very good perl function to parse anything else. 162// but use the very good perl function to parse anything else.
85// note that we never call this function for a ascii codepoints 163// note that we never call this function for a ascii codepoints
86inline UV 164INLINE UV
87decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 165decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
88{ 166{
89 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 167 if (expect_true (len >= 2
90 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 168 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
91 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 169 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
92 { 170 {
93 *clen = 2; 171 *clen = 2;
94 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 172 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
95 } 173 }
96 else 174 else
97 { 175 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
98 *clen = (STRLEN)-1; 176}
99 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 (;;)
100 } 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;
101} 345}
102 346
103///////////////////////////////////////////////////////////////////////////// 347/////////////////////////////////////////////////////////////////////////////
104// encoder 348// encoder
105 349
… …
107typedef struct 351typedef struct
108{ 352{
109 char *cur; // SvPVX (sv) + current output position 353 char *cur; // SvPVX (sv) + current output position
110 char *end; // SvEND (sv) 354 char *end; // SvEND (sv)
111 SV *sv; // result scalar 355 SV *sv; // result scalar
112 U32 flags; // F_* 356 JSON json;
113 U32 indent; // indentation level 357 U32 indent; // indentation level
114 U32 maxdepth; // max. indentation/recursion level 358 UV limit; // escape character values >= this value when encoding
115} enc_t; 359} enc_t;
116 360
117inline void 361INLINE void
118need (enc_t *enc, STRLEN len) 362need (enc_t *enc, STRLEN len)
119{ 363{
120 if (expect_false (enc->cur + len >= enc->end)) 364 if (expect_false (enc->cur + len >= enc->end))
121 { 365 {
122 STRLEN cur = enc->cur - SvPVX (enc->sv); 366 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
123 SvGROW (enc->sv, cur + len + 1); 367 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
124 enc->cur = SvPVX (enc->sv) + cur; 368 enc->cur = SvPVX (enc->sv) + cur;
125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 369 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
126 } 370 }
127} 371}
128 372
129inline void 373INLINE void
130encode_ch (enc_t *enc, char ch) 374encode_ch (enc_t *enc, char ch)
131{ 375{
132 need (enc, 1); 376 need (enc, 1);
133 *enc->cur++ = ch; 377 *enc->cur++ = ch;
134} 378}
… …
178 STRLEN clen; 422 STRLEN clen;
179 UV uch; 423 UV uch;
180 424
181 if (is_utf8) 425 if (is_utf8)
182 { 426 {
183 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
184 uch = decode_utf8 (str, end - str, &clen); 427 uch = decode_utf8 (str, end - str, &clen);
185 if (clen == (STRLEN)-1) 428 if (clen == (STRLEN)-1)
186 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 429 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
187 } 430 }
188 else 431 else
189 { 432 {
190 uch = ch; 433 uch = ch;
191 clen = 1; 434 clen = 1;
192 } 435 }
193 436
194 if (uch > 0x10FFFFUL) 437 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
195 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
196
197 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
198 { 438 {
199 if (uch > 0xFFFFUL) 439 if (uch >= 0x10000UL)
200 { 440 {
441 if (uch >= 0x110000UL)
442 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
443
201 need (enc, len += 11); 444 need (enc, len += 11);
202 sprintf (enc->cur, "\\u%04x\\u%04x", 445 sprintf (enc->cur, "\\u%04x\\u%04x",
203 (int)((uch - 0x10000) / 0x400 + 0xD800), 446 (int)((uch - 0x10000) / 0x400 + 0xD800),
204 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 447 (int)((uch - 0x10000) % 0x400 + 0xDC00));
205 enc->cur += 12; 448 enc->cur += 12;
206 } 449 }
207 else 450 else
208 { 451 {
209 static char hexdigit [16] = "0123456789abcdef";
210 need (enc, len += 5); 452 need (enc, len += 5);
211 *enc->cur++ = '\\'; 453 *enc->cur++ = '\\';
212 *enc->cur++ = 'u'; 454 *enc->cur++ = 'u';
213 *enc->cur++ = hexdigit [ uch >> 12 ]; 455 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
214 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 456 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
215 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 457 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
216 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 458 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
217 } 459 }
218 460
219 str += clen; 461 str += clen;
220 } 462 }
221 else if (enc->flags & F_LATIN1) 463 else if (enc->json.flags & F_LATIN1)
222 { 464 {
223 *enc->cur++ = uch; 465 *enc->cur++ = uch;
224 str += clen; 466 str += clen;
225 } 467 }
226 else if (is_utf8) 468 else if (is_utf8)
… …
233 while (--clen); 475 while (--clen);
234 } 476 }
235 else 477 else
236 { 478 {
237 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
238 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 480 enc->cur = encode_utf8 (enc->cur, uch);
239 ++str; 481 ++str;
240 } 482 }
241 } 483 }
242 } 484 }
243 } 485 }
244 486
245 --len; 487 --len;
246 } 488 }
247} 489}
248 490
249inline void 491INLINE void
250encode_indent (enc_t *enc) 492encode_indent (enc_t *enc)
251{ 493{
252 if (enc->flags & F_INDENT) 494 if (enc->json.flags & F_INDENT)
