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 1.36 by root, Wed Jun 6 17:27:33 2007 UTC vs.
Revision 1.129 by root, Wed Nov 16 18:06:34 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 512 // special-case strings of up to this size 57#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
58
59#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
38 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
175 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
176}
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 (;;)
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 }
97 { 277 }
98 *clen = (STRLEN)-1; 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 return -1;
326}
327
328// returns whether scalar is not a reference in the sense of allow_nonref
329static int
330json_nonref (SV *scalar)
331{
332 if (!SvROK (scalar))
333 return 1;
334
335 scalar = SvRV (scalar);
336
337 if (SvTYPE (scalar) >= SVt_PVMG)
338 {
339 if (SvSTASH (scalar) == bool_stash)
99 return (UV)-1; 340 return 1;
341
342 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
343 return 1;
100 } 344 }
345
346 return 0;
101} 347}
102 348
103///////////////////////////////////////////////////////////////////////////// 349/////////////////////////////////////////////////////////////////////////////
104// encoder 350// encoder
105 351
… …
107typedef struct 353typedef struct
108{ 354{
109 char *cur; // SvPVX (sv) + current output position 355 char *cur; // SvPVX (sv) + current output position
110 char *end; // SvEND (sv) 356 char *end; // SvEND (sv)
111 SV *sv; // result scalar 357 SV *sv; // result scalar
112 U32 flags; // F_* 358 JSON json;
113 U32 indent; // indentation level 359 U32 indent; // indentation level
114 U32 maxdepth; // max. indentation/recursion level 360 UV limit; // escape character values >= this value when encoding
115} enc_t; 361} enc_t;
116 362
117inline void 363INLINE void
118need (enc_t *enc, STRLEN len) 364need (enc_t *enc, STRLEN len)
119{ 365{
120 if (expect_false (enc->cur + len >= enc->end)) 366 if (expect_false (enc->cur + len >= enc->end))
121 { 367 {
122 STRLEN cur = enc->cur - SvPVX (enc->sv); 368 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
123 SvGROW (enc->sv, cur + len + 1); 369 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
124 enc->cur = SvPVX (enc->sv) + cur; 370 enc->cur = SvPVX (enc->sv) + cur;
125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 371 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
126 } 372 }
127} 373}
128 374
129inline void 375INLINE void
130encode_ch (enc_t *enc, char ch) 376encode_ch (enc_t *enc, char ch)
131{ 377{
132 need (enc, 1); 378 need (enc, 1);
133 *enc->cur++ = ch; 379 *enc->cur++ = ch;
134} 380}
… …
178 STRLEN clen; 424 STRLEN clen;
179 UV uch; 425 UV uch;
180 426
181 if (is_utf8) 427 if (is_utf8)
182 { 428 {
183 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
184 uch = decode_utf8 (str, end - str, &clen); 429 uch = decode_utf8 (str, end - str, &clen);
185 if (clen == (STRLEN)-1) 430 if (clen == (STRLEN)-1)
186 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 431 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
187 } 432 }
188 else 433 else
189 { 434 {
190 uch = ch; 435 uch = ch;
191 clen = 1; 436 clen = 1;
192 } 437 }
193 438
194 if (uch > 0x10FFFFUL) 439 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 { 440 {
199 if (uch > 0xFFFFUL) 441 if (uch >= 0x10000UL)
200 { 442 {
443 if (uch >= 0x110000UL)
444 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
445
201 need (enc, len += 11); 446 need (enc, len += 11);
202 sprintf (enc->cur, "\\u%04x\\u%04x", 447 sprintf (enc->cur, "\\u%04x\\u%04x",
203 (int)((uch - 0x10000) / 0x400 + 0xD800), 448 (int)((uch - 0x10000) / 0x400 + 0xD800),
204 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 449 (int)((uch - 0x10000) % 0x400 + 0xDC00));
205 enc->cur += 12; 450 enc->cur += 12;
206 } 451 }
207 else 452 else
208 { 453 {
209 static char hexdigit [16] = "0123456789abcdef";
210 need (enc, len += 5); 454 need (enc, len += 5);
211 *enc->cur++ = '\\'; 455 *enc->cur++ = '\\';
212 *enc->cur++ = 'u'; 456 *enc->cur++ = 'u';
213 *enc->cur++ = hexdigit [ uch >> 12 ]; 457 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
214 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 458 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
215 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 459 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
216 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 460 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
217 } 461 }
218 462
219 str += clen; 463 str += clen;
220 } 464 }
221 else if (enc->flags & F_LATIN1) 465 else if (enc->json.flags & F_LATIN1)
222 { 466 {
223 *enc->cur++ = uch; 467 *enc->cur++ = uch;
224 str += clen; 468 str += clen;
225 } 469 }
226 else if (is_utf8) 470 else if (is_utf8)
… …
233 while (--clen); 477 while (--clen);
234 } 478 }
235 else 479 else
236 { 480 {
237 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 481 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
238 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 482 enc->cur = encode_utf8 (enc->cur, uch);
239 ++str; 483 ++str;
240 } 484 }
241 } 485 }
242 } 486 }
243 } 487 }
244 488
245 --len; 489 --len;
246 } 490 }
247} 491}
248 492
249inline void 493INLINE void
250encode_indent (enc_t *enc) 494encode_indent (enc_t *enc)
251{ 495{
252 if (enc->flags & F_INDENT) 496 if (enc->json.flags & F_INDENT)
253 { 497 {
254 int spaces = enc->indent * INDENT_STEP; 498 int spaces = enc->indent * INDENT_STEP;
255 499
256 need (enc, spaces); 500 need (enc, spaces);
257 memset (enc->cur, ' ', spaces); 501 memset (enc->cur, ' ', spaces);
258 enc->cur += spaces; 502 enc->cur += spaces;
259 } 503 }
260} 504}
261 505
262inline void 506INLINE void
263encode_space (enc_t *enc) 507encode_space (enc_t *enc)
264{ 508{
265 need (enc, 1); 509 need (enc, 1);
266 encode_ch (enc, ' '); 510 encode_ch (enc, ' ');
267} 511}
268 512
269inline void 513INLINE void
270encode_nl (enc_t *enc) 514encode_nl (enc_t *enc)
271{ 515{
272 if (enc->flags & F_INDENT) 516 if (enc->json.flags & F_INDENT)
273 { 517 {
274 need (enc, 1); 518 need (enc, 1);
275 encode_ch (enc, '\n'); 519 encode_ch (enc, '\n');
276 } 520 }
277} 521}
278 522
279inline void 523INLINE void
280encode_comma (enc_t *enc) 524encode_comma (enc_t *enc)
281{ 525{
282 encode_ch (enc, ','); 526 encode_ch (enc, ',');
283 527
284 if (enc->flags & F_INDENT) 528 if (enc->json.flags & F_INDENT)
285 encode_nl (enc); 529 encode_nl (enc);
286 else if (enc->flags & F_SPACE_AFTER) 530 else if (enc->json.flags & F_SPACE_AFTER)
287 encode_space (enc); 531 encode_space (enc);
288} 532}
289 533
290static void encode_sv (enc_t *enc, SV *sv); 534static void encode_sv (enc_t *enc, SV *sv);
291 535
292static void 536static void
