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 1.124 by root, Mon Nov 24 18:42:51 2014 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
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#else
85# define JSON_SLOW 0
86# define JSON_STASH json_stash
87#endif
88
89// the amount of HEs to allocate on the stack, when sorting keys
90#define STACK_HES 64
91
61static HV *json_stash, *json_boolean_stash; // JSON::XS:: 92static HV *json_stash, *types_boolean_stash; // JSON::XS::
62static SV *json_true, *json_false; 93static SV *types_true, *types_false, *sv_json;
94
95enum {
96 INCR_M_WS = 0, // initial whitespace skipping, must be 0
97 INCR_M_STR, // inside string
98 INCR_M_BS, // inside backslash
99 INCR_M_C0, // inside comment in initial whitespace sequence
100 INCR_M_C1, // inside comment in other places
101 INCR_M_JSON // outside anything, count nesting
102};
103
104#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
105
106typedef struct {
107 U32 flags;
108 U32 max_depth;
109 STRLEN max_size;
110
111 SV *cb_object;
112 HV *cb_sk_object;
113
114 // for the incremental parser
115 SV *incr_text; // the source text so far
116 STRLEN incr_pos; // the current offset into the text
117 int incr_nest; // {[]}-nesting level
118 unsigned char incr_mode;
119} JSON;
120
121INLINE void
122json_init (JSON *json)
123{
124 Zero (json, 1, JSON);
125 json->max_depth = 512;
126}
63 127
64///////////////////////////////////////////////////////////////////////////// 128/////////////////////////////////////////////////////////////////////////////
65// utility functions 129// utility functions
66 130
67static UV * 131INLINE SV *
68SvJSON (SV *sv) 132get_bool (const char *name)
69{ 133{
70 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 134 SV *sv = get_sv (name, 1);
71 croak ("object is not of type JSON::XS");
72 135
73 return &SvUVX (SvRV (sv)); 136 SvREADONLY_on (sv);
74} 137 SvREADONLY_on (SvRV (sv));
75 138
76static void 139 return sv;
140}
141
142INLINE void
77shrink (SV *sv) 143shrink (SV *sv)
78{ 144{
79 sv_utf8_downgrade (sv, 1); 145 sv_utf8_downgrade (sv, 1);
146
80 if (SvLEN (sv) > SvCUR (sv) + 1) 147 if (SvLEN (sv) > SvCUR (sv) + 1)
81 { 148 {
82#ifdef SvPV_shrink_to_cur 149#ifdef SvPV_shrink_to_cur
83 SvPV_shrink_to_cur (sv); 150 SvPV_shrink_to_cur (sv);
84#elif defined (SvPV_renew) 151#elif defined (SvPV_renew)
… …
90// decode an utf-8 character and return it, or (UV)-1 in 157// decode an utf-8 character and return it, or (UV)-1 in
91// case of an error. 158// case of an error.
92// we special-case "safe" characters from U+80 .. U+7FF, 159// we special-case "safe" characters from U+80 .. U+7FF,
93// but use the very good perl function to parse anything else. 160// but use the very good perl function to parse anything else.