253 { 495 {
254 int spaces = enc->indent * INDENT_STEP; 496 int spaces = enc->indent * INDENT_STEP;
255 497
256 need (enc, spaces); 498 need (enc, spaces);
257 memset (enc->cur, ' ', spaces); 499 memset (enc->cur, ' ', spaces);
258 enc->cur += spaces; 500 enc->cur += spaces;
259 } 501 }
260} 502}
261 503
262inline void 504INLINE void
263encode_space (enc_t *enc) 505encode_space (enc_t *enc)
264{ 506{
265 need (enc, 1); 507 need (enc, 1);
266 encode_ch (enc, ' '); 508 encode_ch (enc, ' ');
267} 509}
268 510
269inline void 511INLINE void
270encode_nl (enc_t *enc) 512encode_nl (enc_t *enc)
271{ 513{
272 if (enc->flags & F_INDENT) 514 if (enc->json.flags & F_INDENT)
273 { 515 {
274 need (enc, 1); 516 need (enc, 1);
275 encode_ch (enc, '\n'); 517 encode_ch (enc, '\n');
276 } 518 }
277} 519}
278 520
279inline void 521INLINE void
280encode_comma (enc_t *enc) 522encode_comma (enc_t *enc)
281{ 523{
282 encode_ch (enc, ','); 524 encode_ch (enc, ',');
283 525
284 if (enc->flags & F_INDENT) 526 if (enc->json.flags & F_INDENT)
285 encode_nl (enc); 527 encode_nl (enc);
286 else if (enc->flags & F_SPACE_AFTER) 528 else if (enc->json.flags & F_SPACE_AFTER)
287 encode_space (enc); 529 encode_space (enc);
288} 530}
289 531
290static void encode_sv (enc_t *enc, SV *sv); 532static void encode_sv (enc_t *enc, SV *sv);
291 533
292static void 534static void
293encode_av (enc_t *enc, AV *av) 535encode_av (enc_t *enc, AV *av)
294{ 536{
295 int i, len = av_len (av); 537 int i, len = av_len (av);
296 538
297 if (enc->indent >= enc->maxdepth) 539 if (enc->indent >= enc->json.max_depth)
298 croak ("data structure too deep (hit recursion limit)"); 540 croak (ERR_NESTING_EXCEEDED);
299 541
300 encode_ch (enc, '['); encode_nl (enc); 542 encode_ch (enc, '[');
301 ++enc->indent;
302 543
544 if (len >= 0)
545 {
546 encode_nl (enc); ++enc->indent;
547
303 for (i = 0; i <= len; ++i) 548 for (i = 0; i <= len; ++i)
304 { 549 {
550 SV **svp = av_fetch (av, i, 0);
551
305 encode_indent (enc); 552 encode_indent (enc);
306 encode_sv (enc, *av_fetch (av, i, 0));
307 553
554 if (svp)
555 encode_sv (enc, *svp);
556 else
557 encode_str (enc, "null", 4, 0);
558
308 if (i < len) 559 if (i < len)
309 encode_comma (enc); 560 encode_comma (enc);
310 } 561 }
311 562
563 encode_nl (enc); --enc->indent; encode_indent (enc);
564 }
565
312 encode_nl (enc); 566 encode_ch (enc, ']');
313
314 --enc->indent;
315 encode_indent (enc); encode_ch (enc, ']');
316} 567}
317 568
318static void 569static void
319encode_he (enc_t *enc, HE *he) 570encode_hk (enc_t *enc, HE *he)
320{ 571{
321 encode_ch (enc, '"'); 572 encode_ch (enc, '"');
322 573
323 if (HeKLEN (he) == HEf_SVKEY) 574 if (HeKLEN (he) == HEf_SVKEY)
324 { 575 {
325 SV *sv = HeSVKEY (he); 576 SV *sv = HeSVKEY (he);
326 STRLEN len; 577 STRLEN len;
327 char *str; 578 char *str;
328 579
329 SvGETMAGIC (sv); 580 SvGETMAGIC (sv);
330 str = SvPV (sv, len); 581 str = SvPV (sv, len);
331 582
332 encode_str (enc, str, len, SvUTF8 (sv)); 583 encode_str (enc, str, len, SvUTF8 (sv));
333 } 584 }
334 else 585 else
335 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 586 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
336 587
337 encode_ch (enc, '"'); 588 encode_ch (enc, '"');
338 589
339 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 590 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
340 encode_ch (enc, ':'); 591 encode_ch (enc, ':');
341 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 592 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
342 encode_sv (enc, HeVAL (he));
343} 593}
344 594
345// compare hash entries, used when all keys are bytestrings 595// compare hash entries, used when all keys are bytestrings
346static int 596static int
347he_cmp_fast (const void *a_, const void *b_) 597he_cmp_fast (const void *a_, const void *b_)
… …
352 HE *b = *(HE **)b_; 602 HE *b = *(HE **)b_;
353 603
354 STRLEN la = HeKLEN (a); 604 STRLEN la = HeKLEN (a);
355 STRLEN lb = HeKLEN (b); 605 STRLEN lb = HeKLEN (b);
356 606
357 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 607 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
358 cmp = la - lb; 608 cmp = lb - la;
359 609
360 return cmp; 610 return cmp;
361} 611}
362 612
363// 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
364static int 614static int
365he_cmp_slow (const void *a, const void *b) 615he_cmp_slow (const void *a, const void *b)
366{ 616{
367 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 617 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
368} 618}
369 619
370static void 620static void
371encode_hv (enc_t *enc, HV *hv) 621encode_hv (enc_t *enc, HV *hv)
372{ 622{
373 int count, i; 623 HE *he;
374 624
375 if (enc->indent >= enc->maxdepth) 625 if (enc->indent >= enc->json.max_depth)
376 croak ("data structure too deep (hit recursion limit)"); 626 croak (ERR_NESTING_EXCEEDED);
377 627
378 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 628 encode_ch (enc, '{');
379 629
380 if ((count = hv_iterinit (hv)))
381 {
382 // for canonical output we have to sort by keys first 630 // for canonical output we have to sort by keys first
383 // actually, this is mostly due to the stupid so-called 631 // actually, this is mostly due to the stupid so-called
384 // security workaround added somewhere in 5.8.x. 632 // security workaround added somewhere in 5.8.x
385 // that randomises hash orderings 633 // that randomises hash orderings
386 if (enc->flags & F_CANONICAL) 634 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
