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 1.45 by root, Mon Jun 25 06:57:42 2007 UTC vs.
Revision 1.128 by root, Fri Oct 7 05:18:48 2016 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include <assert.h>
6#include "string.h" 6#include <string.h>
7#include "stdlib.h" 7#include <stdlib.h>
8#include "stdio.h" 8#include <stdio.h>
9#include <limits.h>
10#include <float.h>
9 11
10#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 14#endif
13 15
14// some old perls do not have this, try to make it work, no 16// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces. 17// guarantees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
18#endif 20#endif
21
22// compatibility with perl <5.18
23#ifndef HvNAMELEN_get
24# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
25#endif
26#ifndef HvNAMELEN
27# define HvNAMELEN(hv) HvNAMELEN_get (hv)
28#endif
29#ifndef HvNAMEUTF8
30# define HvNAMEUTF8(hv) 0
31#endif
32
33// three extra for rounding, sign, and end of string
34#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
19 35
20#define F_ASCII 0x00000001UL 36#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL 37#define F_LATIN1 0x00000002UL
22#define F_UTF8 0x00000004UL 38#define F_UTF8 0x00000004UL
23#define F_INDENT 0x00000008UL 39#define F_INDENT 0x00000008UL
… …
25#define F_SPACE_BEFORE 0x00000020UL 41#define F_SPACE_BEFORE 0x00000020UL
26#define F_SPACE_AFTER 0x00000040UL 42#define F_SPACE_AFTER 0x00000040UL
27#define F_ALLOW_NONREF 0x00000100UL 43#define F_ALLOW_NONREF 0x00000100UL
28#define F_SHRINK 0x00000200UL 44#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL 45#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL // NYI 46#define F_CONV_BLESSED 0x00000800UL
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL 47#define F_RELAXED 0x00001000UL
34#define S_MAXSIZE 20 48#define F_ALLOW_UNKNOWN 0x00002000UL
35 49#define F_ALLOW_TAGS 0x00004000UL
36#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 50#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
37#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
38 51
39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 52#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
40#define F_DEFAULT (9UL << S_MAXDEPTH)
41 53
42#define INIT_SIZE 32 // initial scalar size to be allocated 54#define INIT_SIZE 32 // initial scalar size to be allocated
43#define INDENT_STEP 3 // spaces per indentation level 55#define INDENT_STEP 3 // spaces per indentation level
44 56
45#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
46 58
59#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
60
47#define SB do { 61#define SB do {
48#define SE } while (0) 62#define SE } while (0)
49 63
50#if __GNUC__ >= 3 64#if __GNUC__ >= 3
51# define expect(expr,value) __builtin_expect ((expr),(value)) 65# define expect(expr,value) __builtin_expect ((expr), (value))
52# define inline inline 66# define INLINE static inline
53#else 67#else
54# define expect(expr,value) (expr) 68# define expect(expr,value) (expr)
55# define inline static 69# define INLINE static
56#endif 70#endif
57 71
58#define expect_false(expr) expect ((expr) != 0, 0) 72#define expect_false(expr) expect ((expr) != 0, 0)
59#define expect_true(expr) expect ((expr) != 0, 1) 73#define expect_true(expr) expect ((expr) != 0, 1)
60 74
61static HV *json_stash, *json_boolean_stash; // JSON::XS:: 75#define IN_RANGE_INC(type,val,beg,end) \
62static SV *json_true, *json_false; 76 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
77 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
78
79#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
80
81#ifdef USE_ITHREADS
82# define JSON_SLOW 1
83# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
84# define BOOL_STASH (bool_stash ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1))
85#else
86# define JSON_SLOW 0
87# define JSON_STASH json_stash
88# define BOOL_STASH bool_stash
89#endif
90
91// the amount of HEs to allocate on the stack, when sorting keys
92#define STACK_HES 64
93
94static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean::
95static SV *bool_true, *bool_false, *sv_json;
96
97enum {
98 INCR_M_WS = 0, // initial whitespace skipping, must be 0
99 INCR_M_STR, // inside string
100 INCR_M_BS, // inside backslash
101 INCR_M_C0, // inside comment in initial whitespace sequence
102 INCR_M_C1, // inside comment in other places
103 INCR_M_JSON // outside anything, count nesting
104};
105
106#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
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}
63 129
64///////////////////////////////////////////////////////////////////////////// 130/////////////////////////////////////////////////////////////////////////////
65// utility functions 131// utility functions
66 132
67static UV * 133INLINE SV *
68SvJSON (SV *sv) 134get_bool (const char *name)
69{ 135{
70 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 136 SV *sv = get_sv (name, 1);
71 croak ("object is not of type JSON::XS");
72 137
73 return &SvUVX (SvRV (sv)); 138 SvREADONLY_on (sv);
74} 139 SvREADONLY_on (SvRV (sv));
75 140
76static void 141 return sv;
142}
143
144INLINE void
77shrink (SV *sv) 145shrink (SV *sv)
78{ 146{
79 sv_utf8_downgrade (sv, 1); 147 sv_utf8_downgrade (sv, 1);
148
80 if (SvLEN (sv) > SvCUR (sv) + 1) 149 if (SvLEN (sv) > SvCUR (sv) + 1)
81 { 150 {
82#ifdef SvPV_shrink_to_cur 151#ifdef SvPV_shrink_to_cur
83 SvPV_shrink_to_cur (sv); 152 SvPV_shrink_to_cur (sv);
84#elif defined (SvPV_renew) 153#elif defined (SvPV_renew)
… …
90// 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
91// case of an error. 160// case of an error.
92// we special-case "safe" characters from U+80 .. U+7FF, 161// we special-case "safe" characters from U+80 .. U+7FF,
93// but use the very good perl function to parse anything else. 162// but use the very good perl function to parse anything else.