94// note that we never call this function for a ascii codepoints 161// note that we never call this function for a ascii codepoints
95inline UV 162INLINE UV
96decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 163decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
97{ 164{
98 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 165 if (expect_true (len >= 2
99 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 166 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
100 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 167 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
101 { 168 {
102 *clen = 2; 169 *clen = 2;
103 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 170 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
104 } 171 }
105 else 172 else
106 { 173 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
107 *clen = (STRLEN)-1; 174}
108 return (UV)-1; 175
176// likewise for encoding, also never called for ascii codepoints
177// this function takes advantage of this fact, although current gccs
178// seem to optimise the check for >= 0x80 away anyways
179INLINE unsigned char *
180encode_utf8 (unsigned char *s, UV ch)
181{
182 if (expect_false (ch < 0x000080))
183 *s++ = ch;
184 else if (expect_true (ch < 0x000800))
185 *s++ = 0xc0 | ( ch >> 6),
186 *s++ = 0x80 | ( ch & 0x3f);
187 else if ( ch < 0x010000)
188 *s++ = 0xe0 | ( ch >> 12),
189 *s++ = 0x80 | ((ch >> 6) & 0x3f),
190 *s++ = 0x80 | ( ch & 0x3f);
191 else if ( ch < 0x110000)
192 *s++ = 0xf0 | ( ch >> 18),
193 *s++ = 0x80 | ((ch >> 12) & 0x3f),
194 *s++ = 0x80 | ((ch >> 6) & 0x3f),
195 *s++ = 0x80 | ( ch & 0x3f);
196
197 return s;
198}
199
200// convert offset pointer to character index, sv must be string
201static STRLEN
202ptr_to_index (SV *sv, char *offset)
203{
204 return SvUTF8 (sv)
205 ? utf8_distance (offset, SvPVX (sv))
206 : offset - SvPVX (sv);
207}
208
209/////////////////////////////////////////////////////////////////////////////
210// fp hell
211
212// scan a group of digits, and a trailing exponent
213static void
214json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
215{
216 UV uaccum = 0;
217 int eaccum = 0;
218
219 // if we recurse too deep, skip all remaining digits
220 // to avoid a stack overflow attack
221 if (expect_false (--maxdepth <= 0))
222 while (((U8)*s - '0') < 10)
223 ++s;
224
225 for (;;)
109 } 226 {
110} 227 U8 dig = (U8)*s - '0';
111 228
229 if (expect_false (dig >= 10))
230 {
231 if (dig == (U8)((U8)'.' - (U8)'0'))
232 {
233 ++s;
234 json_atof_scan1 (s, accum, expo, 1, maxdepth);
235 }
236 else if ((dig | ' ') == 'e' - '0')
237 {
238 int exp2 = 0;
239 int neg = 0;
240
241 ++s;
242
243 if (*s == '-')
244 {
245 ++s;
246 neg = 1;
247 }
248 else if (*s == '+')
249 ++s;
250
251 while ((dig = (U8)*s - '0') < 10)
252 exp2 = exp2 * 10 + *s++ - '0';
253
254 *expo += neg ? -exp2 : exp2;
255 }
256
257 break;
258 }
259
260 ++s;
261
262 uaccum = uaccum * 10 + dig;
263 ++eaccum;
264
265 // if we have too many digits, then recurse for more
266 // we actually do this for rather few digits
267 if (uaccum >= (UV_MAX - 9) / 10)
268 {
269 if (postdp) *expo -= eaccum;
270 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
271 if (postdp) *expo += eaccum;
272
273 break;
274 }
275 }
276
277 // this relies greatly on the quality of the pow ()
278 // implementation of the platform, but a good
279 // implementation is hard to beat.
280 // (IEEE 754 conformant ones are required to be exact)
281 if (postdp) *expo -= eaccum;
282 *accum += uaccum * Perl_pow (10., *expo);
283 *expo += eaccum;
284}
285
286static NV
287json_atof (const char *s)
288{
289 NV accum = 0.;
290 int expo = 0;
291 int neg = 0;
292
293 if (*s == '-')
294 {
295 ++s;
296 neg = 1;
297 }
298
299 // a recursion depth of ten gives us >>500 bits
300 json_atof_scan1 (s, &accum, &expo, 0, 10);
301
302 return neg ? -accum : accum;
303}
112///////////////////////////////////////////////////////////////////////////// 304/////////////////////////////////////////////////////////////////////////////
113// encoder 305// encoder
114 306
115// structure used for encoding JSON 307// structure used for encoding JSON
116typedef struct 308typedef struct