635 {
636 int count = hv_iterinit (hv);
637
638 if (SvMAGICAL (hv))
387 { 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 {
388 int fast = 1; 654 int i, fast = 1;
389 HE *he; 655 HE *hes_stack [STACK_HES];
390#if defined(__BORLANDC__) || defined(_MSC_VER) 656 HE **hes = hes_stack;
391 HE **hes = _alloca (count * sizeof (HE)); 657
392#else 658 // allocate larger arrays on the heap
393 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 659 if (count > STACK_HES)
394#endif 660 {
661 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
662 hes = (HE **)SvPVX (sv);
663 }
395 664
396 i = 0; 665 i = 0;
397 while ((he = hv_iternext (hv))) 666 while ((he = hv_iternext (hv)))
398 { 667 {
399 hes [i++] = he; 668 hes [i++] = he;
… …
421 690
422 FREETMPS; 691 FREETMPS;
423 LEAVE; 692 LEAVE;
424 } 693 }
425 694
426 for (i = 0; i < count; ++i) 695 encode_nl (enc); ++enc->indent;
696
697 while (count--)
427 { 698 {
428 encode_indent (enc); 699 encode_indent (enc);
700 he = hes [count];
429 encode_he (enc, hes [i]); 701 encode_hk (enc, he);
702 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
430 703
431 if (i < count - 1) 704 if (count)
432 encode_comma (enc); 705 encode_comma (enc);
433 } 706 }
434 707
435 encode_nl (enc); 708 encode_nl (enc); --enc->indent; encode_indent (enc);
436 } 709 }
710 }
437 else 711 else
712 {
713 if (hv_iterinit (hv) || SvMAGICAL (hv))
714 if ((he = hv_iternext (hv)))
438 { 715 {
439 HE *he = hv_iternext (hv); 716 encode_nl (enc); ++enc->indent;
440 717
441 for (;;) 718 for (;;)
442 { 719 {
443 encode_indent (enc); 720 encode_indent (enc);
444 encode_he (enc, he); 721 encode_hk (enc, he);
722 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
445 723
446 if (!(he = hv_iternext (hv))) 724 if (!(he = hv_iternext (hv)))
447 break; 725 break;
448 726
449 encode_comma (enc); 727 encode_comma (enc);
450 } 728 }
451 729
452 encode_nl (enc); 730 encode_nl (enc); --enc->indent; encode_indent (enc);
453 } 731 }
454 } 732 }
455 733
456 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 734 encode_ch (enc, '}');
457} 735}
458 736
459// encode objects, arrays and special \0=false and \1=true values. 737// encode objects, arrays and special \0=false and \1=true values.
460static void 738static void
461encode_rv (enc_t *enc, SV *sv) 739encode_rv (enc_t *enc, SV *sv)
462{ 740{
463 svtype svt; 741 svtype svt;
742 GV *method;
464 743
465 SvGETMAGIC (sv); 744 SvGETMAGIC (sv);
466 svt = SvTYPE (sv); 745 svt = SvTYPE (sv);
467 746
747 if (expect_false (SvOBJECT (sv)))
748 {
749 HV *stash = SvSTASH (sv);
750
751 if (stash == bool_stash)
752 {
753 if (SvIV (sv))
754 encode_str (enc, "true", 4, 0);
755 else
756 encode_str (enc, "false", 5, 0);
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);
825 else
826 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
827 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
828 }
468 if (svt == SVt_PVHV) 829 else if (svt == SVt_PVHV)
469 encode_hv (enc, (HV *)sv); 830 encode_hv (enc, (HV *)sv);
470 else if (svt == SVt_PVAV) 831 else if (svt == SVt_PVAV)
471 encode_av (enc, (AV *)sv); 832 encode_av (enc, (AV *)sv);
472 else if (svt < SVt_PVAV) 833 else if (svt < SVt_PVAV)
473 { 834 {
474 if (SvNIOK (sv) && SvIV (sv) == 0) 835 int bool_type = ref_bool_type (sv);
836
837 if (bool_type == 1)
838 encode_str (enc, "true", 4, 0);
839 else if (bool_type == 0)
475 encode_str (enc, "false", 5, 0); 840 encode_str (enc, "false", 5, 0);
476 else if (SvNIOK (sv) && SvIV (sv) == 1) 841 else if (enc->json.flags & F_ALLOW_UNKNOWN)
477 encode_str (enc, "true", 4, 0); 842 encode_str (enc, "null", 4, 0);
478 else 843 else
479 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",
480 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 845 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
481 } 846 }
847 else if (enc->json.flags & F_ALLOW_UNKNOWN)
848 encode_str (enc, "null", 4, 0);
482 else 849 else
483 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",
484 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 851 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
485} 852}
486 853
… …
504 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 871 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
505 enc->cur += strlen (enc->cur); 872 enc->cur += strlen (enc->cur);
506 } 873 }
507 else if (SvIOKp (sv)) 874 else if (SvIOKp (sv))
508 { 875 {
509 // 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
510 if (SvUV (sv) & ~(UV)0x7fff) 877 // we assume two's complement
511 { 878 // we assume no aliasing issues in the union
512 // large integer, use the (rather slow) snprintf way. 879 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
513 need (enc, sizeof (UV) * 3); 880 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
514 enc->cur +=
515 SvIsUV(sv)
516 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
517 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
518 }
519 else
520 { 881 {
521 // optimise the "small number case" 882 // optimise the "small number case"
522 // 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
523 I32 i = SvIV (sv); 885 I32 i = SvIVX (sv);
524 U32 u; 886 U32 u;
525 char digit, nz = 0; 887 char digit, nz = 0;
526 888
527 need (enc, 6); 889 need (enc, 6);
528 890
… …
534 896
535 // 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
536 // 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,
537 // resulting in a net multiplication by 10. 899 // resulting in a net multiplication by 10.