293encode_av (enc_t *enc, AV *av) 537encode_av (enc_t *enc, AV *av)
294{ 538{
295 int i, len = av_len (av); 539 int i, len = av_len (av);
296 540
297 if (enc->indent >= enc->maxdepth) 541 if (enc->indent >= enc->json.max_depth)
298 croak ("data structure too deep (hit recursion limit)"); 542 croak (ERR_NESTING_EXCEEDED);
299 543
300 encode_ch (enc, '['); encode_nl (enc); 544 encode_ch (enc, '[');
301 ++enc->indent;
302 545
546 if (len >= 0)
547 {
548 encode_nl (enc); ++enc->indent;
549
303 for (i = 0; i <= len; ++i) 550 for (i = 0; i <= len; ++i)
304 { 551 {
552 SV **svp = av_fetch (av, i, 0);
553
305 encode_indent (enc); 554 encode_indent (enc);
306 encode_sv (enc, *av_fetch (av, i, 0));
307 555
556 if (svp)
557 encode_sv (enc, *svp);
558 else
559 encode_str (enc, "null", 4, 0);
560
308 if (i < len) 561 if (i < len)
309 encode_comma (enc); 562 encode_comma (enc);
310 } 563 }
311 564
565 encode_nl (enc); --enc->indent; encode_indent (enc);
566 }
567
312 encode_nl (enc); 568 encode_ch (enc, ']');
313
314 --enc->indent;
315 encode_indent (enc); encode_ch (enc, ']');
316} 569}
317 570
318static void 571static void
319encode_he (enc_t *enc, HE *he) 572encode_hk (enc_t *enc, HE *he)
320{ 573{
321 encode_ch (enc, '"'); 574 encode_ch (enc, '"');
322 575
323 if (HeKLEN (he) == HEf_SVKEY) 576 if (HeKLEN (he) == HEf_SVKEY)
324 { 577 {
325 SV *sv = HeSVKEY (he); 578 SV *sv = HeSVKEY (he);
326 STRLEN len; 579 STRLEN len;
327 char *str; 580 char *str;
328 581
329 SvGETMAGIC (sv); 582 SvGETMAGIC (sv);
330 str = SvPV (sv, len); 583 str = SvPV (sv, len);
331 584
332 encode_str (enc, str, len, SvUTF8 (sv)); 585 encode_str (enc, str, len, SvUTF8 (sv));
333 } 586 }
334 else 587 else
335 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 588 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
336 589
337 encode_ch (enc, '"'); 590 encode_ch (enc, '"');
338 591
339 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 592 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
340 encode_ch (enc, ':'); 593 encode_ch (enc, ':');
341 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 594 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
342 encode_sv (enc, HeVAL (he));
343} 595}
344 596
345// compare hash entries, used when all keys are bytestrings 597// compare hash entries, used when all keys are bytestrings
346static int 598static int
347he_cmp_fast (const void *a_, const void *b_) 599he_cmp_fast (const void *a_, const void *b_)
… …
352 HE *b = *(HE **)b_; 604 HE *b = *(HE **)b_;
353 605
354 STRLEN la = HeKLEN (a); 606 STRLEN la = HeKLEN (a);
355 STRLEN lb = HeKLEN (b); 607 STRLEN lb = HeKLEN (b);
356 608
357 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 609 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
358 cmp = la - lb; 610 cmp = lb - la;
359 611
360 return cmp; 612 return cmp;
361} 613}
362 614
363// compare hash entries, used when some keys are sv's or utf-x 615// compare hash entries, used when some keys are sv's or utf-x
364static int 616static int
365he_cmp_slow (const void *a, const void *b) 617he_cmp_slow (const void *a, const void *b)
366{ 618{
367 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 619 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
368} 620}
369 621
370static void 622static void
371encode_hv (enc_t *enc, HV *hv) 623encode_hv (enc_t *enc, HV *hv)
372{ 624{
373 int count, i; 625 HE *he;
374 626
375 if (enc->indent >= enc->maxdepth) 627 if (enc->indent >= enc->json.max_depth)
376 croak ("data structure too deep (hit recursion limit)"); 628 croak (ERR_NESTING_EXCEEDED);
377 629
378 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 630 encode_ch (enc, '{');
379 631
380 if ((count = hv_iterinit (hv)))
381 {
382 // for canonical output we have to sort by keys first 632 // for canonical output we have to sort by keys first
383 // actually, this is mostly due to the stupid so-called 633 // actually, this is mostly due to the stupid so-called
384 // security workaround added somewhere in 5.8.x. 634 // security workaround added somewhere in 5.8.x
385 // that randomises hash orderings 635 // that randomises hash orderings
386 if (enc->flags & F_CANONICAL) 636 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
637 {
638 int count = hv_iterinit (hv);
639
640 if (SvMAGICAL (hv))
387 { 641 {
642 // need to count by iterating. could improve by dynamically building the vector below
643 // but I don't care for the speed of this special case.
644 // note also that we will run into undefined behaviour when the two iterations
645 // do not result in the same count, something I might care for in some later release.
646
647 count = 0;
648 while (hv_iternext (hv))
649 ++count;
650
651 hv_iterinit (hv);
652 }
653
654 if (count)
655 {
388 int fast = 1; 656 int i, fast = 1;
389 HE *he; 657 HE *hes_stack [STACK_HES];
390#if defined(__BORLANDC__) || defined(_MSC_VER) 658 HE **hes = hes_stack;
391 HE **hes = _alloca (count * sizeof (HE)); 659
392#else 660 // allocate larger arrays on the heap
393 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 661 if (count > STACK_HES)
394#endif 662 {
663 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
664 hes = (HE **)SvPVX (sv);
665 }
395 666
396 i = 0; 667 i = 0;
397 while ((he = hv_iternext (hv))) 668 while ((he = hv_iternext (hv)))
398 { 669 {
399 hes [i++] = he; 670 hes [i++] = he;
… …
421 692
422 FREETMPS; 693 FREETMPS;
423 LEAVE; 694 LEAVE;
424 } 695 }
425 696
426 for (i = 0; i < count; ++i) 697 encode_nl (enc); ++enc->indent;
698
699 while (count--)
427 { 700 {
428 encode_indent (enc); 701 encode_indent (enc);
702 he = hes [count];
429 encode_he (enc, hes [i]); 703 encode_hk (enc, he);
704 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
430 705
431 if (i < count - 1) 706 if (count)
432 encode_comma (enc); 707 encode_comma (enc);
433 } 708 }
434 709
435 encode_nl (enc); 710 encode_nl (enc); --enc->indent; encode_indent (enc);
436 } 711 }
712 }
437 else 713 else
714 {
715 if (hv_iterinit (hv) || SvMAGICAL (hv))
716 if ((he = hv_iternext (hv)))
438 { 717 {
439 HE *he = hv_iternext (hv); 718 encode_nl (enc); ++enc->indent;
440 719
441 for (;;) 720 for (;;)
442 { 721 {
443 encode_indent (enc); 722 encode_indent (enc);
444 encode_he (enc, he); 723 encode_hk (enc, he);
724 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
445 725
446 if (!(he = hv_iternext (hv))) 726 if (!(he = hv_iternext (hv)))
447 break; 727 break;
448 728
449 encode_comma (enc); 729 encode_comma (enc);
450 } 730 }
451 731
452 encode_nl (enc); 732 encode_nl (enc); --enc->indent; encode_indent (enc);
453 } 733 }
454 } 734 }
455 735
456 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 736 encode_ch (enc, '}');
457} 737}
458 738
459// encode objects, arrays and special \0=false and \1=true values. 739// encode objects, arrays and special \0=false and \1=true values.