94// note that we never call this function for a ascii codepoints 163// note that we never call this function for a ascii codepoints
95inline UV 164INLINE UV
96decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 165decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
97{ 166{
98 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 167 if (expect_true (len >= 2
99 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 168 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
100 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 169 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
101 { 170 {
102 *clen = 2; 171 *clen = 2;
103 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 172 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
104 } 173 }
105 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 }
106 { 277 }
107 *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)
108 return (UV)-1; 340 return 1;
341
342 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
343 return 1;
109 } 344 }
345
346 return 0;
110} 347}
111 348
112///////////////////////////////////////////////////////////////////////////// 349/////////////////////////////////////////////////////////////////////////////
113// encoder 350// encoder
114 351
… …
116typedef struct 353typedef struct
117{ 354{
118 char *cur; // SvPVX (sv) + current output position 355 char *cur; // SvPVX (sv) + current output position
119 char *end; // SvEND (sv) 356 char *end; // SvEND (sv)
120 SV *sv; // result scalar 357 SV *sv; // result scalar
121 U32 flags; // F_* 358 JSON json;
122 U32 indent; // indentation level 359 U32 indent; // indentation level
123 U32 maxdepth; // max. indentation/recursion level 360 UV limit; // escape character values >= this value when encoding
124} enc_t; 361} enc_t;
125 362
126inline void 363INLINE void
127need (enc_t *enc, STRLEN len) 364need (enc_t *enc, STRLEN len)
128{ 365{
129 if (expect_false (enc->cur + len >= enc->end)) 366 if (expect_false (enc->cur + len >= enc->end))
130 { 367 {
131 STRLEN cur = enc->cur - SvPVX (enc->sv); 368 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
132 SvGROW (enc->sv, cur + len + 1); 369 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
133 enc->cur = SvPVX (enc->sv) + cur; 370 enc->cur = SvPVX (enc->sv) + cur;
134 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 371 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
135 } 372 }
136} 373}
137 374
138inline void 375INLINE void
139encode_ch (enc_t *enc, char ch) 376encode_ch (enc_t *enc, char ch)
140{ 377{
141 need (enc, 1); 378 need (enc, 1);
142 *enc->cur++ = ch; 379 *enc->cur++ = ch;
143} 380}
… …
197 { 434 {
198 uch = ch; 435 uch = ch;
199 clen = 1; 436 clen = 1;
200 } 437 }
201 438
202 if (uch > 0x10FFFFUL) 439 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
203 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
204
205 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
206 { 440 {
207 if (uch > 0xFFFFUL) 441 if (uch >= 0x10000UL)
208 { 442 {
443 if (uch >= 0x110000UL)
444 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
445
209 need (enc, len += 11); 446 need (enc, len += 11);
210 sprintf (enc->cur, "\\u%04x\\u%04x", 447 sprintf (enc->cur, "\\u%04x\\u%04x",
211 (int)((uch - 0x10000) / 0x400 + 0xD800), 448 (int)((uch - 0x10000) / 0x400 + 0xD800),
212 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 449 (int)((uch - 0x10000) % 0x400 + 0xDC00));
213 enc->cur += 12; 450 enc->cur += 12;
214 } 451 }
215 else 452 else
216 { 453 {
217 static char hexdigit [16] = "0123456789abcdef";
218 need (enc, len += 5); 454 need (enc, len += 5);
219 *enc->cur++ = '\\'; 455 *enc->cur++ = '\\';
220 *enc->cur++ = 'u'; 456 *enc->cur++ = 'u';
221 *enc->cur++ = hexdigit [ uch >> 12 ]; 457 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
222 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 458 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
223 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 459 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
224 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 460 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
225 } 461 }
226 462
227 str += clen; 463 str += clen;
228 } 464 }
229 else if (enc->flags & F_LATIN1) 465 else if (enc->json.flags & F_LATIN1)
230 { 466 {
231 *enc->cur++ = uch; 467 *enc->cur++ = uch;
232 str += clen; 468 str += clen;
233 } 469 }
234 else if (is_utf8) 470 else if (is_utf8)
… …
241 while (--clen); 477 while (--clen);
242 } 478 }
243 else 479 else
244 { 480 {
245 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
246 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 482 enc->cur = encode_utf8 (enc->cur, uch);
247 ++str; 483 ++str;
248 } 484 }
249 } 485 }
250 } 486 }
251 } 487 }
252 488
253 --len; 489 --len;
254 } 490 }
255} 491}
256 492
257inline void 493INLINE void
258encode_indent (enc_t *enc) 494encode_indent (enc_t *enc)
259{ 495{
260 if (enc->flags & F_INDENT) 496 if (enc->json.flags & F_INDENT)
261 { 497 {
262 int spaces = enc->indent * INDENT_STEP; 498 int spaces = enc->indent * INDENT_STEP;
263 499
264 need (enc, spaces); 500 need (enc, spaces);
265 memset (enc->cur, ' ', spaces); 501 memset (enc->cur, ' ', spaces);
266 enc->cur += spaces; 502 enc->cur += spaces;
267 } 503 }
268} 504}
269 505
270inline void 506INLINE void
271encode_space (enc_t *enc) 507encode_space (enc_t *enc)
272{ 508{
273 need (enc, 1); 509 need (enc, 1);
274 encode_ch (enc, ' '); 510 encode_ch (enc, ' ');
275} 511}
276 512
277inline void 513INLINE void
278encode_nl (enc_t *enc) 514encode_nl (enc_t *enc)
279{ 515{
280 if (enc->flags & F_INDENT) 516 if (enc->json.flags & F_INDENT)
281 { 517 {
282 need (enc, 1); 518 need (enc, 1);
283 encode_ch (enc, '\n'); 519 encode_ch (enc, '\n');
284 } 520 }
285} 521}
286 522
287inline void 523INLINE void
288encode_comma (enc_t *enc) 524encode_comma (enc_t *enc)
289{ 525{
290 encode_ch (enc, ','); 526 encode_ch (enc, ',');
291 527
292 if (enc->flags & F_INDENT) 528 if (enc->json.flags & F_INDENT)
293 encode_nl (enc); 529 encode_nl (enc);
294 else if (enc->flags & F_SPACE_AFTER) 530 else if (enc->json.flags & F_SPACE_AFTER)
295 encode_space (enc); 531 encode_space (enc);
296} 532}
297 533
298static void encode_sv (enc_t *enc, SV *sv); 534static void encode_sv (enc_t *enc, SV *sv);
299 535
300static void 536static void
301encode_av (enc_t *enc, AV *av) 537encode_av (enc_t *enc, AV *av)
302{ 538{
303 int i, len = av_len (av); 539 int i, len = av_len (av);
304 540
305 if (enc->indent >= enc->maxdepth) 541 if (enc->indent >= enc->json.max_depth)
306 croak ("data structure too deep (hit recursion limit)"); 542 croak (ERR_NESTING_EXCEEDED);
307 543
308 encode_ch (enc, '['); encode_nl (enc); 544 encode_ch (enc, '[');
309 ++enc->indent;
310 545
546 if (len >= 0)
547 {
548 encode_nl (enc); ++enc->indent;
549
311 for (i = 0; i <= len; ++i) 550 for (i = 0; i <= len; ++i)
312 { 551 {
552 SV **svp = av_fetch (av, i, 0);