538 // we always write the digit to memory but conditionally increment 900 // we always write the digit to memory but conditionally increment
539 // the pointer, to ease the usage of conditional move instructions. 901 // the pointer, to enable the use of conditional move instructions.
540 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;
541 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;
542 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;
543 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;
544 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'
545 } 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 }
546 } 917 }
547 else if (SvROK (sv)) 918 else if (SvROK (sv))
548 encode_rv (enc, SvRV (sv)); 919 encode_rv (enc, SvRV (sv));
549 else if (!SvOK (sv)) 920 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
550 encode_str (enc, "null", 4, 0); 921 encode_str (enc, "null", 4, 0);
551 else 922 else
552 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",
553 SvPV_nolen (sv), SvFLAGS (sv)); 924 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
554} 925}
555 926
556static SV * 927static SV *
557encode_json (SV *scalar, U32 flags) 928encode_json (SV *scalar, JSON *json)
558{ 929{
559 enc_t enc; 930 enc_t enc;
560 931
561 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 932 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
562 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)");
563 934
564 enc.flags = flags; 935 enc.json = *json;
565 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 936 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
566 enc.cur = SvPVX (enc.sv); 937 enc.cur = SvPVX (enc.sv);
567 enc.end = SvEND (enc.sv); 938 enc.end = SvEND (enc.sv);
568 enc.indent = 0; 939 enc.indent = 0;
569 enc.maxdepth = DEC_DEPTH (flags); 940 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
941 : enc.json.flags & F_LATIN1 ? 0x000100UL
942 : 0x110000UL;
570 943
571 SvPOK_only (enc.sv); 944 SvPOK_only (enc.sv);
572 encode_sv (&enc, scalar); 945 encode_sv (&enc, scalar);
946 encode_nl (&enc);
573 947
574 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 948 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
575 *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
576 950
577 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 951 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
578 SvUTF8_on (enc.sv); 952 SvUTF8_on (enc.sv);
579 953
580 if (enc.flags & F_SHRINK) 954 if (enc.json.flags & F_SHRINK)
581 shrink (enc.sv); 955 shrink (enc.sv);
582 956
583 return enc.sv; 957 return enc.sv;
584} 958}
585 959
… …
590typedef struct 964typedef struct
591{ 965{
592 char *cur; // current parser pointer 966 char *cur; // current parser pointer
593 char *end; // end of input string 967 char *end; // end of input string
594 const char *err; // parse error, if != 0 968 const char *err; // parse error, if != 0
595 U32 flags; // F_* 969 JSON json;
596 U32 depth; // recursion depth 970 U32 depth; // recursion depth
597 U32 maxdepth; // recursion depth limit 971 U32 maxdepth; // recursion depth limit
598} dec_t; 972} dec_t;
599 973
600inline 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
601decode_ws (dec_t *dec) 984decode_ws (dec_t *dec)
602{ 985{
603 for (;;) 986 for (;;)
604 { 987 {
605 char ch = *dec->cur; 988 char ch = *dec->cur;
606 989
607 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 }
608 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 1002 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
609 break; 1003 break; // parse error, but let higher level handle it, gives better error messages
610 1004
611 ++dec->cur; 1005 ++dec->cur;
612 } 1006 }
613} 1007}
614 1008
… …
618 if (*dec->cur != ch) \ 1012 if (*dec->cur != ch) \
619 ERR (# ch " expected"); \ 1013 ERR (# ch " expected"); \
620 ++dec->cur; \ 1014 ++dec->cur; \
621 SE 1015 SE
622 1016
623#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)
624#define DEC_DEC_DEPTH --dec->depth 1018#define DEC_DEC_DEPTH --dec->depth
625 1019
626static SV *decode_sv (dec_t *dec); 1020static SV *decode_sv (dec_t *dec);
627 1021
628static signed char decode_hexdigit[256]; 1022static signed char decode_hexdigit[256];
… …
631decode_4hex (dec_t *dec) 1025decode_4hex (dec_t *dec)
632{ 1026{
633 signed char d1, d2, d3, d4; 1027 signed char d1, d2, d3, d4;
634 unsigned char *cur = (unsigned char *)dec->cur; 1028 unsigned char *cur = (unsigned char *)dec->cur;
635 1029
636 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected"); 1030 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
637 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected"); 1031 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
638 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected"); 1032 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
639 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected"); 1033 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
640 1034
641 dec->cur += 4; 1035 dec->cur += 4;
642 1036