117{ 309{
118 char *cur; // SvPVX (sv) + current output position 310 char *cur; // SvPVX (sv) + current output position
119 char *end; // SvEND (sv) 311 char *end; // SvEND (sv)
120 SV *sv; // result scalar 312 SV *sv; // result scalar
121 U32 flags; // F_* 313 JSON json;
122 U32 indent; // indentation level 314 U32 indent; // indentation level
123 U32 maxdepth; // max. indentation/recursion level 315 UV limit; // escape character values >= this value when encoding
124} enc_t; 316} enc_t;
125 317
126inline void 318INLINE void
127need (enc_t *enc, STRLEN len) 319need (enc_t *enc, STRLEN len)
128{ 320{
129 if (expect_false (enc->cur + len >= enc->end)) 321 if (expect_false (enc->cur + len >= enc->end))
130 { 322 {
131 STRLEN cur = enc->cur - SvPVX (enc->sv); 323 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
132 SvGROW (enc->sv, cur + len + 1); 324 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
133 enc->cur = SvPVX (enc->sv) + cur; 325 enc->cur = SvPVX (enc->sv) + cur;
134 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 326 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
135 } 327 }
136} 328}
137 329
138inline void 330INLINE void
139encode_ch (enc_t *enc, char ch) 331encode_ch (enc_t *enc, char ch)
140{ 332{
141 need (enc, 1); 333 need (enc, 1);
142 *enc->cur++ = ch; 334 *enc->cur++ = ch;
143} 335}
… …
197 { 389 {
198 uch = ch; 390 uch = ch;
199 clen = 1; 391 clen = 1;
200 } 392 }
201 393
202 if (uch > 0x10FFFFUL) 394 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 { 395 {
207 if (uch > 0xFFFFUL) 396 if (uch >= 0x10000UL)
208 { 397 {
398 if (uch >= 0x110000UL)
399 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
400
209 need (enc, len += 11); 401 need (enc, len += 11);
210 sprintf (enc->cur, "\\u%04x\\u%04x", 402 sprintf (enc->cur, "\\u%04x\\u%04x",
211 (int)((uch - 0x10000) / 0x400 + 0xD800), 403 (int)((uch - 0x10000) / 0x400 + 0xD800),
212 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 404 (int)((uch - 0x10000) % 0x400 + 0xDC00));
213 enc->cur += 12; 405 enc->cur += 12;
214 } 406 }
215 else 407 else
216 { 408 {
217 static char hexdigit [16] = "0123456789abcdef";
218 need (enc, len += 5); 409 need (enc, len += 5);
219 *enc->cur++ = '\\'; 410 *enc->cur++ = '\\';
220 *enc->cur++ = 'u'; 411 *enc->cur++ = 'u';
221 *enc->cur++ = hexdigit [ uch >> 12 ]; 412 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
222 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 413 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
223 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 414 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
224 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 415 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
225 } 416 }
226 417
227 str += clen; 418 str += clen;
228 } 419 }
229 else if (enc->flags & F_LATIN1) 420 else if (enc->json.flags & F_LATIN1)
230 { 421 {
231 *enc->cur++ = uch; 422 *enc->cur++ = uch;
232 str += clen; 423 str += clen;
233 } 424 }
234 else if (is_utf8) 425 else if (is_utf8)
… …
241 while (--clen); 432 while (--clen);
242 } 433 }
243 else 434 else
244 { 435 {
245 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 436 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
246 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 437 enc->cur = encode_utf8 (enc->cur, uch);
247 ++str; 438 ++str;
248 } 439 }
249 } 440 }
250 } 441 }
251 } 442 }
252 443
253 --len; 444 --len;
254 } 445 }
255} 446}
256 447
257inline void 448INLINE void
258encode_indent (enc_t *enc) 449encode_indent (enc_t *enc)
259{ 450{
260 if (enc->flags & F_INDENT) 451 if (enc->json.flags & F_INDENT)
261 { 452 {
262 int spaces = enc->indent * INDENT_STEP; 453 int spaces = enc->indent * INDENT_STEP;
263 454
264 need (enc, spaces); 455 need (enc, spaces);
265 memset (enc->cur, ' ', spaces); 456 memset (enc->cur, ' ', spaces);
266 enc->cur += spaces; 457 enc->cur += spaces;
267 } 458 }
268} 459}
269 460
270inline void 461INLINE void
271encode_space (enc_t *enc) 462encode_space (enc_t *enc)
272{ 463{
273 need (enc, 1); 464 need (enc, 1);
274 encode_ch (enc, ' '); 465 encode_ch (enc, ' ');
275} 466}
276 467
277inline void 468INLINE void