460static void 740static void
461encode_rv (enc_t *enc, SV *sv) 741encode_rv (enc_t *enc, SV *sv)
462{ 742{
463 svtype svt; 743 svtype svt;
744 GV *method;
464 745
465 SvGETMAGIC (sv); 746 SvGETMAGIC (sv);
466 svt = SvTYPE (sv); 747 svt = SvTYPE (sv);
467 748
749 if (expect_false (SvOBJECT (sv)))
750 {
751 HV *stash = SvSTASH (sv);
752
753 if (stash == bool_stash)
754 {
755 if (SvIV (sv))
756 encode_str (enc, "true", 4, 0);
757 else
758 encode_str (enc, "false", 5, 0);
759 }
760 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
761 {
762 int count;
763 dSP;
764
765 ENTER; SAVETMPS;
766 SAVESTACK_POS ();
767 PUSHMARK (SP);
768 EXTEND (SP, 2);
769 // we re-bless the reference to get overload and other niceties right
770 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
771 PUSHs (sv_json);
772
773 PUTBACK;
774 count = call_sv ((SV *)GvCV (method), G_ARRAY);
775 SPAGAIN;
776
777 // catch this surprisingly common error
778 if (SvROK (TOPs) && SvRV (TOPs) == sv)
779 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
780
781 encode_ch (enc, '(');
782 encode_ch (enc, '"');
783 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
784 encode_ch (enc, '"');
785 encode_ch (enc, ')');
786 encode_ch (enc, '[');
787
788 while (count)
789 {
790 encode_sv (enc, SP[1 - count--]);
791
792 if (count)
793 encode_ch (enc, ',');
794 }
795
796 encode_ch (enc, ']');
797
798 FREETMPS; LEAVE;
799 }
800 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
801 {
802 dSP;
803
804 ENTER; SAVETMPS;
805 PUSHMARK (SP);
806 // we re-bless the reference to get overload and other niceties right
807 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
808
809 // calling with G_SCALAR ensures that we always get a 1 return value
810 PUTBACK;
811 call_sv ((SV *)GvCV (method), G_SCALAR);
812 SPAGAIN;
813
814 // catch this surprisingly common error
815 if (SvROK (TOPs) && SvRV (TOPs) == sv)
816 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
817
818 sv = POPs;
819 PUTBACK;
820
821 encode_sv (enc, sv);
822
823 FREETMPS; LEAVE;
824 }
825 else if (enc->json.flags & F_ALLOW_BLESSED)
826 encode_str (enc, "null", 4, 0);
827 else
828 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
829 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
830 }
468 if (svt == SVt_PVHV) 831 else if (svt == SVt_PVHV)
469 encode_hv (enc, (HV *)sv); 832 encode_hv (enc, (HV *)sv);
470 else if (svt == SVt_PVAV) 833 else if (svt == SVt_PVAV)
471 encode_av (enc, (AV *)sv); 834 encode_av (enc, (AV *)sv);
472 else if (svt < SVt_PVAV) 835 else if (svt < SVt_PVAV)
473 { 836 {
474 if (SvNIOK (sv) && SvIV (sv) == 0) 837 int bool_type = ref_bool_type (sv);
838
839 if (bool_type == 1)
840 encode_str (enc, "true", 4, 0);
841 else if (bool_type == 0)
475 encode_str (enc, "false", 5, 0); 842 encode_str (enc, "false", 5, 0);
476 else if (SvNIOK (sv) && SvIV (sv) == 1) 843 else if (enc->json.flags & F_ALLOW_UNKNOWN)
477 encode_str (enc, "true", 4, 0); 844 encode_str (enc, "null", 4, 0);
478 else 845 else
479 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 846 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
480 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 847 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
481 } 848 }
849 else if (enc->json.flags & F_ALLOW_UNKNOWN)
850 encode_str (enc, "null", 4, 0);
482 else 851 else
483 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 852 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
484 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 853 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
485} 854}
486 855
… …
497 encode_str (enc, str, len, SvUTF8 (sv)); 866 encode_str (enc, str, len, SvUTF8 (sv));
498 encode_ch (enc, '"'); 867 encode_ch (enc, '"');
499 } 868 }
500 else if (SvNOKp (sv)) 869 else if (SvNOKp (sv))
501 { 870 {
871 // trust that perl will do the right thing w.r.t. JSON syntax.
502 need (enc, NV_DIG + 32); 872 need (enc, NV_DIG + 32);
503 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 873 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
504 enc->cur += strlen (enc->cur); 874 enc->cur += strlen (enc->cur);
505 } 875 }
506 else if (SvIOKp (sv)) 876 else if (SvIOKp (sv))
507 { 877 {
508 // we assume we can always read an IV as a UV 878 // we assume we can always read an IV as a UV and vice versa
509 if (SvUV (sv) & ~(UV)0x7fff) 879 // we assume two's complement
880 // we assume no aliasing issues in the union
881 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
882 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
510 { 883 {
884 // optimise the "small number case"
885 // code will likely be branchless and use only a single multiplication
886 // works for numbers up to 59074
887 I32 i = SvIVX (sv);
888 U32 u;
889 char digit, nz = 0;
890
511 need (enc, 64); 891 need (enc, 6);
512 enc->cur += 892
513 SvIsUV(sv) 893 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
514 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 894 u = i < 0 ? -i : i;
515 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 895
896 // convert to 4.28 fixed-point representation
897 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
898
899 // now output digit by digit, each time masking out the integer part
900 // and multiplying by 5 while moving the decimal point one to the right,
901 // resulting in a net multiplication by 10.
902 // we always write the digit to memory but conditionally increment
903 // the pointer, to enable the use of conditional move instructions.
904 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
905 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
906 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
907 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
908 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
516 } 909 }
517 else 910 else
518 { 911 {
519 // optimise the "small number case" 912 // large integer, use the (rather slow) snprintf way.
520 // code will likely be branchless and use only a single multiplication
521 I32 i = SvIV (sv);
522 U32 u;
523
524 need (enc, 6); 913 need (enc, IVUV_MAXCHARS);
525 914 enc->cur +=
526 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0; 915 SvIsUV(sv)
527 u = i < 0 ? -i : i; 916 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
528 917 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
529 // convert to 4.28 fixed-point representation
530 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
531
532 char digit, nz = 0;
533
534 digit = u >> 28; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0xfffffff) * 5;
535 digit = u >> 27; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0x7ffffff) * 5;
536 digit = u >> 26; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0x3ffffff) * 5;
537 digit = u >> 25; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0x1ffffff) * 5;
538 digit = u >> 24; *enc->cur = digit + '0'; nz |= digit; enc->cur += 1;
539 } 918 }
540 } 919 }
541 else if (SvROK (sv)) 920 else if (SvROK (sv))
542 encode_rv (enc, SvRV (sv)); 921 encode_rv (enc, SvRV (sv));
543 else if (!SvOK (sv)) 922 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
544 encode_str (enc, "null", 4, 0); 923 encode_str (enc, "null", 4, 0);
545 else 924 else
546 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 925 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
547 SvPV_nolen (sv), SvFLAGS (sv)); 926 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
548} 927}
549 928
550static SV * 929static SV *
551encode_json (SV *scalar, U32 flags) 930encode_json (SV *scalar, JSON *json)
552{ 931{
553 enc_t enc; 932 enc_t enc;
554 933
555 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 934 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
556 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 935 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
557 936
558 enc.flags = flags; 937 enc.json = *json;
559 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 938 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
560 enc.cur = SvPVX (enc.sv); 939 enc.cur = SvPVX (enc.sv);
561 enc.end = SvEND (enc.sv); 940 enc.end = SvEND (enc.sv);
562 enc.indent = 0; 941 enc.indent = 0;
563 enc.maxdepth = DEC_DEPTH (flags); 942 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
943 : enc.json.flags & F_LATIN1 ? 0x000100UL
944 : 0x110000UL;
564 945
565 SvPOK_only (enc.sv); 946 SvPOK_only (enc.sv);
566 encode_sv (&enc, scalar); 947 encode_sv (&enc, scalar);
948 encode_nl (&enc);
567 949
568 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 950 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
569 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 951 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
570 952
571 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 953 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
572 SvUTF8_on (enc.sv); 954 SvUTF8_on (enc.sv);
573 955
574 if (enc.flags & F_SHRINK) 956 if (enc.json.flags & F_SHRINK)
575 shrink (enc.sv); 957 shrink (enc.sv);
576 958
577 return enc.sv; 959 return enc.sv;
578} 960}
579 961
… …
584typedef struct 966typedef struct
585{ 967{
586 char *cur; // current parser pointer 968 char *cur; // current parser pointer
587 char *end; // end of input string 969 char *end; // end of input string
588 const char *err; // parse error, if != 0 970 const char *err; // parse error, if != 0
589 U32 flags; // F_* 971 JSON json;
590 U32 depth; // recursion depth 972 U32 depth; // recursion depth
591 U32 maxdepth; // recursion depth limit 973 U32 maxdepth; // recursion depth limit
592} dec_t; 974} dec_t;
593 975
594inline void 976INLINE void
977decode_comment (dec_t *dec)
978{
979 // only '#'-style comments allowed a.t.m.