553
313 encode_indent (enc); 554 encode_indent (enc);
314 encode_sv (enc, *av_fetch (av, i, 0));
315 555
556 if (svp)
557 encode_sv (enc, *svp);
558 else
559 encode_str (enc, "null", 4, 0);
560
316 if (i < len) 561 if (i < len)
317 encode_comma (enc); 562 encode_comma (enc);
318 } 563 }
319 564
565 encode_nl (enc); --enc->indent; encode_indent (enc);
566 }
567
320 encode_nl (enc); 568 encode_ch (enc, ']');
321
322 --enc->indent;
323 encode_indent (enc); encode_ch (enc, ']');
324} 569}
325 570
326static void 571static void
327encode_he (enc_t *enc, HE *he) 572encode_hk (enc_t *enc, HE *he)
328{ 573{
329 encode_ch (enc, '"'); 574 encode_ch (enc, '"');
330 575
331 if (HeKLEN (he) == HEf_SVKEY) 576 if (HeKLEN (he) == HEf_SVKEY)
332 { 577 {
333 SV *sv = HeSVKEY (he); 578 SV *sv = HeSVKEY (he);
334 STRLEN len; 579 STRLEN len;
335 char *str; 580 char *str;
336 581
337 SvGETMAGIC (sv); 582 SvGETMAGIC (sv);
338 str = SvPV (sv, len); 583 str = SvPV (sv, len);
339 584
340 encode_str (enc, str, len, SvUTF8 (sv)); 585 encode_str (enc, str, len, SvUTF8 (sv));
341 } 586 }
342 else 587 else
343 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 588 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
344 589
345 encode_ch (enc, '"'); 590 encode_ch (enc, '"');
346 591
347 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 592 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
348 encode_ch (enc, ':'); 593 encode_ch (enc, ':');
349 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 594 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
350 encode_sv (enc, HeVAL (he));
351} 595}
352 596
353// compare hash entries, used when all keys are bytestrings 597// compare hash entries, used when all keys are bytestrings
354static int 598static int
355he_cmp_fast (const void *a_, const void *b_) 599he_cmp_fast (const void *a_, const void *b_)
… …
360 HE *b = *(HE **)b_; 604 HE *b = *(HE **)b_;
361 605
362 STRLEN la = HeKLEN (a); 606 STRLEN la = HeKLEN (a);
363 STRLEN lb = HeKLEN (b); 607 STRLEN lb = HeKLEN (b);
364 608
365 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 609 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
366 cmp = la - lb; 610 cmp = lb - la;
367 611
368 return cmp; 612 return cmp;
369} 613}
370 614
371// 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
372static int 616static int
373he_cmp_slow (const void *a, const void *b) 617he_cmp_slow (const void *a, const void *b)
374{ 618{
375 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 619 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
376} 620}
377 621
378static void 622static void
379encode_hv (enc_t *enc, HV *hv) 623encode_hv (enc_t *enc, HV *hv)
380{ 624{
381 int count, i; 625 HE *he;
382 626
383 if (enc->indent >= enc->maxdepth) 627 if (enc->indent >= enc->json.max_depth)
384 croak ("data structure too deep (hit recursion limit)"); 628 croak (ERR_NESTING_EXCEEDED);
385 629
386 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 630 encode_ch (enc, '{');
387 631
388 if ((count = hv_iterinit (hv)))
389 {
390 // for canonical output we have to sort by keys first 632 // for canonical output we have to sort by keys first
391 // actually, this is mostly due to the stupid so-called 633 // actually, this is mostly due to the stupid so-called
392 // security workaround added somewhere in 5.8.x. 634 // security workaround added somewhere in 5.8.x
393 // that randomises hash orderings 635 // that randomises hash orderings
394 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))
395 { 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 {
396 int fast = 1; 656 int i, fast = 1;
397 HE *he; 657 HE *hes_stack [STACK_HES];
398#if defined(__BORLANDC__) || defined(_MSC_VER) 658 HE **hes = hes_stack;
399 HE **hes = _alloca (count * sizeof (HE)); 659
400#else 660 // allocate larger arrays on the heap
401 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 661 if (count > STACK_HES)
402#endif 662 {
663 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
664 hes = (HE **)SvPVX (sv);
665 }
403 666
404 i = 0; 667 i = 0;
405 while ((he = hv_iternext (hv))) 668 while ((he = hv_iternext (hv)))
406 { 669 {
407 hes [i++] = he; 670 hes [i++] = he;
… …
429 692
430 FREETMPS; 693 FREETMPS;
431 LEAVE; 694 LEAVE;
432 } 695 }
433 696
434 for (i = 0; i < count; ++i) 697 encode_nl (enc); ++enc->indent;
698
699 while (count--)
435 { 700 {
436 encode_indent (enc); 701 encode_indent (enc);
702 he = hes [count];
437 encode_he (enc, hes [i]); 703 encode_hk (enc, he);
704 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
438 705
439 if (i < count - 1) 706 if (count)
440 encode_comma (enc); 707 encode_comma (enc);
441 } 708 }
442 709
443 encode_nl (enc); 710 encode_nl (enc); --enc->indent; encode_indent (enc);
444 } 711 }
712 }
445 else 713 else
714 {
715 if (hv_iterinit (hv) || SvMAGICAL (hv))
716 if ((he = hv_iternext (hv)))
446 { 717 {
447 HE *he = hv_iternext (hv); 718 encode_nl (enc); ++enc->indent;
448 719
449 for (;;) 720 for (;;)
450 { 721 {
451 encode_indent (enc); 722 encode_indent (enc);
452 encode_he (enc, he); 723 encode_hk (enc, he);
724 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
453 725
454 if (!(he = hv_iternext (hv))) 726 if (!(he = hv_iternext (hv)))
455 break; 727 break;
456 728
457 encode_comma (enc); 729 encode_comma (enc);
458 } 730 }
459 731
460 encode_nl (enc); 732 encode_nl (enc); --enc->indent; encode_indent (enc);
461 } 733 }
462 } 734 }
463 735
464 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 736 encode_ch (enc, '}');
465} 737}
466 738
467// encode objects, arrays and special \0=false and \1=true values. 739// encode objects, arrays and special \0=false and \1=true values.
468static void 740static void
469encode_rv (enc_t *enc, SV *sv) 741encode_rv (enc_t *enc, SV *sv)
470{ 742{
471 svtype svt; 743 svtype svt;
744 GV *method;
472 745
473 SvGETMAGIC (sv); 746 SvGETMAGIC (sv);
474 svt = SvTYPE (sv); 747 svt = SvTYPE (sv);
475 748
476 if (expect_false (SvOBJECT (sv))) 749 if (expect_false (SvOBJECT (sv)))
477 { 750 {
478 if (SvSTASH (sv) == json_boolean_stash) 751 HV *stash = SvSTASH (sv);
752
753 if (stash == bool_stash)
479 { 754 {
480 if (SvIV (sv) == 0) 755 if (SvIV (sv))
756 encode_str (enc, "true", 4, 0);
757 else
481 encode_str (enc, "false", 5, 0); 758 encode_str (enc, "false", 5, 0);
482 else
483 encode_str (enc, "true", 4, 0);
484 } 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);
485 else 827 else
486 {
487#if 0
488 if (0 && sv_derived_from (rv, "JSON::Literal"))
489 {
490 // not yet
491 }
492#endif
493 if (enc->flags & F_CONV_BLESSED)
494 {
495 // we re-bless the reference to get overload and other niceties right
496 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
497
498 if (to_json)
499 {
500 dSP;
501 ENTER;
502 SAVETMPS;
503 PUSHMARK (SP);
504 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
505
506 // calling with G_SCALAR ensures that we always get a 1 reutrn value
507 // check anyways.