643 return ((UV)d1) << 12 1037 return ((UV)d1) << 12
644 | ((UV)d2) << 8 1038 | ((UV)d2) << 8
… …
720 1114
721 if (hi >= 0x80) 1115 if (hi >= 0x80)
722 { 1116 {
723 utf8 = 1; 1117 utf8 = 1;
724 1118
725 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1119 cur = encode_utf8 (cur, hi);
726 } 1120 }
727 else 1121 else
728 *cur++ = hi; 1122 *cur++ = hi;
729 } 1123 }
730 break; 1124 break;
… …
732 default: 1126 default:
733 --dec_cur; 1127 --dec_cur;
734 ERR ("illegal backslash escape sequence in string"); 1128 ERR ("illegal backslash escape sequence in string");
735 } 1129 }
736 } 1130 }
737 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1131 else if (expect_true (ch >= 0x20 && ch < 0x80))
738 *cur++ = ch; 1132 *cur++ = ch;
739 else if (ch >= 0x80) 1133 else if (ch >= 0x80)
740 { 1134 {
741 STRLEN clen; 1135 STRLEN clen;
742 UV uch;
743 1136
744 --dec_cur; 1137 --dec_cur;
745 1138
746 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1139 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
747 if (clen == (STRLEN)-1) 1140 if (clen == (STRLEN)-1)
748 ERR ("malformed UTF-8 character in JSON string"); 1141 ERR ("malformed UTF-8 character in JSON string");
749 1142
750 do 1143 do
751 *cur++ = *dec_cur++; 1144 *cur++ = *dec_cur++;
752 while (--clen); 1145 while (--clen);
753 1146
754 utf8 = 1; 1147 utf8 = 1;
755 } 1148 }
1149 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1150 *cur++ = ch;
756 else 1151 else
757 { 1152 {
758 --dec_cur; 1153 --dec_cur;
759 1154
760 if (!ch) 1155 if (!ch)
… …
768 { 1163 {
769 STRLEN len = cur - buf; 1164 STRLEN len = cur - buf;
770 1165
771 if (sv) 1166 if (sv)
772 { 1167 {
773 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
774 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1173 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
775 SvCUR_set (sv, SvCUR (sv) + len); 1174 SvCUR_set (sv, SvCUR (sv) + len);
776 } 1175 }
777 else 1176 else
778 sv = newSVpvn (buf, len); 1177 sv = newSVpvn (buf, len);
… …
863 is_nv = 1; 1262 is_nv = 1;
864 } 1263 }
865 1264
866 if (!is_nv) 1265 if (!is_nv)
867 { 1266 {
1267 int len = dec->cur - start;
1268
868 // 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
869 if (*start == '-') 1270 if (*start == '-')
870 switch (dec->cur - start) 1271 switch (len)
871 { 1272 {
872 case 2: return newSViv (-( start [1] - '0' )); 1273 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
873 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1274 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
874 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));
875 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));
876 } 1278 }
877 else 1279 else
878 switch (dec->cur - start) 1280 switch (len)
879 { 1281 {
880 case 1: return newSViv ( start [0] - '0' ); 1282 case 1: return newSViv ( start [0] - '0' * 1);
881 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1283 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
882 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);
883 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);
884 } 1287 }
885 1288
886 { 1289 {
887 UV uv; 1290 UV uv;
888 int numtype = grok_number (start, dec->cur - start, &uv); 1291 int numtype = grok_number (start, len, &uv);
889 if (numtype & IS_NUMBER_IN_UV) 1292 if (numtype & IS_NUMBER_IN_UV)
890 if (numtype & IS_NUMBER_NEG) 1293 if (numtype & IS_NUMBER_NEG)
891 { 1294 {
892 if (uv < (UV)IV_MIN) 1295 if (uv < (UV)IV_MIN)
893 return newSViv (-(IV)uv); 1296 return newSViv (-(IV)uv);
894 } 1297 }
895 else 1298 else
896 return newSVuv (uv); 1299 return newSVuv (uv);
897 } 1300 }
898 }
899 1301
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
900 return newSVnv (Atof (start)); 1314 return newSVnv (json_atof (start));
901 1315
902fail: 1316fail:
903 return 0; 1317 return 0;
904} 1318}
905 1319
… …
929 if (*dec->cur == ']') 1343 if (*dec->cur == ']')
930 { 1344 {
931 ++dec->cur; 1345 ++dec->cur;
932 break; 1346 break;
933 } 1347 }
934 1348
935 if (*dec->cur != ',') 1349 if (*dec->cur != ',')
936 ERR (", or ] expected while parsing array"); 1350 ERR (", or ] expected while parsing array");
937 1351
938 ++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 }
939 } 1361 }
940 1362
941 DEC_DEC_DEPTH; 1363 DEC_DEC_DEPTH;
942 return newRV_noinc ((SV *)av); 1364 return newRV_noinc ((SV *)av);
943 1365
… …
948} 1370}
949 1371
950static SV * 1372static SV *
951decode_hv (dec_t *dec) 1373decode_hv (dec_t *dec)
952{ 1374{
1375 SV *sv;
953 HV *hv = newHV (); 1376 HV *hv = newHV ();
954 1377
955 DEC_INC_DEPTH; 1378 DEC_INC_DEPTH;
956 decode_ws (dec); 1379 decode_ws (dec);
957 1380
958 if (*dec->cur == '}') 1381 if (*dec->cur == '}')
959 ++dec->cur; 1382 ++dec->cur;
960 else 1383 else
961 for (;;) 1384 for (;;)
962 { 1385 {
1386 EXPECT_CH ('"');
1387
1388 // heuristic: assume that
1389 // a) decode_str + hv_store_ent are abysmally slow.
1390 // b) most hash keys are short, simple ascii text.
1391 // => try to "fast-match" such strings to avoid
1392 // the overhead of decode_str + hv_store_ent.