980
981 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
982 ++dec->cur;
983}
984
985INLINE void
595decode_ws (dec_t *dec) 986decode_ws (dec_t *dec)
596{ 987{
597 for (;;) 988 for (;;)
598 { 989 {
599 char ch = *dec->cur; 990 char ch = *dec->cur;
600 991
601 if (ch > 0x20 992 if (ch > 0x20)
993 {
994 if (expect_false (ch == '#'))
995 {
996 if (dec->json.flags & F_RELAXED)
997 decode_comment (dec);
998 else
999 break;
1000 }
1001 else
1002 break;
1003 }
602 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 1004 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
603 break; 1005 break; // parse error, but let higher level handle it, gives better error messages
604 1006
605 ++dec->cur; 1007 ++dec->cur;
606 } 1008 }
607} 1009}
608 1010
… …
612 if (*dec->cur != ch) \ 1014 if (*dec->cur != ch) \
613 ERR (# ch " expected"); \ 1015 ERR (# ch " expected"); \
614 ++dec->cur; \ 1016 ++dec->cur; \
615 SE 1017 SE
616 1018
617#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 1019#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
618#define DEC_DEC_DEPTH --dec->depth 1020#define DEC_DEC_DEPTH --dec->depth
619 1021
620static SV *decode_sv (dec_t *dec); 1022static SV *decode_sv (dec_t *dec);
621 1023
622static signed char decode_hexdigit[256]; 1024static signed char decode_hexdigit[256];
… …
625decode_4hex (dec_t *dec) 1027decode_4hex (dec_t *dec)
626{ 1028{
627 signed char d1, d2, d3, d4; 1029 signed char d1, d2, d3, d4;
628 unsigned char *cur = (unsigned char *)dec->cur; 1030 unsigned char *cur = (unsigned char *)dec->cur;
629 1031
630 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected"); 1032 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
631 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected"); 1033 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
632 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected"); 1034 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
633 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected"); 1035 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
634 1036
635 dec->cur += 4; 1037 dec->cur += 4;
636 1038
637 return ((UV)d1) << 12 1039 return ((UV)d1) << 12
638 | ((UV)d2) << 8 1040 | ((UV)d2) << 8
… …
646static SV * 1048static SV *
647decode_str (dec_t *dec) 1049decode_str (dec_t *dec)
648{ 1050{
649 SV *sv = 0; 1051 SV *sv = 0;
650 int utf8 = 0; 1052 int utf8 = 0;
1053 char *dec_cur = dec->cur;
651 1054
652 do 1055 do
653 { 1056 {
654 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; 1057 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
655 char *cur = buf; 1058 char *cur = buf;
656 1059
657 do 1060 do
658 { 1061 {
659 unsigned char ch = *(unsigned char *)dec->cur++; 1062 unsigned char ch = *(unsigned char *)dec_cur++;
660 1063
661 if (expect_false (ch == '"')) 1064 if (expect_false (ch == '"'))
662 { 1065 {
663 --dec->cur; 1066 --dec_cur;
664 break; 1067 break;
665 } 1068 }
666 else if (expect_false (ch == '\\')) 1069 else if (expect_false (ch == '\\'))
667 { 1070 {
668 switch (*dec->cur) 1071 switch (*dec_cur)
669 { 1072 {
670 case '\\': 1073 case '\\':
671 case '/': 1074 case '/':
672 case '"': *cur++ = *dec->cur++; break; 1075 case '"': *cur++ = *dec_cur++; break;
673 1076
674 case 'b': ++dec->cur; *cur++ = '\010'; break; 1077 case 'b': ++dec_cur; *cur++ = '\010'; break;
675 case 't': ++dec->cur; *cur++ = '\011'; break; 1078 case 't': ++dec_cur; *cur++ = '\011'; break;
676 case 'n': ++dec->cur; *cur++ = '\012'; break; 1079 case 'n': ++dec_cur; *cur++ = '\012'; break;
677 case 'f': ++dec->cur; *cur++ = '\014'; break; 1080 case 'f': ++dec_cur; *cur++ = '\014'; break;
678 case 'r': ++dec->cur; *cur++ = '\015'; break; 1081 case 'r': ++dec_cur; *cur++ = '\015'; break;
679 1082
680 case 'u': 1083 case 'u':
681 { 1084 {
682 UV lo, hi; 1085 UV lo, hi;
683 ++dec->cur; 1086 ++dec_cur;
684 1087
1088 dec->cur = dec_cur;
685 hi = decode_4hex (dec); 1089 hi = decode_4hex (dec);
1090 dec_cur = dec->cur;
686 if (hi == (UV)-1) 1091 if (hi == (UV)-1)
687 goto fail; 1092 goto fail;
688 1093
689 // possibly a surrogate pair 1094 // possibly a surrogate pair
690 if (hi >= 0xd800) 1095 if (hi >= 0xd800)
691 if (hi < 0xdc00) 1096 if (hi < 0xdc00)
692 { 1097 {
693 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 1098 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
694 ERR ("missing low surrogate character in surrogate pair"); 1099 ERR ("missing low surrogate character in surrogate pair");
695 1100
696 dec->cur += 2; 1101 dec_cur += 2;
697 1102
1103 dec->cur = dec_cur;
698 lo = decode_4hex (dec); 1104 lo = decode_4hex (dec);
1105 dec_cur = dec->cur;
699 if (lo == (UV)-1) 1106 if (lo == (UV)-1)
700 goto fail; 1107 goto fail;
701 1108
702 if (lo < 0xdc00 || lo >= 0xe000) 1109 if (lo < 0xdc00 || lo >= 0xe000)
703 ERR ("surrogate pair expected"); 1110 ERR ("surrogate pair expected");
… …
709 1116
710 if (hi >= 0x80) 1117 if (hi >= 0x80)
711 { 1118 {
712 utf8 = 1; 1119 utf8 = 1;
713 1120
714 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1121 cur = encode_utf8 (cur, hi);
715 } 1122 }
716 else 1123 else
717 *cur++ = hi; 1124 *cur++ = hi;
718 } 1125 }
719 break; 1126 break;
720 1127
721 default: 1128 default:
722 --dec->cur; 1129 --dec_cur;
723 ERR ("illegal backslash escape sequence in string"); 1130 ERR ("illegal backslash escape sequence in string");
724 } 1131 }
725 } 1132 }
726 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1133 else if (expect_true (ch >= 0x20 && ch < 0x80))
727 *cur++ = ch; 1134 *cur++ = ch;
728 else if (ch >= 0x80) 1135 else if (ch >= 0x80)
729 { 1136 {
730 STRLEN clen; 1137 STRLEN clen;
731 UV uch;
732 1138
733 --dec->cur; 1139 --dec_cur;
734 1140
735 uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 1141 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
736 if (clen == (STRLEN)-1) 1142 if (clen == (STRLEN)-1)
737 ERR ("malformed UTF-8 character in JSON string"); 1143 ERR ("malformed UTF-8 character in JSON string");
738 1144
739 do 1145 do
740 *cur++ = *dec->cur++; 1146 *cur++ = *dec_cur++;
741 while (--clen); 1147 while (--clen);
742 1148
743 utf8 = 1; 1149 utf8 = 1;
744 } 1150 }
1151 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1152 *cur++ = ch;
745 else 1153 else
746 { 1154 {
747 --dec->cur; 1155 --dec_cur;
748 1156
749 if (!ch) 1157 if (!ch)
750 ERR ("unexpected end of string while parsing JSON string"); 1158 ERR ("unexpected end of string while parsing JSON string");
751 else 1159 else
752 ERR ("invalid character encountered while parsing JSON string"); 1160 ERR ("invalid character encountered while parsing JSON string");
… …
757 { 1165 {
758 STRLEN len = cur - buf; 1166 STRLEN len = cur - buf;
759 1167
760 if (sv) 1168 if (sv)
761 { 1169 {
762 SvGROW (sv, SvCUR (sv) + len + 1); 1170 STRLEN cur = SvCUR (sv);
1171
1172 if (SvLEN (sv) <= cur + len)
1173 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1174
763 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1175 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