508 PUTBACK;
509 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
510 SPAGAIN;
511
512 encode_sv (enc, POPs);
513
514 FREETMPS;
515 LEAVE;
516 }
517 else if (enc->flags & F_ALLOW_BLESSED)
518 encode_str (enc, "null", 4, 0);
519 else
520 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
521 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
522 }
523 else if (enc->flags & F_ALLOW_BLESSED)
524 encode_str (enc, "null", 4, 0);
525 else
526 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 828 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
527 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 829 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
528 }
529 } 830 }
530 else if (svt == SVt_PVHV) 831 else if (svt == SVt_PVHV)
531 encode_hv (enc, (HV *)sv); 832 encode_hv (enc, (HV *)sv);
532 else if (svt == SVt_PVAV) 833 else if (svt == SVt_PVAV)
533 encode_av (enc, (AV *)sv); 834 encode_av (enc, (AV *)sv);
534 else if (svt < SVt_PVAV) 835 else if (svt < SVt_PVAV)
535 { 836 {
536 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)
537 encode_str (enc, "false", 5, 0); 842 encode_str (enc, "false", 5, 0);
538 else if (SvNIOK (sv) && SvIV (sv) == 1) 843 else if (enc->json.flags & F_ALLOW_UNKNOWN)
539 encode_str (enc, "true", 4, 0); 844 encode_str (enc, "null", 4, 0);
540 else 845 else
541 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",
542 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 847 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
543 } 848 }
849 else if (enc->json.flags & F_ALLOW_UNKNOWN)
850 encode_str (enc, "null", 4, 0);
544 else 851 else
545 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",
546 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 853 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
547} 854}
548 855
… …
566 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 873 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
567 enc->cur += strlen (enc->cur); 874 enc->cur += strlen (enc->cur);
568 } 875 }
569 else if (SvIOKp (sv)) 876 else if (SvIOKp (sv))
570 { 877 {
571 // 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
572 if (SvUV (sv) & ~(UV)0x7fff) 879 // we assume two's complement
573 { 880 // we assume no aliasing issues in the union
574 // large integer, use the (rather slow) snprintf way. 881 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
575 need (enc, sizeof (UV) * 3); 882 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
576 enc->cur +=
577 SvIsUV(sv)
578 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
579 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
580 }
581 else
582 { 883 {
583 // optimise the "small number case" 884 // optimise the "small number case"
584 // code will likely be branchless and use only a single multiplication 885 // code will likely be branchless and use only a single multiplication
886 // works for numbers up to 59074
585 I32 i = SvIV (sv); 887 I32 i = SvIVX (sv);
586 U32 u; 888 U32 u;
587 char digit, nz = 0; 889 char digit, nz = 0;
588 890
589 need (enc, 6); 891 need (enc, 6);
590 892
… …
596 898
597 // now output digit by digit, each time masking out the integer part 899 // now output digit by digit, each time masking out the integer part
598 // and multiplying by 5 while moving the decimal point one to the right, 900 // and multiplying by 5 while moving the decimal point one to the right,
599 // resulting in a net multiplication by 10. 901 // resulting in a net multiplication by 10.
600 // we always write the digit to memory but conditionally increment 902 // we always write the digit to memory but conditionally increment
601 // the pointer, to ease the usage of conditional move instructions. 903 // the pointer, to enable the use of conditional move instructions.
602 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 904 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
603 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 905 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
604 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 906 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
605 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 907 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
606 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 908 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
607 } 909 }
910 else
911 {
912 // large integer, use the (rather slow) snprintf way.
913 need (enc, IVUV_MAXCHARS);
914 enc->cur +=
915 SvIsUV(sv)
916 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
917 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
918 }
608 } 919 }
609 else if (SvROK (sv)) 920 else if (SvROK (sv))
610 encode_rv (enc, SvRV (sv)); 921 encode_rv (enc, SvRV (sv));
611 else if (!SvOK (sv)) 922 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
612 encode_str (enc, "null", 4, 0); 923 encode_str (enc, "null", 4, 0);
613 else 924 else
614 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",
615 SvPV_nolen (sv), SvFLAGS (sv)); 926 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
616} 927}
617 928
618static SV * 929static SV *
619encode_json (SV *scalar, U32 flags) 930encode_json (SV *scalar, JSON *json)
620{ 931{
621 enc_t enc; 932 enc_t enc;
622 933
623 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 934 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
624 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)");
625 936
626 enc.flags = flags; 937 enc.json = *json;
627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 938 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
628 enc.cur = SvPVX (enc.sv); 939 enc.cur = SvPVX (enc.sv);
629 enc.end = SvEND (enc.sv); 940 enc.end = SvEND (enc.sv);
630 enc.indent = 0; 941 enc.indent = 0;
631 enc.maxdepth = DEC_DEPTH (flags); 942 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
943 : enc.json.flags & F_LATIN1 ? 0x000100UL
944 : 0x110000UL;
632 945
633 SvPOK_only (enc.sv); 946 SvPOK_only (enc.sv);
634 encode_sv (&enc, scalar); 947 encode_sv (&enc, scalar);
948 encode_nl (&enc);
635 949
636 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 950 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
637 *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
638 952
639 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 953 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
640 SvUTF8_on (enc.sv); 954 SvUTF8_on (enc.sv);
641 955
642 if (enc.flags & F_SHRINK) 956 if (enc.json.flags & F_SHRINK)
643 shrink (enc.sv); 957 shrink (enc.sv);
644 958
645 return enc.sv; 959 return enc.sv;
646} 960}
647 961
… …
652typedef struct 966typedef struct
653{ 967{
654 char *cur; // current parser pointer 968 char *cur; // current parser pointer
655 char *end; // end of input string 969 char *end; // end of input string
656 const char *err; // parse error, if != 0 970 const char *err; // parse error, if != 0
657 U32 flags; // F_* 971 JSON json;
658 U32 depth; // recursion depth 972 U32 depth; // recursion depth
659 U32 maxdepth; // recursion depth limit 973 U32 maxdepth; // recursion depth limit
660} dec_t; 974} dec_t;
661 975
662inline 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
663decode_ws (dec_t *dec) 986decode_ws (dec_t *dec)
664{ 987{
665 for (;;) 988 for (;;)
666 { 989 {
667 char ch = *dec->cur; 990 char ch = *dec->cur;
668 991
669 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 }