1393 {
963 SV *key, *value; 1394 SV *value;
1395 char *p = dec->cur;
1396 char *e = p + 24; // only try up to 24 bytes
964 1397
965 decode_ws (dec); EXPECT_CH ('"'); 1398 for (;;)
966
967 key = decode_str (dec);
968 if (!key)
969 goto fail;
970
971 decode_ws (dec); EXPECT_CH (':');
972
973 value = decode_sv (dec);
974 if (!value)
975 { 1399 {
1400 // the >= 0x80 is false on most architectures
1401 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1402 {
1403 // slow path, back up and use decode_str
1404 SV *key = decode_str (dec);
1405 if (!key)
1406 goto fail;
1407
1408 decode_ws (dec); EXPECT_CH (':');
1409
1410 decode_ws (dec);
1411 value = decode_sv (dec);
1412 if (!value)
1413 {
1414 SvREFCNT_dec (key);
1415 goto fail;
1416 }
1417
1418 hv_store_ent (hv, key, value, 0);
976 SvREFCNT_dec (key); 1419 SvREFCNT_dec (key);
1420
1421 break;
1422 }
1423 else if (*p == '"')
1424 {
1425 // fast path, got a simple key
1426 char *key = dec->cur;
1427 int len = p - key;
1428 dec->cur = p + 1;
1429
1430 decode_ws (dec); EXPECT_CH (':');
1431
1432 decode_ws (dec);
1433 value = decode_sv (dec);
1434 if (!value)
977 goto fail; 1435 goto fail;
1436
1437 hv_store (hv, key, len, value, 0);
1438
1439 break;
1440 }
1441
1442 ++p;
978 } 1443 }
979 1444 }
980 hv_store_ent (hv, key, value, 0);
981 SvREFCNT_dec (key);
982 1445
983 decode_ws (dec); 1446 decode_ws (dec);
984 1447
985 if (*dec->cur == '}') 1448 if (*dec->cur == '}')
986 { 1449 {
… …
990 1453
991 if (*dec->cur != ',') 1454 if (*dec->cur != ',')
992 ERR (", or } expected while parsing object/hash"); 1455 ERR (", or } expected while parsing object/hash");
993 1456
994 ++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 }
995 } 1466 }
996 1467
997 DEC_DEC_DEPTH; 1468 DEC_DEC_DEPTH;
998 return newRV_noinc ((SV *)hv); 1469 sv = newRV_noinc ((SV *)hv);
1470
1471 // check filter callbacks
1472 if (dec->json.flags & F_HOOK)
1473 {
1474 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1475 {
1476 HE *cb, *he;
1477
1478 hv_iterinit (hv);
1479 he = hv_iternext (hv);
1480 hv_iterinit (hv);
1481
1482 // the next line creates a mortal sv each time it's called.
1483 // might want to optimise this for common cases.
1484 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1485
1486 if (cb)
1487 {
1488 dSP;
1489 int count;
1490
1491 ENTER; SAVETMPS;
1492 SAVESTACK_POS ();
1493 PUSHMARK (SP);
1494 XPUSHs (HeVAL (he));
1495 sv_2mortal (sv);
1496
1497 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1498
1499 if (count == 1)
1500 {
1501 sv = newSVsv (POPs);
1502 FREETMPS; LEAVE;
1503 return sv;
1504 }
1505
1506 SvREFCNT_inc (sv);
1507 FREETMPS; LEAVE;
1508 }
1509 }
1510
1511 if (dec->json.cb_object)
1512 {
1513 dSP;
1514 int count;
1515
1516 ENTER; SAVETMPS;
1517 SAVESTACK_POS ();
1518 PUSHMARK (SP);
1519 XPUSHs (sv_2mortal (sv));
1520
1521 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1522
1523 if (count == 1)
1524 {
1525 sv = newSVsv (POPs);
1526 FREETMPS; LEAVE;
1527 return sv;
1528 }
1529
1530 SvREFCNT_inc (sv);
1531 FREETMPS; LEAVE;
1532 }
1533 }
1534
1535 return sv;
999 1536
1000fail: 1537fail:
1001 SvREFCNT_dec (hv); 1538 SvREFCNT_dec (hv);
1002 DEC_DEC_DEPTH; 1539 DEC_DEC_DEPTH;
1003 return 0; 1540 return 0;
1004} 1541}
1005 1542
1006static 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 *
1007decode_sv (dec_t *dec) 1627decode_sv (dec_t *dec)
1008{ 1628{
1009 decode_ws (dec); 1629 // the beauty of JSON: you need exactly one character lookahead
1630 // to parse everything.
1010 switch (*dec->cur) 1631 switch (*dec->cur)
1011 { 1632 {
1012 case '"': ++dec->cur; return decode_str (dec); 1633 case '"': ++dec->cur; return decode_str (dec);
1013 case '[': ++dec->cur; return decode_av (dec); 1634 case '[': ++dec->cur; return decode_av (dec);
1014 case '{': ++dec->cur; return decode_hv (dec); 1635 case '{': ++dec->cur; return decode_hv (dec);
1636 case '(': return decode_tag (dec);
1015 1637
1016 case '-': 1638 case '-':
1017 case '0': case '1': case '2': case '3': case '4': 1639 case '0': case '1': case '2': case '3': case '4':
1018 case '5': case '6': case '7': case '8': case '9': 1640 case '5': case '6': case '7': case '8': case '9':
1019 return decode_num (dec); 1641 return decode_num (dec);
1020 1642
1021 case 't': 1643 case 't':
1022 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1644 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1023 { 1645 {
1024 dec->cur += 4; 1646 dec->cur += 4;
1647#if JSON_SLOW
1648 bool_true = get_bool ("Types::Serialiser::true");
1649#endif
1025 return newSViv (1); 1650 return newSVsv (bool_true);
1026 } 1651 }
1027 else 1652 else
1028 ERR ("'true' expected"); 1653 ERR ("'true' expected");
1029 1654
1030 break; 1655 break;
1031 1656
1032 case 'f': 1657 case 'f':
1033 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1658 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1034 { 1659 {
1035 dec->cur += 5; 1660 dec->cur += 5;
1661#if JSON_SLOW
1662 bool_false = get_bool ("Types::Serialiser::false");
1663#endif
1036 return newSViv (0); 1664 return newSVsv (bool_false);