764 SvCUR_set (sv, SvCUR (sv) + len); 1176 SvCUR_set (sv, SvCUR (sv) + len);
765 } 1177 }
766 else 1178 else
767 sv = newSVpvn (buf, len); 1179 sv = newSVpvn (buf, len);
768 } 1180 }
769 } 1181 }
770 while (*dec->cur != '"'); 1182 while (*dec_cur != '"');
771 1183
772 ++dec->cur; 1184 ++dec_cur;
773 1185
774 if (sv) 1186 if (sv)
775 { 1187 {
776 SvPOK_only (sv); 1188 SvPOK_only (sv);
777 *SvEND (sv) = 0; 1189 *SvEND (sv) = 0;
… …
780 SvUTF8_on (sv); 1192 SvUTF8_on (sv);
781 } 1193 }
782 else 1194 else
783 sv = newSVpvn ("", 0); 1195 sv = newSVpvn ("", 0);
784 1196
1197 dec->cur = dec_cur;
785 return sv; 1198 return sv;
786 1199
787fail: 1200fail:
1201 dec->cur = dec_cur;
788 return 0; 1202 return 0;
789} 1203}
790 1204
791static SV * 1205static SV *
792decode_num (dec_t *dec) 1206decode_num (dec_t *dec)
… …
850 is_nv = 1; 1264 is_nv = 1;
851 } 1265 }
852 1266
853 if (!is_nv) 1267 if (!is_nv)
854 { 1268 {
1269 int len = dec->cur - start;
1270
855 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1271 // special case the rather common 1..5-digit-int case
856 if (*start == '-') 1272 if (*start == '-')
857 switch (dec->cur - start) 1273 switch (len)
858 { 1274 {
859 case 2: return newSViv (-( start [1] - '0' )); 1275 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
860 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1276 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
861 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1277 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
862 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1278 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1279 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
863 } 1280 }
864 else 1281 else
865 switch (dec->cur - start) 1282 switch (len)
866 { 1283 {
867 case 1: return newSViv ( start [0] - '0' ); 1284 case 1: return newSViv ( start [0] - '0' * 1);
868 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1285 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
869 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1286 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
870 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1287 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1288 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
871 } 1289 }
872 1290
873 { 1291 {
874 UV uv; 1292 UV uv;
875 int numtype = grok_number (start, dec->cur - start, &uv); 1293 int numtype = grok_number (start, len, &uv);
876 if (numtype & IS_NUMBER_IN_UV) 1294 if (numtype & IS_NUMBER_IN_UV)
877 if (numtype & IS_NUMBER_NEG) 1295 if (numtype & IS_NUMBER_NEG)
878 { 1296 {
879 if (uv < (UV)IV_MIN) 1297 if (uv < (UV)IV_MIN)
880 return newSViv (-(IV)uv); 1298 return newSViv (-(IV)uv);
881 } 1299 }
882 else 1300 else
883 return newSVuv (uv); 1301 return newSVuv (uv);
884 } 1302 }
885 }
886 1303
1304 len -= *start == '-' ? 1 : 0;
1305
1306 // does not fit into IV or UV, try NV
1307 if (len <= NV_DIG)
1308 // fits into NV without loss of precision
1309 return newSVnv (json_atof (start));
1310
1311 // everything else fails, convert it to a string
1312 return newSVpvn (start, dec->cur - start);
1313 }
1314
1315 // loss of precision here
887 return newSVnv (Atof (start)); 1316 return newSVnv (json_atof (start));
888 1317
889fail: 1318fail:
890 return 0; 1319 return 0;
891} 1320}
892 1321
… …
916 if (*dec->cur == ']') 1345 if (*dec->cur == ']')
917 { 1346 {
918 ++dec->cur; 1347 ++dec->cur;
919 break; 1348 break;
920 } 1349 }
921 1350
922 if (*dec->cur != ',') 1351 if (*dec->cur != ',')
923 ERR (", or ] expected while parsing array"); 1352 ERR (", or ] expected while parsing array");
924 1353
925 ++dec->cur; 1354 ++dec->cur;
1355
1356 decode_ws (dec);
1357
1358 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1359 {
1360 ++dec->cur;
1361 break;
1362 }
926 } 1363 }
927 1364
928 DEC_DEC_DEPTH; 1365 DEC_DEC_DEPTH;
929 return newRV_noinc ((SV *)av); 1366 return newRV_noinc ((SV *)av);
930 1367
… …
935} 1372}
936 1373
937static SV * 1374static SV *
938decode_hv (dec_t *dec) 1375decode_hv (dec_t *dec)
939{ 1376{
1377 SV *sv;
940 HV *hv = newHV (); 1378 HV *hv = newHV ();
941 1379
942 DEC_INC_DEPTH; 1380 DEC_INC_DEPTH;
943 decode_ws (dec); 1381 decode_ws (dec);
944 1382
945 if (*dec->cur == '}') 1383 if (*dec->cur == '}')
946 ++dec->cur; 1384 ++dec->cur;
947 else 1385 else
948 for (;;) 1386 for (;;)
949 { 1387 {
1388 EXPECT_CH ('"');
1389
1390 // heuristic: assume that
1391 // a) decode_str + hv_store_ent are abysmally slow.
1392 // b) most hash keys are short, simple ascii text.
1393 // => try to "fast-match" such strings to avoid
1394 // the overhead of decode_str + hv_store_ent.
1395 {
950 SV *key, *value; 1396 SV *value;
1397 char *p = dec->cur;
1398 char *e = p + 24; // only try up to 24 bytes
951 1399
952 decode_ws (dec); EXPECT_CH ('"'); 1400 for (;;)
953
954 key = decode_str (dec);
955 if (!key)
956 goto fail;
957
958 decode_ws (dec); EXPECT_CH (':');
959
960 value = decode_sv (dec);
961 if (!value)
962 { 1401 {
1402 // the >= 0x80 is false on most architectures
1403 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1404 {
1405 // slow path, back up and use decode_str
1406 SV *key = decode_str (dec);
1407 if (!key)
1408 goto fail;
1409
1410 decode_ws (dec); EXPECT_CH (':');
1411
1412 decode_ws (dec);
1413 value = decode_sv (dec);
1414 if (!value)
1415 {
1416 SvREFCNT_dec (key);
1417 goto fail;
1418 }
1419
1420 hv_store_ent (hv, key, value, 0);
963 SvREFCNT_dec (key); 1421 SvREFCNT_dec (key);
1422
1423 break;
1424 }
1425 else if (*p == '"')
1426 {
1427 // fast path, got a simple key
1428 char *key = dec->cur;
1429 int len = p - key;
1430 dec->cur = p + 1;
1431
1432 decode_ws (dec); EXPECT_CH (':');
1433
1434 decode_ws (dec);
1435 value = decode_sv (dec);
1436 if (!value)
964 goto fail; 1437 goto fail;
1438
1439 hv_store (hv, key, len, value, 0);
1440
1441 break;
1442 }
1443
1444 ++p;
965 } 1445 }
966 1446 }
967 hv_store_ent (hv, key, value, 0);
968 SvREFCNT_dec (key);
969 1447
970 decode_ws (dec); 1448 decode_ws (dec);
971 1449
972 if (*dec->cur == '}') 1450 if (*dec->cur == '}')
973 { 1451 {
… …
977 1455
978 if (*dec->cur != ',') 1456 if (*dec->cur != ',')
979 ERR (", or } expected while parsing object/hash"); 1457 ERR (", or } expected while parsing object/hash");
980 1458
981 ++dec->cur; 1459 ++dec->cur;
1460
1461 decode_ws (dec);
1462
1463 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1464 {
1465 ++dec->cur;
1466 break;
1467 }
982 } 1468 }
983 1469
984 DEC_DEC_DEPTH; 1470 DEC_DEC_DEPTH;
985 return newRV_noinc ((SV *)hv); 1471 sv = newRV_noinc ((SV *)hv);
1472
1473 // check filter callbacks
1474 if (dec->json.flags & F_HOOK)
1475 {
1476 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1477 {
1478 HE *cb, *he;
1479
1480 hv_iterinit (hv);
1481 he = hv_iternext (hv);
1482 hv_iterinit (hv);
1483
1484 // the next line creates a mortal sv each time it's called.
1485 // might want to optimise this for common cases.