670 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 1004 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
671 break; 1005 break; // parse error, but let higher level handle it, gives better error messages
672 1006
673 ++dec->cur; 1007 ++dec->cur;
674 } 1008 }
675} 1009}
676 1010
… …
680 if (*dec->cur != ch) \ 1014 if (*dec->cur != ch) \
681 ERR (# ch " expected"); \ 1015 ERR (# ch " expected"); \
682 ++dec->cur; \ 1016 ++dec->cur; \
683 SE 1017 SE
684 1018
685#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)
686#define DEC_DEC_DEPTH --dec->depth 1020#define DEC_DEC_DEPTH --dec->depth
687 1021
688static SV *decode_sv (dec_t *dec); 1022static SV *decode_sv (dec_t *dec);
689 1023
690static signed char decode_hexdigit[256]; 1024static signed char decode_hexdigit[256];
… …
782 1116
783 if (hi >= 0x80) 1117 if (hi >= 0x80)
784 { 1118 {
785 utf8 = 1; 1119 utf8 = 1;
786 1120
787 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1121 cur = encode_utf8 (cur, hi);
788 } 1122 }
789 else 1123 else
790 *cur++ = hi; 1124 *cur++ = hi;
791 } 1125 }
792 break; 1126 break;
… …
794 default: 1128 default:
795 --dec_cur; 1129 --dec_cur;
796 ERR ("illegal backslash escape sequence in string"); 1130 ERR ("illegal backslash escape sequence in string");
797 } 1131 }
798 } 1132 }
799 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1133 else if (expect_true (ch >= 0x20 && ch < 0x80))
800 *cur++ = ch; 1134 *cur++ = ch;
801 else if (ch >= 0x80) 1135 else if (ch >= 0x80)
802 { 1136 {
803 STRLEN clen; 1137 STRLEN clen;
804 UV uch;
805 1138
806 --dec_cur; 1139 --dec_cur;
807 1140
808 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1141 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
809 if (clen == (STRLEN)-1) 1142 if (clen == (STRLEN)-1)
810 ERR ("malformed UTF-8 character in JSON string"); 1143 ERR ("malformed UTF-8 character in JSON string");
811 1144
812 do 1145 do
813 *cur++ = *dec_cur++; 1146 *cur++ = *dec_cur++;
814 while (--clen); 1147 while (--clen);
815 1148
816 utf8 = 1; 1149 utf8 = 1;
817 } 1150 }
1151 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1152 *cur++ = ch;
818 else 1153 else
819 { 1154 {
820 --dec_cur; 1155 --dec_cur;
821 1156
822 if (!ch) 1157 if (!ch)
… …
830 { 1165 {
831 STRLEN len = cur - buf; 1166 STRLEN len = cur - buf;
832 1167
833 if (sv) 1168 if (sv)
834 { 1169 {
835 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
836 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1175 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
837 SvCUR_set (sv, SvCUR (sv) + len); 1176 SvCUR_set (sv, SvCUR (sv) + len);
838 } 1177 }
839 else 1178 else
840 sv = newSVpvn (buf, len); 1179 sv = newSVpvn (buf, len);
… …
925 is_nv = 1; 1264 is_nv = 1;
926 } 1265 }
927 1266
928 if (!is_nv) 1267 if (!is_nv)
929 { 1268 {
1269 int len = dec->cur - start;
1270
930 // 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
931 if (*start == '-') 1272 if (*start == '-')
932 switch (dec->cur - start) 1273 switch (len)
933 { 1274 {
934 case 2: return newSViv (-( start [1] - '0' * 1)); 1275 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
935 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1276 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
936 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));
937 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));
938 } 1280 }
939 else 1281 else
940 switch (dec->cur - start) 1282 switch (len)
941 { 1283 {
942 case 1: return newSViv ( start [0] - '0' * 1); 1284 case 1: return newSViv ( start [0] - '0' * 1);
943 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1285 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
944 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);
945 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);
946 } 1289 }
947 1290
948 { 1291 {
949 UV uv; 1292 UV uv;
950 int numtype = grok_number (start, dec->cur - start, &uv); 1293 int numtype = grok_number (start, len, &uv);
951 if (numtype & IS_NUMBER_IN_UV) 1294 if (numtype & IS_NUMBER_IN_UV)
952 if (numtype & IS_NUMBER_NEG) 1295 if (numtype & IS_NUMBER_NEG)
953 { 1296 {
954 if (uv < (UV)IV_MIN) 1297 if (uv < (UV)IV_MIN)
955 return newSViv (-(IV)uv); 1298 return newSViv (-(IV)uv);
956 } 1299 }
957 else 1300 else
958 return newSVuv (uv); 1301 return newSVuv (uv);
959
960 // here would likely be the place for bigint support
961 } 1302 }
962 }
963 1303
964 // if we ever support bigint or bigfloat, this is the place for bigfloat 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
965 return newSVnv (Atof (start)); 1316 return newSVnv (json_atof (start));
966 1317
967fail: 1318fail:
968 return 0; 1319 return 0;
969} 1320}
970 1321
… …
994 if (*dec->cur == ']') 1345 if (*dec->cur == ']')
995 { 1346 {
996 ++dec->cur; 1347 ++dec->cur;
997 break; 1348 break;
998 } 1349 }
999 1350
1000 if (*dec->cur != ',') 1351 if (*dec->cur != ',')
1001 ERR (", or ] expected while parsing array"); 1352 ERR (", or ] expected while parsing array");
1002 1353
1003 ++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 }
1004 } 1363 }
1005 1364
1006 DEC_DEC_DEPTH; 1365 DEC_DEC_DEPTH;
1007 return newRV_noinc ((SV *)av); 1366 return newRV_noinc ((SV *)av);
1008 1367
… …
1013} 1372}
1014 1373
1015static SV * 1374static SV *
1016decode_hv (dec_t *dec) 1375decode_hv (dec_t *dec)
1017{ 1376{
1377 SV *sv;
1018 HV *hv = newHV (); 1378 HV *hv = newHV ();
1019 1379
1020 DEC_INC_DEPTH; 1380 DEC_INC_DEPTH;
1021 decode_ws (dec); 1381 decode_ws (dec);
1022 1382
1023 if (*dec->cur == '}') 1383 if (*dec->cur == '}')
1024 ++dec->cur; 1384 ++dec->cur;
1025 else 1385 else
1026 for (;;) 1386 for (;;)
1027 { 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 {
1028 SV *key, *value; 1396 SV *value;
1397 char *p = dec->cur;
1398 char *e = p + 24; // only try up to 24 bytes
1029 1399
1030 decode_ws (dec); EXPECT_CH ('"'); 1400 for (;;)
1031
1032 key = decode_str (dec);
1033 if (!key)
1034 goto fail;
1035
1036 decode_ws (dec); EXPECT_CH (':');
1037
1038 value = decode_sv (dec);
1039 if (!value)
1040 { 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);
1041 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)
1042 goto fail; 1437 goto fail;
1438
1439 hv_store (hv, key, len, value, 0);
1440
1441 break;
1442 }
1443
1444 ++p;
1043 } 1445 }
1044 1446 }
1045 hv_store_ent (hv, key, value, 0);
1046 SvREFCNT_dec (key);
1047 1447
1048 decode_ws (dec); 1448 decode_ws (dec);
1049 1449
1050 if (*dec->cur == '}') 1450 if (*dec->cur == '}')
1051 { 1451 {
… …
1055 1455
1056 if (*dec->cur != ',') 1456 if (*dec->cur != ',')
1057 ERR (", or } expected while parsing object/hash"); 1457 ERR (", or } expected while parsing object/hash");
1058 1458
1059 ++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 }
1060 } 1468 }
1061 1469
1062 DEC_DEC_DEPTH; 1470 DEC_DEC_DEPTH;
1063 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;
1064 1538
1065fail: 1539fail:
1066 SvREFCNT_dec (hv); 1540 SvREFCNT_dec (hv);
1067 DEC_DEC_DEPTH; 1541 DEC_DEC_DEPTH;
1068 return 0; 1542 return 0;
1069} 1543}
1070 1544
1071static 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 *
1072decode_sv (dec_t *dec) 1629decode_sv (dec_t *dec)
1073{ 1630{
1074 decode_ws (dec);
1075
1076 // the beauty of JSON: you need exactly one character lookahead 1631 // the beauty of JSON: you need exactly one character lookahead
1077 // to parse anything. 1632 // to parse everything.