1037 } 1665 }
1038 else 1666 else
1039 ERR ("'false' expected"); 1667 ERR ("'false' expected");
1040 1668
1041 break; 1669 break;
… …
1050 ERR ("'null' expected"); 1678 ERR ("'null' expected");
1051 1679
1052 break; 1680 break;
1053 1681
1054 default: 1682 default:
1055 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1683 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1056 break; 1684 break;
1057 } 1685 }
1058 1686
1059fail: 1687fail:
1060 return 0; 1688 return 0;
1061} 1689}
1062 1690
1063static SV * 1691static SV *
1064decode_json (SV *string, U32 flags, UV *offset_return) 1692decode_json (SV *string, JSON *json, char **offset_return)
1065{ 1693{
1066 dec_t dec; 1694 dec_t dec;
1067 UV offset;
1068 SV *sv; 1695 SV *sv;
1069 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 */
1070 SvGETMAGIC (string); 1702 /*SvGETMAGIC (string);*/
1703 if (SvMAGICAL (string) || !SvPOK (string))
1704 string = sv_2mortal (newSVsv (string));
1705
1071 SvUPGRADE (string, SVt_PV); 1706 SvUPGRADE (string, SVt_PV);
1072 1707
1073 if (flags & F_UTF8) 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)
1726 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1727 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1728 }
1729
1730 if (DECODE_WANTS_OCTETS (json))
1074 sv_utf8_downgrade (string, 0); 1731 sv_utf8_downgrade (string, 0);
1075 else 1732 else
1076 sv_utf8_upgrade (string); 1733 sv_utf8_upgrade (string);
1077 1734
1078 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1735 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1079 1736
1080 dec.flags = flags; 1737 dec.json = *json;
1081 dec.cur = SvPVX (string); 1738 dec.cur = SvPVX (string);
1082 dec.end = SvEND (string); 1739 dec.end = SvEND (string);
1083 dec.err = 0; 1740 dec.err = 0;
1084 dec.depth = 0; 1741 dec.depth = 0;
1085 dec.maxdepth = DEC_DEPTH (dec.flags); 1742
1743 if (dec.json.cb_object || dec.json.cb_sk_object)
1744 dec.json.flags |= F_HOOK;
1086 1745
1087 *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);
1088 sv = decode_sv (&dec); 1749 sv = decode_sv (&dec);
1750
1751 if (offset_return)
1752 *offset_return = dec.cur;
1089 1753
1090 if (!(offset_return || !sv)) 1754 if (!(offset_return || !sv))
1091 { 1755 {
1092 // check for trailing garbage 1756 // check for trailing garbage
1093 decode_ws (&dec); 1757 decode_ws (&dec);
… …
1096 { 1760 {
1097 dec.err = "garbage after JSON object"; 1761 dec.err = "garbage after JSON object";
1098 SvREFCNT_dec (sv); 1762 SvREFCNT_dec (sv);
1099 sv = 0; 1763 sv = 0;
1100 } 1764 }
1101 }
1102
1103 if (offset_return || !sv)
1104 {
1105 offset = dec.flags & F_UTF8
1106 ? dec.cur - SvPVX (string)
1107 : utf8_distance (dec.cur, SvPVX (string));
1108
1109 if (offset_return)
1110 *offset_return = offset;
1111 } 1765 }
1112 1766
1113 if (!sv) 1767 if (!sv)
1114 { 1768 {
1115 SV *uni = sv_newmortal (); 1769 SV *uni = sv_newmortal ();
… …
1121 SAVEVPTR (PL_curcop); 1775 SAVEVPTR (PL_curcop);
1122 PL_curcop = &cop; 1776 PL_curcop = &cop;
1123 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);
1124 LEAVE; 1778 LEAVE;
1125 1779
1126 croak ("%s, at character offset %d [\"%s\"]", 1780 croak ("%s, at character offset %d (before \"%s\")",
1127 dec.err, 1781 dec.err,
1128 (int)offset, 1782 (int)ptr_to_index (string, dec.cur),
1129 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1783 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1130 } 1784 }
1131 1785
1132 sv = sv_2mortal (sv); 1786 sv = sv_2mortal (sv);
1133 1787
1134 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1788 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1135 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)");
1136 1790
1137 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
1138} 1950}
1139 1951
1140///////////////////////////////////////////////////////////////////////////// 1952/////////////////////////////////////////////////////////////////////////////
1141// XS interface functions 1953// XS interface functions
1142 1954
1143MODULE = JSON::XS PACKAGE = JSON::XS 1955MODULE = JSON::XS PACKAGE = JSON::XS
1144 1956
1145BOOT: 1957BOOT:
1146{ 1958{
1147 int i; 1959 int i;
1148
1149 memset (decode_hexdigit, 0xff, 256);
1150 1960
1151 for (i = 0; i < 256; ++i) 1961 for (i = 0; i < 256; ++i)
1152 decode_hexdigit [i] = 1962 decode_hexdigit [i] =
1153 i >= '0' && i <= '9' ? i - '0' 1963 i >= '0' && i <= '9' ? i - '0'
1154 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1964 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1155 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1965 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1156 : -1; 1966 : -1;
1157 1967
1158 json_stash = gv_stashpv ("JSON::XS", 1); 1968 json_stash = gv_stashpv ("JSON::XS" , 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");
1972
1973 sv_json = newSVpv ("JSON", 0);
1974 SvREADONLY_on (sv_json);
1975
1976 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1159} 1977}
1160 1978
1161PROTOTYPES: DISABLE 1979PROTOTYPES: DISABLE
1162 1980
1163SV *new (char *dummy) 1981void CLONE (...)