1486 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1487
1488 if (cb)
1489 {
1490 dSP;
1491 int count;
1492
1493 ENTER; SAVETMPS;
1494 SAVESTACK_POS ();
1495 PUSHMARK (SP);
1496 XPUSHs (HeVAL (he));
1497 sv_2mortal (sv);
1498
1499 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1500
1501 if (count == 1)
1502 {
1503 sv = newSVsv (POPs);
1504 FREETMPS; LEAVE;
1505 return sv;
1506 }
1507
1508 SvREFCNT_inc (sv);
1509 FREETMPS; LEAVE;
1510 }
1511 }
1512
1513 if (dec->json.cb_object)
1514 {
1515 dSP;
1516 int count;
1517
1518 ENTER; SAVETMPS;
1519 SAVESTACK_POS ();
1520 PUSHMARK (SP);
1521 XPUSHs (sv_2mortal (sv));
1522
1523 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1524
1525 if (count == 1)
1526 {
1527 sv = newSVsv (POPs);
1528 FREETMPS; LEAVE;
1529 return sv;
1530 }
1531
1532 SvREFCNT_inc (sv);
1533 FREETMPS; LEAVE;
1534 }
1535 }
1536
1537 return sv;
986 1538
987fail: 1539fail:
988 SvREFCNT_dec (hv); 1540 SvREFCNT_dec (hv);
989 DEC_DEC_DEPTH; 1541 DEC_DEC_DEPTH;
990 return 0; 1542 return 0;
991} 1543}
992 1544
993static SV * 1545static SV *
1546decode_tag (dec_t *dec)
1547{
1548 SV *tag = 0;
1549 SV *val = 0;
1550
1551 if (!(dec->json.flags & F_ALLOW_TAGS))
1552 ERR ("malformed JSON string, neither array, object, number, string or atom");
1553
1554 ++dec->cur;
1555
1556 decode_ws (dec);
1557
1558 tag = decode_sv (dec);
1559 if (!tag)
1560 goto fail;
1561
1562 if (!SvPOK (tag))
1563 ERR ("malformed JSON string, (tag) must be a string");
1564
1565 decode_ws (dec);
1566
1567 if (*dec->cur != ')')
1568 ERR (") expected after tag");
1569
1570 ++dec->cur;
1571
1572 decode_ws (dec);
1573
1574 val = decode_sv (dec);
1575 if (!val)
1576 goto fail;
1577
1578 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1579 ERR ("malformed JSON string, tag value must be an array");
1580
1581 {
1582 AV *av = (AV *)SvRV (val);
1583 int i, len = av_len (av) + 1;
1584 HV *stash = gv_stashsv (tag, 0);
1585 SV *sv;
1586
1587 if (!stash)
1588 ERR ("cannot decode perl-object (package does not exist)");
1589
1590 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1591
1592 if (!method)
1593 ERR ("cannot decode perl-object (package does not have a THAW method)");
1594
1595 dSP;
1596
1597 ENTER; SAVETMPS;
1598 PUSHMARK (SP);
1599 EXTEND (SP, len + 2);
1600 // we re-bless the reference to get overload and other niceties right
1601 PUSHs (tag);
1602 PUSHs (sv_json);
1603
1604 for (i = 0; i < len; ++i)
1605 PUSHs (*av_fetch (av, i, 1));
1606
1607 PUTBACK;
1608 call_sv ((SV *)GvCV (method), G_SCALAR);
1609 SPAGAIN;
1610
1611 SvREFCNT_dec (tag);
1612 SvREFCNT_dec (val);
1613 sv = SvREFCNT_inc (POPs);
1614
1615 PUTBACK;
1616
1617 FREETMPS; LEAVE;
1618
1619 return sv;
1620 }
1621
1622fail:
1623 SvREFCNT_dec (tag);
1624 SvREFCNT_dec (val);
1625 return 0;
1626}
1627
1628static SV *
994decode_sv (dec_t *dec) 1629decode_sv (dec_t *dec)
995{ 1630{
996 decode_ws (dec); 1631 // the beauty of JSON: you need exactly one character lookahead
1632 // to parse everything.
997 switch (*dec->cur) 1633 switch (*dec->cur)
998 { 1634 {
999 case '"': ++dec->cur; return decode_str (dec); 1635 case '"': ++dec->cur; return decode_str (dec);
1000 case '[': ++dec->cur; return decode_av (dec); 1636 case '[': ++dec->cur; return decode_av (dec);
1001 case '{': ++dec->cur; return decode_hv (dec); 1637 case '{': ++dec->cur; return decode_hv (dec);
1638 case '(': return decode_tag (dec);
1002 1639
1003 case '-': 1640 case '-':
1004 case '0': case '1': case '2': case '3': case '4': 1641 case '0': case '1': case '2': case '3': case '4':
1005 case '5': case '6': case '7': case '8': case '9': 1642 case '5': case '6': case '7': case '8': case '9':
1006 return decode_num (dec); 1643 return decode_num (dec);
1007 1644
1008 case 't': 1645 case 't':
1009 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1646 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1010 { 1647 {
1011 dec->cur += 4; 1648 dec->cur += 4;
1649#if JSON_SLOW
1650 bool_true = get_bool ("Types::Serialiser::true");
1651#endif
1012 return newSViv (1); 1652 return newSVsv (bool_true);
1013 } 1653 }
1014 else 1654 else
1015 ERR ("'true' expected"); 1655 ERR ("'true' expected");
1016 1656
1017 break; 1657 break;
1018 1658
1019 case 'f': 1659 case 'f':
1020 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1660 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1021 { 1661 {
1022 dec->cur += 5; 1662 dec->cur += 5;
1663#if JSON_SLOW
1664 bool_false = get_bool ("Types::Serialiser::false");
1665#endif
1023 return newSViv (0); 1666 return newSVsv (bool_false);
1024 } 1667 }
1025 else 1668 else
1026 ERR ("'false' expected"); 1669 ERR ("'false' expected");
1027 1670
1028 break; 1671 break;
… …
1037 ERR ("'null' expected"); 1680 ERR ("'null' expected");
1038 1681
1039 break; 1682 break;
1040 1683
1041 default: 1684 default:
1042 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1685 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1043 break; 1686 break;
1044 } 1687 }
1045 1688
1046fail: 1689fail:
1047 return 0; 1690 return 0;
1048} 1691}
1049 1692
1050static SV * 1693static SV *
1051decode_json (SV *string, U32 flags, UV *offset_return) 1694decode_json (SV *string, JSON *json, STRLEN *offset_return)
1052{ 1695{
1053 dec_t dec; 1696 dec_t dec;
1054 UV offset;
1055 SV *sv; 1697 SV *sv;
1056 1698
1699 /* work around bugs in 5.10 where manipulating magic values
1700 * makes perl ignore the magic in subsequent accesses.
1701 * also make a copy of non-PV values, to get them into a clean
1702 * state (SvPV should do that, but it's buggy, see below).
1703 *
1704 * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
1705 * as reported by Reini Urban.
1706 */
1057 SvGETMAGIC (string); 1707 /*SvGETMAGIC (string);*/
1708 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1709 string = sv_2mortal (newSVsv (string));
1710
1058 SvUPGRADE (string, SVt_PV); 1711 SvUPGRADE (string, SVt_PV);
1059 1712
1060 if (flags & F_UTF8) 1713 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1714 * assertion with -DDEBUGGING, although SvCUR is documented to
1715 * return the xpv_cur field which certainly exists after upgrading.
1716 * according to nicholas clark, calling SvPOK fixes this.
1717 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1718 * and hope for the best.
1719 * Damnit, SvPV_nolen still trips over yet another assertion. This
1720 * assertion business is seriously broken, try yet another workaround
1721 * for the broken -DDEBUGGING.
1722 */
1723 {
1724#ifdef DEBUGGING
1725 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1726#else
1727 STRLEN offset = SvCUR (string);
1728#endif
1729
1730 if (offset > json->max_size && json->max_size)
1731 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1732 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1733 }
1734
1735 if (DECODE_WANTS_OCTETS (json))
1061 sv_utf8_downgrade (string, 0); 1736 sv_utf8_downgrade (string, 0);
1062 else 1737 else
1063 sv_utf8_upgrade (string); 1738 sv_utf8_upgrade (string);
1064 1739
1065 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1740 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1066 1741
1067 dec.flags = flags; 1742 dec.json = *json;
1068 dec.cur = SvPVX (string); 1743 dec.cur = SvPVX (string);
1069 dec.end = SvEND (string); 1744 dec.end = SvEND (string);
1070 dec.err = 0; 1745 dec.err = 0;
1071 dec.depth = 0; 1746 dec.depth = 0;
1072 dec.maxdepth = DEC_DEPTH (dec.flags); 1747
1748 if (dec.json.cb_object || dec.json.cb_sk_object)
1749 dec.json.flags |= F_HOOK;
1073 1750
1074 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1751 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1752
1753 decode_ws (&dec);
1075 sv = decode_sv (&dec); 1754 sv = decode_sv (&dec);
1076 1755
1077 if (!(offset_return || !sv)) 1756 if (offset_return)
1757 *offset_return = dec.cur - SvPVX (string);
1758 else if (sv)
1078 { 1759 {
1079 // check for trailing garbage 1760 // check for trailing garbage
1080 decode_ws (&dec); 1761 decode_ws (&dec);
1081 1762
1082 if (*dec.cur) 1763 if (*dec.cur)
1083 { 1764 {
1084 dec.err = "garbage after JSON object"; 1765 dec.err = "garbage after JSON object";
1085 SvREFCNT_dec (sv); 1766 SvREFCNT_dec (sv);
1086 sv = 0; 1767 sv = 0;
1087 } 1768 }
1088 }
1089
1090 if (offset_return || !sv)
1091 {
1092 offset = dec.flags & F_UTF8
1093 ? dec.cur - SvPVX (string)
1094 : utf8_distance (dec.cur, SvPVX (string));
1095
1096 if (offset_return)
1097 *offset_return = offset;
1098 } 1769 }
1099 1770
1100 if (!sv) 1771 if (!sv)
1101 { 1772 {
1102 SV *uni = sv_newmortal (); 1773 SV *uni = sv_newmortal ();
… …
1108 SAVEVPTR (PL_curcop); 1779 SAVEVPTR (PL_curcop);
1109 PL_curcop = &cop; 1780 PL_curcop = &cop;
1110 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1781 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1111 LEAVE; 1782 LEAVE;
1112 1783
1113 croak ("%s, at character offset %d [\"%s\"]", 1784 croak ("%s, at character offset %d (before \"%s\")",
1114 dec.err, 1785 dec.err,
1115 (int)offset, 1786 (int)ptr_to_index (string, dec.cur),
1116 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1787 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1117 } 1788 }
1118 1789
1119 sv = sv_2mortal (sv); 1790 sv = sv_2mortal (sv);
1120 1791
1121 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1792 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1122 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1793 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1123 1794
1124 return sv; 1795 return sv;
1796}
1797
1798/////////////////////////////////////////////////////////////////////////////
1799// incremental parser
1800
1801static void
1802incr_parse (JSON *self)
1803{
1804 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1805
1806 // the state machine here is a bit convoluted and could be simplified a lot
1807 // but this would make it slower, so...