1078 switch (*dec->cur) 1633 switch (*dec->cur)
1079 { 1634 {
1080 case '"': ++dec->cur; return decode_str (dec); 1635 case '"': ++dec->cur; return decode_str (dec);
1081 case '[': ++dec->cur; return decode_av (dec); 1636 case '[': ++dec->cur; return decode_av (dec);
1082 case '{': ++dec->cur; return decode_hv (dec); 1637 case '{': ++dec->cur; return decode_hv (dec);
1638 case '(': return decode_tag (dec);
1083 1639
1084 case '-': 1640 case '-':
1085 case '0': case '1': case '2': case '3': case '4': 1641 case '0': case '1': case '2': case '3': case '4':
1086 case '5': case '6': case '7': case '8': case '9': 1642 case '5': case '6': case '7': case '8': case '9':
1087 return decode_num (dec); 1643 return decode_num (dec);
1088 1644
1089 case 't': 1645 case 't':
1090 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1646 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1091 { 1647 {
1092 dec->cur += 4; 1648 dec->cur += 4;
1093 return SvREFCNT_inc (json_true); 1649#if JSON_SLOW
1650 bool_true = get_bool ("Types::Serialiser::true");
1651#endif
1652 return newSVsv (bool_true);
1094 } 1653 }
1095 else 1654 else
1096 ERR ("'true' expected"); 1655 ERR ("'true' expected");
1097 1656
1098 break; 1657 break;
1099 1658
1100 case 'f': 1659 case 'f':
1101 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1660 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1102 { 1661 {
1103 dec->cur += 5; 1662 dec->cur += 5;
1663#if JSON_SLOW
1664 bool_false = get_bool ("Types::Serialiser::false");
1665#endif
1104 return SvREFCNT_inc (json_false); 1666 return newSVsv (bool_false);
1105 } 1667 }
1106 else 1668 else
1107 ERR ("'false' expected"); 1669 ERR ("'false' expected");
1108 1670
1109 break; 1671 break;
… …
1118 ERR ("'null' expected"); 1680 ERR ("'null' expected");
1119 1681
1120 break; 1682 break;
1121 1683
1122 default: 1684 default:
1123 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1685 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1124 break; 1686 break;
1125 } 1687 }
1126 1688
1127fail: 1689fail:
1128 return 0; 1690 return 0;
1129} 1691}
1130 1692
1131static SV * 1693static SV *
1132decode_json (SV *string, U32 flags, UV *offset_return) 1694decode_json (SV *string, JSON *json, char **offset_return)
1133{ 1695{
1134 dec_t dec; 1696 dec_t dec;
1135 UV offset;
1136 SV *sv; 1697 SV *sv;
1137 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 */
1138 SvGETMAGIC (string); 1707 /*SvGETMAGIC (string);*/
1708 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1709 string = sv_2mortal (newSVsv (string));
1710
1139 SvUPGRADE (string, SVt_PV); 1711 SvUPGRADE (string, SVt_PV);
1140 1712
1141 if (flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (flags)) 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)
1142 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1731 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1143 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (flags)); 1732 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1733 }
1144 1734
1145 if (flags & F_UTF8) 1735 if (DECODE_WANTS_OCTETS (json))
1146 sv_utf8_downgrade (string, 0); 1736 sv_utf8_downgrade (string, 0);
1147 else 1737 else
1148 sv_utf8_upgrade (string); 1738 sv_utf8_upgrade (string);
1149 1739
1150 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1740 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1151 1741
1152 dec.flags = flags; 1742 dec.json = *json;
1153 dec.cur = SvPVX (string); 1743 dec.cur = SvPVX (string);
1154 dec.end = SvEND (string); 1744 dec.end = SvEND (string);
1155 dec.err = 0; 1745 dec.err = 0;
1156 dec.depth = 0; 1746 dec.depth = 0;
1157 dec.maxdepth = DEC_DEPTH (dec.flags); 1747
1748 if (dec.json.cb_object || dec.json.cb_sk_object)
1749 dec.json.flags |= F_HOOK;
1158 1750
1159 *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);
1160 sv = decode_sv (&dec); 1754 sv = decode_sv (&dec);
1755
1756 if (offset_return)
1757 *offset_return = dec.cur;
1161 1758
1162 if (!(offset_return || !sv)) 1759 if (!(offset_return || !sv))
1163 { 1760 {
1164 // check for trailing garbage 1761 // check for trailing garbage
1165 decode_ws (&dec); 1762 decode_ws (&dec);
… …
1168 { 1765 {
1169 dec.err = "garbage after JSON object"; 1766 dec.err = "garbage after JSON object";
1170 SvREFCNT_dec (sv); 1767 SvREFCNT_dec (sv);
1171 sv = 0; 1768 sv = 0;
1172 } 1769 }
1173 }
1174
1175 if (offset_return || !sv)
1176 {
1177 offset = dec.flags & F_UTF8
1178 ? dec.cur - SvPVX (string)
1179 : utf8_distance (dec.cur, SvPVX (string));
1180
1181 if (offset_return)
1182 *offset_return = offset;
1183 } 1770 }
1184 1771
1185 if (!sv) 1772 if (!sv)
1186 { 1773 {
1187 SV *uni = sv_newmortal (); 1774 SV *uni = sv_newmortal ();
… …
1193 SAVEVPTR (PL_curcop); 1780 SAVEVPTR (PL_curcop);
1194 PL_curcop = &cop; 1781 PL_curcop = &cop;
1195 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1782 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1196 LEAVE; 1783 LEAVE;
1197 1784
1198 croak ("%s, at character offset %d [\"%s\"]", 1785 croak ("%s, at character offset %d (before \"%s\")",
1199 dec.err, 1786 dec.err,
1200 (int)offset, 1787 (int)ptr_to_index (string, dec.cur),
1201 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1788 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1202 } 1789 }
1203 1790
1204 sv = sv_2mortal (sv); 1791 sv = sv_2mortal (sv);
1205 1792
1206 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1793 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1207 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1794 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1208 1795
1209 return sv; 1796 return sv;
1797}
1798
1799/////////////////////////////////////////////////////////////////////////////
1800// incremental parser
1801
1802static void
1803incr_parse (JSON *self)
1804{
1805 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1806
1807 // the state machine here is a bit convoluted and could be simplified a lot
1808 // but this would make it slower, so...