1164 CODE: 1982 CODE:
1165 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1983 json_stash = 0;
1984 bool_stash = 0;
1985
1986void new (char *klass)
1987 PPCODE:
1988{
1989 SV *pv = NEWSV (0, sizeof (JSON));
1990 SvPOK_only (pv);
1991 json_init ((JSON *)SvPVX (pv));
1992 XPUSHs (sv_2mortal (sv_bless (
1993 newRV_noinc (pv),
1994 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1995 )));
1996}
1997
1998void ascii (JSON *self, int enable = 1)
1999 ALIAS:
2000 ascii = F_ASCII
2001 latin1 = F_LATIN1
2002 utf8 = F_UTF8
2003 indent = F_INDENT
2004 canonical = F_CANONICAL
2005 space_before = F_SPACE_BEFORE
2006 space_after = F_SPACE_AFTER
2007 pretty = F_PRETTY
2008 allow_nonref = F_ALLOW_NONREF
2009 shrink = F_SHRINK
2010 allow_blessed = F_ALLOW_BLESSED
2011 convert_blessed = F_CONV_BLESSED
2012 relaxed = F_RELAXED
2013 allow_unknown = F_ALLOW_UNKNOWN
2014 allow_tags = F_ALLOW_TAGS
2015 PPCODE:
2016{
2017 if (enable)
2018 self->flags |= ix;
2019 else
2020 self->flags &= ~ix;
2021
2022 XPUSHs (ST (0));
2023}
2024
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
2041 PPCODE:
2042 XPUSHs (boolSV (self->flags & ix));
2043
2044void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
2045 PPCODE:
2046 self->max_depth = max_depth;
2047 XPUSHs (ST (0));
2048
2049U32 get_max_depth (JSON *self)
2050 CODE:
2051 RETVAL = self->max_depth;
1166 OUTPUT: 2052 OUTPUT:
1167 RETVAL 2053 RETVAL
1168 2054
1169SV *ascii (SV *self, int enable = 1) 2055void max_size (JSON *self, U32 max_size = 0)
1170 ALIAS: 2056 PPCODE:
1171 ascii = F_ASCII 2057 self->max_size = max_size;
1172 latin1 = F_LATIN1 2058 XPUSHs (ST (0));
1173 utf8 = F_UTF8 2059
1174 indent = F_INDENT 2060int get_max_size (JSON *self)
1175 canonical = F_CANONICAL
1176 space_before = F_SPACE_BEFORE
1177 space_after = F_SPACE_AFTER
1178 pretty = F_PRETTY
1179 allow_nonref = F_ALLOW_NONREF
1180 shrink = F_SHRINK
1181 CODE: 2061 CODE:
1182{ 2062 RETVAL = self->max_size;
1183 UV *uv = SvJSON (self);
1184 if (enable)
1185 *uv |= ix;
1186 else
1187 *uv &= ~ix;
1188
1189 RETVAL = newSVsv (self);
1190}
1191 OUTPUT: 2063 OUTPUT:
1192 RETVAL 2064 RETVAL
1193 2065
1194SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 2066void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
2067 PPCODE:
2068{
2069 SvREFCNT_dec (self->cb_object);
2070 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
2071
2072 XPUSHs (ST (0));
2073}
2074
2075void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
2076 PPCODE:
2077{
2078 if (!self->cb_sk_object)
2079 self->cb_sk_object = newHV ();
2080
2081 if (SvOK (cb))
2082 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
2083 else
2084 {
2085 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
2086
2087 if (!HvKEYS (self->cb_sk_object))
2088 {
2089 SvREFCNT_dec (self->cb_sk_object);
2090 self->cb_sk_object = 0;
2091 }
2092 }
2093
2094 XPUSHs (ST (0));
2095}
2096
2097void encode (JSON *self, SV *scalar)
2098 PPCODE:
2099 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2100 XPUSHs (scalar);
2101
2102void decode (JSON *self, SV *jsonstr)
2103 PPCODE:
2104 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2105 XPUSHs (jsonstr);
2106
2107void decode_prefix (JSON *self, SV *jsonstr)
2108 PPCODE:
2109{
2110 SV *sv;
2111 char *offset;
2112 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
2113 EXTEND (SP, 2);
2114 PUSHs (sv);
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
1195 CODE: 2209 CODE:
1196{ 2210{
1197 UV *uv = SvJSON (self); 2211 if (self->incr_pos)
1198 UV log2 = 0; 2212 croak ("incr_text can not be called when the incremental parser already started parsing");
1199 2213
1200 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 2214 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1201
1202 while ((1UL << log2) < max_depth)
1203 ++log2;
1204
1205 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1206
1207 RETVAL = newSVsv (self);
1208} 2215}
1209 OUTPUT: 2216 OUTPUT:
1210 RETVAL 2217 RETVAL
1211 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;
2239}
2240
2241void DESTROY (JSON *self)
2242 CODE:
2243 SvREFCNT_dec (self->cb_sk_object);
2244 SvREFCNT_dec (self->cb_object);
2245 SvREFCNT_dec (self->incr_text);
2246
2247PROTOTYPES: ENABLE
2248
1212void encode (SV *self, SV *scalar) 2249void encode_json (SV *scalar)
1213 PPCODE: 2250 PPCODE:
1214 XPUSHs (encode_json (scalar, *SvJSON (self))); 2251{
2252 JSON json;
2253 json_init (&json);
2254 json.flags |= F_UTF8;
2255 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2256 XPUSHs (scalar);
2257}
1215 2258
1216void decode (SV *self, SV *jsonstr) 2259void decode_json (SV *jsonstr)
1217 PPCODE: 2260 PPCODE:
1218 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1219
1220void decode_prefix (SV *self, SV *jsonstr)
1221 PPCODE:
1222{ 2261{
1223 UV offset; 2262 JSON json;
1224 EXTEND (SP, 2); 2263 json_init (&json);
1225 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 2264 json.flags |= F_UTF8;
1226 PUSHs (sv_2mortal (newSVuv (offset))); 2265 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2266 XPUSHs (jsonstr);
1227} 2267}
1228 2268
1229PROTOTYPES: ENABLE
1230
1231void to_json (SV *scalar)
1232 ALIAS:
1233 objToJson = 0
1234 PPCODE:
1235 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1236
1237void from_json (SV *jsonstr)
1238 ALIAS:
1239 jsonToObj = 0
1240 PPCODE:
1241 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1242

Diff Legend

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