1808
1809 for (;;)
1810 {
1811 //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D
1812 switch (self->incr_mode)
1813 {
1814 // only used for initial whitespace skipping
1815 case INCR_M_WS:
1816 for (;;)
1817 {
1818 if (*p > 0x20)
1819 {
1820 if (*p == '#')
1821 {
1822 self->incr_mode = INCR_M_C0;
1823 goto incr_m_c;
1824 }
1825 else
1826 {
1827 self->incr_mode = INCR_M_JSON;
1828 goto incr_m_json;
1829 }
1830 }
1831 else if (!*p)
1832 goto interrupt;
1833
1834 ++p;
1835 }
1836
1837 // skip a single char inside a string (for \\-processing)
1838 case INCR_M_BS:
1839 if (!*p)
1840 goto interrupt;
1841
1842 ++p;
1843 self->incr_mode = INCR_M_STR;
1844 goto incr_m_str;
1845
1846 // inside #-style comments
1847 case INCR_M_C0:
1848 case INCR_M_C1:
1849 incr_m_c:
1850 for (;;)
1851 {
1852 if (*p == '\n')
1853 {
1854 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1855 break;
1856 }
1857 else if (!*p)
1858 goto interrupt;
1859
1860 ++p;
1861 }
1862
1863 break;
1864
1865 // inside a string
1866 case INCR_M_STR:
1867 incr_m_str:
1868 for (;;)
1869 {
1870 if (*p == '"')
1871 {
1872 ++p;
1873 self->incr_mode = INCR_M_JSON;
1874
1875 if (!self->incr_nest)
1876 goto interrupt;
1877
1878 goto incr_m_json;
1879 }
1880 else if (*p == '\\')
1881 {
1882 ++p; // "virtually" consumes character after \
1883
1884 if (!*p) // if at end of string we have to switch modes
1885 {
1886 self->incr_mode = INCR_M_BS;
1887 goto interrupt;
1888 }
1889 }
1890 else if (!*p)
1891 goto interrupt;
1892
1893 ++p;
1894 }
1895
1896 // after initial ws, outside string
1897 case INCR_M_JSON:
1898 incr_m_json:
1899 for (;;)
1900 {
1901 switch (*p++)
1902 {
1903 case 0:
1904 --p;
1905 goto interrupt;
1906
1907 case 0x09:
1908 case 0x0a:
1909 case 0x0d:
1910 case 0x20:
1911 if (!self->incr_nest)
1912 {
1913 --p; // do not eat the whitespace, let the next round do it
1914 goto interrupt;
1915 }
1916 break;
1917
1918 case '"':
1919 self->incr_mode = INCR_M_STR;
1920 goto incr_m_str;
1921
1922 case '[':
1923 case '{':
1924 case '(':
1925 if (++self->incr_nest > self->max_depth)
1926 croak (ERR_NESTING_EXCEEDED);
1927 break;
1928
1929 case ']':
1930 case '}':
1931 if (--self->incr_nest <= 0)
1932 goto interrupt;
1933 break;
1934
1935 case ')':
1936 --self->incr_nest;
1937 break;
1938
1939 case '#':
1940 self->incr_mode = INCR_M_C1;
1941 goto incr_m_c;
1942 }
1943 }
1944 }
1945
1946 modechange:
1947 ;
1948 }
1949
1950interrupt:
1951 self->incr_pos = p - SvPVX (self->incr_text);
1952 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1953 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1125} 1954}
1126 1955
1127///////////////////////////////////////////////////////////////////////////// 1956/////////////////////////////////////////////////////////////////////////////
1128// XS interface functions 1957// XS interface functions
1129 1958
1130MODULE = JSON::XS PACKAGE = JSON::XS 1959MODULE = JSON::XS PACKAGE = JSON::XS
1131 1960
1132BOOT: 1961BOOT:
1133{ 1962{
1134 int i; 1963 int i;
1135
1136 memset (decode_hexdigit, 0xff, 256);
1137 1964
1138 for (i = 0; i < 256; ++i) 1965 for (i = 0; i < 256; ++i)
1139 decode_hexdigit [i] = 1966 decode_hexdigit [i] =
1140 i >= '0' && i <= '9' ? i - '0' 1967 i >= '0' && i <= '9' ? i - '0'
1141 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1968 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1142 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1969 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1143 : -1; 1970 : -1;
1144 1971
1145 json_stash = gv_stashpv ("JSON::XS", 1); 1972 json_stash = gv_stashpv ("JSON::XS" , 1);
1973 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1974 bool_true = get_bool ("Types::Serialiser::true");
1975 bool_false = get_bool ("Types::Serialiser::false");
1976
1977 sv_json = newSVpv ("JSON", 0);
1978 SvREADONLY_on (sv_json);
1979
1980 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1146} 1981}
1147 1982
1148PROTOTYPES: DISABLE 1983PROTOTYPES: DISABLE
1149 1984
1150SV *new (char *dummy) 1985void CLONE (...)