1809
1810 for (;;)
1811 {
1812 //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
1813 switch (self->incr_mode)
1814 {
1815 // only used for initial whitespace skipping
1816 case INCR_M_WS:
1817 for (;;)
1818 {
1819 if (*p > 0x20)
1820 {
1821 if (*p == '#')
1822 {
1823 self->incr_mode = INCR_M_C0;
1824 goto incr_m_c;
1825 }
1826 else
1827 {
1828 self->incr_mode = INCR_M_JSON;
1829 goto incr_m_json;
1830 }
1831 }
1832 else if (!*p)
1833 goto interrupt;
1834
1835 ++p;
1836 }
1837
1838 // skip a single char inside a string (for \\-processing)
1839 case INCR_M_BS:
1840 if (!*p)
1841 goto interrupt;
1842
1843 ++p;
1844 self->incr_mode = INCR_M_STR;
1845 goto incr_m_str;
1846
1847 // inside #-style comments
1848 case INCR_M_C0:
1849 case INCR_M_C1:
1850 incr_m_c:
1851 for (;;)
1852 {
1853 if (*p == '\n')
1854 {
1855 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1856 break;
1857 }
1858 else if (!*p)
1859 goto interrupt;
1860
1861 ++p;
1862 }
1863
1864 break;
1865
1866 // inside a string
1867 case INCR_M_STR:
1868 incr_m_str:
1869 for (;;)
1870 {
1871 if (*p == '"')
1872 {
1873 ++p;
1874 self->incr_mode = INCR_M_JSON;
1875
1876 if (!self->incr_nest)
1877 goto interrupt;
1878
1879 goto incr_m_json;
1880 }
1881 else if (*p == '\\')
1882 {
1883 ++p; // "virtually" consumes character after \
1884
1885 if (!*p) // if at end of string we have to switch modes
1886 {
1887 self->incr_mode = INCR_M_BS;
1888 goto interrupt;
1889 }
1890 }
1891 else if (!*p)
1892 goto interrupt;
1893
1894 ++p;
1895 }
1896
1897 // after initial ws, outside string
1898 case INCR_M_JSON:
1899 incr_m_json:
1900 for (;;)
1901 {
1902 switch (*p++)
1903 {
1904 case 0:
1905 --p;
1906 goto interrupt;
1907
1908 case 0x09:
1909 case 0x0a:
1910 case 0x0d:
1911 case 0x20:
1912 if (!self->incr_nest)
1913 {
1914 --p; // do not eat the whitespace, let the next round do it
1915 goto interrupt;
1916 }
1917 break;
1918
1919 case '"':
1920 self->incr_mode = INCR_M_STR;
1921 goto incr_m_str;
1922
1923 case '[':
1924 case '{':
1925 case '(':
1926 if (++self->incr_nest > self->max_depth)
1927 croak (ERR_NESTING_EXCEEDED);
1928 break;
1929
1930 case ']':
1931 case '}':
1932 if (--self->incr_nest <= 0)
1933 goto interrupt;
1934 break;
1935
1936 case ')':
1937 --self->incr_nest;
1938 break;
1939
1940 case '#':
1941 self->incr_mode = INCR_M_C1;
1942 goto incr_m_c;
1943 }
1944 }
1945 }
1946
1947 modechange:
1948 ;
1949 }
1950
1951interrupt:
1952 self->incr_pos = p - SvPVX (self->incr_text);
1953 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1954 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1210} 1955}
1211 1956
1212///////////////////////////////////////////////////////////////////////////// 1957/////////////////////////////////////////////////////////////////////////////
1213// XS interface functions 1958// XS interface functions
1214 1959
… …
1223 i >= '0' && i <= '9' ? i - '0' 1968 i >= '0' && i <= '9' ? i - '0'
1224 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1969 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1225 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1970 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1226 : -1; 1971 : -1;
1227 1972
1228 json_stash = gv_stashpv ("JSON::XS" , 1); 1973 json_stash = gv_stashpv ("JSON::XS" , 1);
1229 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1974 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1975 bool_true = get_bool ("Types::Serialiser::true");
1976 bool_false = get_bool ("Types::Serialiser::false");
1230 1977
1231 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1978 sv_json = newSVpv ("JSON", 0);
1232 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1979 SvREADONLY_on (sv_json);
1980
1981 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1233} 1982}
1234 1983
1235PROTOTYPES: DISABLE 1984PROTOTYPES: DISABLE
1236 1985
1237SV *new (char *dummy) 1986void CLONE (...)
1238 CODE: 1987 CODE:
1239 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1988 json_stash = 0;
1240 OUTPUT: 1989 bool_stash = 0;
1241 RETVAL
1242 1990
1991void new (char *klass)
1992 PPCODE:
1993{
1994 SV *pv = NEWSV (0, sizeof (JSON));
1995 SvPOK_only (pv);
1996 json_init ((JSON *)SvPVX (pv));
1997 XPUSHs (sv_2mortal (sv_bless (
1998 newRV_noinc (pv),
1999 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
2000 )));
2001}
2002
1243SV *ascii (SV *self, int enable = 1) 2003void ascii (JSON *self, int enable = 1)
1244 ALIAS: 2004 ALIAS:
1245 ascii = F_ASCII 2005 ascii = F_ASCII
1246 latin1 = F_LATIN1 2006 latin1 = F_LATIN1
1247 utf8 = F_UTF8 2007 utf8 = F_UTF8
1248 indent = F_INDENT 2008 indent = F_INDENT
… …
1252 pretty = F_PRETTY 2012 pretty = F_PRETTY
1253 allow_nonref = F_ALLOW_NONREF 2013 allow_nonref = F_ALLOW_NONREF
1254 shrink = F_SHRINK 2014 shrink = F_SHRINK
1255 allow_blessed = F_ALLOW_BLESSED 2015 allow_blessed = F_ALLOW_BLESSED
1256 convert_blessed = F_CONV_BLESSED 2016 convert_blessed = F_CONV_BLESSED
2017 relaxed = F_RELAXED
2018 allow_unknown = F_ALLOW_UNKNOWN
2019 allow_tags = F_ALLOW_TAGS
2020 PPCODE:
2021{
2022 if (enable)
2023 self->flags |= ix;
2024 else
2025 self->flags &= ~ix;
2026
2027 XPUSHs (ST (0));
2028}
2029
2030void get_ascii (JSON *self)
2031 ALIAS:
2032 get_ascii = F_ASCII
2033 get_latin1 = F_LATIN1
2034 get_utf8 = F_UTF8
2035 get_indent = F_INDENT
2036 get_canonical = F_CANONICAL
2037 get_space_before = F_SPACE_BEFORE
2038 get_space_after = F_SPACE_AFTER
2039 get_allow_nonref = F_ALLOW_NONREF
2040 get_shrink = F_SHRINK
2041 get_allow_blessed = F_ALLOW_BLESSED
2042 get_convert_blessed = F_CONV_BLESSED
2043 get_relaxed = F_RELAXED
2044 get_allow_unknown = F_ALLOW_UNKNOWN
2045 get_allow_tags = F_ALLOW_TAGS
2046 PPCODE:
2047 XPUSHs (boolSV (self->flags & ix));
2048
2049void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
2050 PPCODE:
2051 self->max_depth = max_depth;
2052 XPUSHs (ST (0));
2053
2054U32 get_max_depth (JSON *self)
1257 CODE: 2055 CODE:
1258{ 2056 RETVAL = self->max_depth;
1259 UV *uv = SvJSON (self);
1260 if (enable)
1261 *uv |= ix;
1262 else
1263 *uv &= ~ix;
1264
1265 RETVAL = newSVsv (self);
1266}
1267 OUTPUT: 2057 OUTPUT:
1268 RETVAL 2058 RETVAL
1269 2059
1270SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 2060void max_size (JSON *self, U32 max_size = 0)
2061 PPCODE:
2062 self->max_size = max_size;
2063 XPUSHs (ST (0));
2064
2065int get_max_size (JSON *self)
1271 CODE: 2066 CODE:
1272{ 2067 RETVAL = self->max_size;