1151 CODE: 1986 CODE:
1152 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1987 json_stash = 0;
1988 bool_stash = 0;
1989
1990void new (char *klass)
1991 PPCODE:
1992{
1993 SV *pv = NEWSV (0, sizeof (JSON));
1994 SvPOK_only (pv);
1995 json_init ((JSON *)SvPVX (pv));
1996 XPUSHs (sv_2mortal (sv_bless (
1997 newRV_noinc (pv),
1998 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1999 )));
2000}
2001
2002void ascii (JSON *self, int enable = 1)
2003 ALIAS:
2004 ascii = F_ASCII
2005 latin1 = F_LATIN1
2006 utf8 = F_UTF8
2007 indent = F_INDENT
2008 canonical = F_CANONICAL
2009 space_before = F_SPACE_BEFORE
2010 space_after = F_SPACE_AFTER
2011 pretty = F_PRETTY
2012 allow_nonref = F_ALLOW_NONREF
2013 shrink = F_SHRINK
2014 allow_blessed = F_ALLOW_BLESSED
2015 convert_blessed = F_CONV_BLESSED
2016 relaxed = F_RELAXED
2017 allow_unknown = F_ALLOW_UNKNOWN
2018 allow_tags = F_ALLOW_TAGS
2019 PPCODE:
2020{
2021 if (enable)
2022 self->flags |= ix;
2023 else
2024 self->flags &= ~ix;
2025
2026 XPUSHs (ST (0));
2027}
2028
2029void get_ascii (JSON *self)
2030 ALIAS:
2031 get_ascii = F_ASCII
2032 get_latin1 = F_LATIN1
2033 get_utf8 = F_UTF8
2034 get_indent = F_INDENT
2035 get_canonical = F_CANONICAL
2036 get_space_before = F_SPACE_BEFORE
2037 get_space_after = F_SPACE_AFTER
2038 get_allow_nonref = F_ALLOW_NONREF
2039 get_shrink = F_SHRINK
2040 get_allow_blessed = F_ALLOW_BLESSED
2041 get_convert_blessed = F_CONV_BLESSED
2042 get_relaxed = F_RELAXED
2043 get_allow_unknown = F_ALLOW_UNKNOWN
2044 get_allow_tags = F_ALLOW_TAGS
2045 PPCODE:
2046 XPUSHs (boolSV (self->flags & ix));
2047
2048void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
2049 PPCODE:
2050 self->max_depth = max_depth;
2051 XPUSHs (ST (0));
2052
2053U32 get_max_depth (JSON *self)
2054 CODE:
2055 RETVAL = self->max_depth;
1153 OUTPUT: 2056 OUTPUT:
1154 RETVAL 2057 RETVAL
1155 2058
1156SV *ascii (SV *self, int enable = 1) 2059void max_size (JSON *self, U32 max_size = 0)
1157 ALIAS: 2060 PPCODE:
1158 ascii = F_ASCII 2061 self->max_size = max_size;
1159 latin1 = F_LATIN1 2062 XPUSHs (ST (0));
1160 utf8 = F_UTF8 2063
1161 indent = F_INDENT 2064int get_max_size (JSON *self)
1162 canonical = F_CANONICAL
1163 space_before = F_SPACE_BEFORE
1164 space_after = F_SPACE_AFTER
1165 pretty = F_PRETTY
1166 allow_nonref = F_ALLOW_NONREF
1167 shrink = F_SHRINK
1168 CODE: 2065 CODE:
1169{ 2066 RETVAL = self->max_size;
1170 UV *uv = SvJSON (self);
1171 if (enable)
1172 *uv |= ix;
1173 else
1174 *uv &= ~ix;
1175
1176 RETVAL = newSVsv (self);
1177}
1178 OUTPUT: 2067 OUTPUT:
1179 RETVAL 2068 RETVAL
1180 2069
1181SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 2070void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
2071 PPCODE:
2072{
2073 SvREFCNT_dec (self->cb_object);
2074 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
2075
2076 XPUSHs (ST (0));
2077}
2078
2079void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
2080 PPCODE:
2081{
2082 if (!self->cb_sk_object)
2083 self->cb_sk_object = newHV ();
2084
2085 if (SvOK (cb))
2086 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
2087 else
2088 {
2089 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
2090
2091 if (!HvKEYS (self->cb_sk_object))
2092 {
2093 SvREFCNT_dec (self->cb_sk_object);
2094 self->cb_sk_object = 0;
2095 }
2096 }
2097
2098 XPUSHs (ST (0));
2099}
2100
2101void encode (JSON *self, SV *scalar)
2102 PPCODE:
2103 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2104 XPUSHs (scalar);
2105
2106void decode (JSON *self, SV *jsonstr)
2107 PPCODE:
2108 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2109 XPUSHs (jsonstr);
2110
2111void decode_prefix (JSON *self, SV *jsonstr)
2112 PPCODE:
2113{
2114 SV *sv;
2115 STRLEN offset;
2116 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
2117 EXTEND (SP, 2);
2118 PUSHs (sv);
2119 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
2120}
2121
2122void incr_parse (JSON *self, SV *jsonstr = 0)
2123 PPCODE:
2124{
2125 if (!self->incr_text)
2126 self->incr_text = newSVpvn ("", 0);
2127
2128 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2129 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2130 if (DECODE_WANTS_OCTETS (self))
2131 {
2132 if (self->incr_pos)
2133 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2134 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2135
2136 sv_utf8_downgrade (self->incr_text, 0);
2137 }
2138 else
2139 {
2140 sv_utf8_upgrade (self->incr_text);
2141
2142 if (self->incr_pos)
2143 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2144 - (U8 *)SvPVX (self->incr_text);
2145 }
2146
2147 // append data, if any
2148 if (jsonstr)
2149 {
2150 /* make sure both strings have same encoding */
2151 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2152 if (SvUTF8 (jsonstr))
2153 sv_utf8_downgrade (jsonstr, 0);
2154 else
2155 sv_utf8_upgrade (jsonstr);
2156
2157 /* and then just blindly append */
2158 {
2159 STRLEN len;
2160 const char *str = SvPV (jsonstr, len);
2161 STRLEN cur = SvCUR (self->incr_text);
2162
2163 if (SvLEN (self->incr_text) <= cur + len)
2164 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
2165
2166 Move (str, SvEND (self->incr_text), len, char);
2167 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2168 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2169 }
2170 }
2171
2172 if (GIMME_V != G_VOID)
2173 do
2174 {
2175 SV *sv;
2176 STRLEN offset;
2177
2178 if (!INCR_DONE (self))
2179 {
2180 incr_parse (self);
2181
2182 if (self->incr_pos > self->max_size && self->max_size)
2183 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2184 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2185
2186 if (!INCR_DONE (self))
2187 {
2188 // as an optimisation, do not accumulate white space in the incr buffer
2189 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2190 {
2191 self->incr_pos = 0;
2192 SvCUR_set (self->incr_text, 0);
2193 }
2194
2195 break;
2196 }
2197 }
2198
2199 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2200 XPUSHs (sv);
2201
2202 self->incr_pos -= offset;
2203 self->incr_nest = 0;
2204 self->incr_mode = 0;
2205
2206 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
2207 }
2208 while (GIMME_V == G_ARRAY);
2209}
2210
2211SV *incr_text (JSON *self)
2212 ATTRS: lvalue
1182 CODE: 2213 CODE:
1183{ 2214{
1184 UV *uv = SvJSON (self); 2215 if (self->incr_pos)
1185 UV log2 = 0; 2216 croak ("incr_text can not be called when the incremental parser already started parsing");
1186 2217
1187 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 2218 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1188
1189 while ((1UL << log2) < max_depth)
1190 ++log2;
1191
1192 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1193
1194 RETVAL = newSVsv (self);
1195} 2219}
1196 OUTPUT: 2220 OUTPUT:
1197 RETVAL 2221 RETVAL
1198 2222
2223void incr_skip (JSON *self)
2224 CODE:
2225{
2226 if (self->incr_pos)
2227 {
2228 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2229 self->incr_pos = 0;
2230 self->incr_nest = 0;
2231 self->incr_mode = 0;
2232 }
2233}
2234
2235void incr_reset (JSON *self)
2236 CODE:
2237{
2238 SvREFCNT_dec (self->incr_text);
2239 self->incr_text = 0;
2240 self->incr_pos = 0;
2241 self->incr_nest = 0;
2242 self->incr_mode = 0;
2243}
2244
2245void DESTROY (JSON *self)
2246 CODE:
2247 SvREFCNT_dec (self->cb_sk_object);
2248 SvREFCNT_dec (self->cb_object);
2249 SvREFCNT_dec (self->incr_text);
2250
2251PROTOTYPES: ENABLE
2252
1199void encode (SV *self, SV *scalar) 2253void encode_json (SV *scalar)
1200 PPCODE: 2254 PPCODE:
1201 XPUSHs (encode_json (scalar, *SvJSON (self))); 2255{
2256 JSON json;
2257 json_init (&json);
2258 json.flags |= F_UTF8;
2259 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2260 XPUSHs (scalar);
2261}
1202 2262
1203void decode (SV *self, SV *jsonstr) 2263void decode_json (SV *jsonstr)
1204 PPCODE: 2264 PPCODE:
1205 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1206
1207void decode_prefix (SV *self, SV *jsonstr)
1208 PPCODE:
1209{ 2265{
1210 UV offset; 2266 JSON json;
1211 EXTEND (SP, 2); 2267 json_init (&json);
1212 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 2268 json.flags |= F_UTF8;
1213 PUSHs (sv_2mortal (newSVuv (offset))); 2269 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2270 XPUSHs (jsonstr);
1214} 2271}
1215 2272
1216PROTOTYPES: ENABLE
1217
1218void to_json (SV *scalar)
1219 ALIAS:
1220 objToJson = 0
1221 PPCODE:
1222 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1223
1224void from_json (SV *jsonstr)
1225 ALIAS:
1226 jsonToObj = 0
1227 PPCODE:
1228 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1229

Diff Legend

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