1273 UV *uv = SvJSON (self);
1274 UV log2 = 0;
1275
1276 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1277
1278 while ((1UL << log2) < max_depth)
1279 ++log2;
1280
1281 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1282
1283 RETVAL = newSVsv (self);
1284}
1285 OUTPUT: 2068 OUTPUT:
1286 RETVAL 2069 RETVAL
1287 2070
1288SV *max_size (SV *self, UV max_size = 0) 2071void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
2072 PPCODE:
2073{
2074 SvREFCNT_dec (self->cb_object);
2075 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
2076
2077 XPUSHs (ST (0));
2078}
2079
2080void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
2081 PPCODE:
2082{
2083 if (!self->cb_sk_object)
2084 self->cb_sk_object = newHV ();
2085
2086 if (SvOK (cb))
2087 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
2088 else
2089 {
2090 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
2091
2092 if (!HvKEYS (self->cb_sk_object))
2093 {
2094 SvREFCNT_dec (self->cb_sk_object);
2095 self->cb_sk_object = 0;
2096 }
2097 }
2098
2099 XPUSHs (ST (0));
2100}
2101
2102void encode (JSON *self, SV *scalar)
2103 PPCODE:
2104 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2105 XPUSHs (scalar);
2106
2107void decode (JSON *self, SV *jsonstr)
2108 PPCODE:
2109 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2110 XPUSHs (jsonstr);
2111
2112void decode_prefix (JSON *self, SV *jsonstr)
2113 PPCODE:
2114{
2115 SV *sv;
2116 char *offset;
2117 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
2118 EXTEND (SP, 2);
2119 PUSHs (sv);
2120 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
2121}
2122
2123void incr_parse (JSON *self, SV *jsonstr = 0)
2124 PPCODE:
2125{
2126 if (!self->incr_text)
2127 self->incr_text = newSVpvn ("", 0);
2128
2129 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2130 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2131 if (DECODE_WANTS_OCTETS (self))
2132 {
2133 if (self->incr_pos)
2134 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2135 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2136
2137 sv_utf8_downgrade (self->incr_text, 0);
2138 }
2139 else
2140 {
2141 sv_utf8_upgrade (self->incr_text);
2142
2143 if (self->incr_pos)
2144 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2145 - (U8 *)SvPVX (self->incr_text);
2146 }
2147
2148 // append data, if any
2149 if (jsonstr)
2150 {
2151 /* make sure both strings have same encoding */
2152 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2153 if (SvUTF8 (jsonstr))
2154 sv_utf8_downgrade (jsonstr, 0);
2155 else
2156 sv_utf8_upgrade (jsonstr);
2157
2158 /* and then just blindly append */
2159 {
2160 STRLEN len;
2161 const char *str = SvPV (jsonstr, len);
2162 STRLEN cur = SvCUR (self->incr_text);
2163
2164 if (SvLEN (self->incr_text) <= cur + len)
2165 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
2166
2167 Move (str, SvEND (self->incr_text), len, char);
2168 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2169 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2170 }
2171 }
2172
2173 if (GIMME_V != G_VOID)
2174 do
2175 {
2176 SV *sv;
2177 char *offset;
2178
2179 if (!INCR_DONE (self))
2180 {
2181 incr_parse (self);
2182
2183 if (self->incr_pos > self->max_size && self->max_size)
2184 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2185 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2186
2187 if (!INCR_DONE (self))
2188 {
2189 // as an optimisation, do not accumulate white space in the incr buffer
2190 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2191 {
2192 self->incr_pos = 0;
2193 SvCUR_set (self->incr_text, 0);
2194 }
2195
2196 break;
2197 }
2198 }
2199
2200 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2201 XPUSHs (sv);
2202
2203 self->incr_pos -= offset - SvPVX (self->incr_text);
2204 self->incr_nest = 0;
2205 self->incr_mode = 0;
2206
2207 sv_chop (self->incr_text, offset);
2208 }
2209 while (GIMME_V == G_ARRAY);
2210}
2211
2212SV *incr_text (JSON *self)
2213 ATTRS: lvalue
1289 CODE: 2214 CODE:
1290{ 2215{
1291 UV *uv = SvJSON (self); 2216 if (self->incr_pos)
1292 UV log2 = 0; 2217 croak ("incr_text can not be called when the incremental parser already started parsing");
1293 2218
1294 if (max_size > 0x80000000UL) max_size = 0x80000000UL; 2219 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1295 if (max_size == 1) max_size = 2;
1296
1297 while ((1UL << log2) < max_size)
1298 ++log2;
1299
1300 *uv = *uv & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1301
1302 RETVAL = newSVsv (self);
1303} 2220}
1304 OUTPUT: 2221 OUTPUT:
1305 RETVAL 2222 RETVAL
1306 2223
2224void incr_skip (JSON *self)
2225 CODE:
2226{
2227 if (self->incr_pos)
2228 {
2229 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2230 self->incr_pos = 0;
2231 self->incr_nest = 0;
2232 self->incr_mode = 0;
2233 }
2234}
2235
2236void incr_reset (JSON *self)
2237 CODE:
2238{
2239 SvREFCNT_dec (self->incr_text);
2240 self->incr_text = 0;
2241 self->incr_pos = 0;
2242 self->incr_nest = 0;
2243 self->incr_mode = 0;
2244}
2245
2246void DESTROY (JSON *self)
2247 CODE:
2248 SvREFCNT_dec (self->cb_sk_object);
2249 SvREFCNT_dec (self->cb_object);
2250 SvREFCNT_dec (self->incr_text);
2251
2252PROTOTYPES: ENABLE
2253
1307void encode (SV *self, SV *scalar) 2254void encode_json (SV *scalar)
1308 PPCODE: 2255 PPCODE:
1309 XPUSHs (encode_json (scalar, *SvJSON (self))); 2256{
2257 JSON json;
2258 json_init (&json);
2259 json.flags |= F_UTF8;
2260 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2261 XPUSHs (scalar);
2262}
1310 2263
1311void decode (SV *self, SV *jsonstr) 2264void decode_json (SV *jsonstr)
1312 PPCODE: 2265 PPCODE:
1313 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1314
1315void decode_prefix (SV *self, SV *jsonstr)
1316 PPCODE:
1317{ 2266{
1318 UV offset; 2267 JSON json;
1319 EXTEND (SP, 2); 2268 json_init (&json);
1320 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 2269 json.flags |= F_UTF8;
1321 PUSHs (sv_2mortal (newSVuv (offset))); 2270 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2271 XPUSHs (jsonstr);
1322} 2272}
1323 2273
1324PROTOTYPES: ENABLE
1325
1326void to_json (SV *scalar)
1327 ALIAS:
1328 objToJson = 0
1329 PPCODE:
1330 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1331
1332void from_json (SV *jsonstr)
1333 ALIAS:
1334 jsonToObj = 0
1335 PPCODE:
1336 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1337

Diff Legend

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