278encode_nl (enc_t *enc) 469encode_nl (enc_t *enc)
279{ 470{
280 if (enc->flags & F_INDENT) 471 if (enc->json.flags & F_INDENT)
281 { 472 {
282 need (enc, 1); 473 need (enc, 1);
283 encode_ch (enc, '\n'); 474 encode_ch (enc, '\n');
284 } 475 }
285} 476}
286 477
287inline void 478INLINE void
288encode_comma (enc_t *enc) 479encode_comma (enc_t *enc)
289{ 480{
290 encode_ch (enc, ','); 481 encode_ch (enc, ',');
291 482
292 if (enc->flags & F_INDENT) 483 if (enc->json.flags & F_INDENT)
293 encode_nl (enc); 484 encode_nl (enc);
294 else if (enc->flags & F_SPACE_AFTER) 485 else if (enc->json.flags & F_SPACE_AFTER)
295 encode_space (enc); 486 encode_space (enc);
296} 487}
297 488
298static void encode_sv (enc_t *enc, SV *sv); 489static void encode_sv (enc_t *enc, SV *sv);
299 490
300static void 491static void
301encode_av (enc_t *enc, AV *av) 492encode_av (enc_t *enc, AV *av)
302{ 493{
303 int i, len = av_len (av); 494 int i, len = av_len (av);
304 495
305 if (enc->indent >= enc->maxdepth) 496 if (enc->indent >= enc->json.max_depth)
306 croak ("data structure too deep (hit recursion limit)"); 497 croak (ERR_NESTING_EXCEEDED);
307 498
308 encode_ch (enc, '['); encode_nl (enc); 499 encode_ch (enc, '[');
309 ++enc->indent;
310 500
501 if (len >= 0)
502 {
503 encode_nl (enc); ++enc->indent;
504
311 for (i = 0; i <= len; ++i) 505 for (i = 0; i <= len; ++i)
312 { 506 {
507 SV **svp = av_fetch (av, i, 0);
508
313 encode_indent (enc); 509 encode_indent (enc);
314 encode_sv (enc, *av_fetch (av, i, 0));
315 510
511 if (svp)
512 encode_sv (enc, *svp);
513 else
514 encode_str (enc, "null", 4, 0);
515
316 if (i < len) 516 if (i < len)
317 encode_comma (enc); 517 encode_comma (enc);
318 } 518 }
319 519
520 encode_nl (enc); --enc->indent; encode_indent (enc);
521 }
522
320 encode_nl (enc); 523 encode_ch (enc, ']');
321
322 --enc->indent;
323 encode_indent (enc); encode_ch (enc, ']');
324} 524}
325 525
326static void 526static void
327encode_he (enc_t *enc, HE *he) 527encode_hk (enc_t *enc, HE *he)
328{ 528{
329 encode_ch (enc, '"'); 529 encode_ch (enc, '"');
330 530
331 if (HeKLEN (he) == HEf_SVKEY) 531 if (HeKLEN (he) == HEf_SVKEY)
332 { 532 {
333 SV *sv = HeSVKEY (he); 533 SV *sv = HeSVKEY (he);
334 STRLEN len; 534 STRLEN len;
335 char *str; 535 char *str;
336 536
337 SvGETMAGIC (sv); 537 SvGETMAGIC (sv);
338 str = SvPV (sv, len); 538 str = SvPV (sv, len);
339 539
340 encode_str (enc, str, len, SvUTF8 (sv)); 540 encode_str (enc, str, len, SvUTF8 (sv));
341 } 541 }
342 else 542 else
343 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 543 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
344 544
345 encode_ch (enc, '"'); 545 encode_ch (enc, '"');
346 546
347 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 547 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
348 encode_ch (enc, ':'); 548 encode_ch (enc, ':');
349 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 549 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
350 encode_sv (enc, HeVAL (he));
351} 550}
352 551
353// compare hash entries, used when all keys are bytestrings 552// compare hash entries, used when all keys are bytestrings
354static int 553static int
355he_cmp_fast (const void *a_, const void *b_) 554he_cmp_fast (const void *a_, const void *b_)
… …
360 HE *b = *(HE **)b_; 559 HE *b = *(HE **)b_;
361 560
362 STRLEN la = HeKLEN (a); 561 STRLEN la = HeKLEN (a);
363 STRLEN lb = HeKLEN (b); 562 STRLEN lb = HeKLEN (b);
364 563
365 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 564 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
366 cmp = la - lb; 565 cmp = lb - la;
367 566
368 return cmp; 567 return cmp;
369} 568}
370 569
371// compare hash entries, used when some keys are sv's or utf-x 570// compare hash entries, used when some keys are sv's or utf-x
372static int 571static int
373he_cmp_slow (const void *a, const void *b) 572he_cmp_slow (const void *a, const void *b)
374{ 573{
375 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 574 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
376} 575}
377 576
378static void 577static void
379encode_hv (enc_t *enc, HV *hv) 578encode_hv (enc_t *enc, HV *hv)
380{ 579{
381 int count, i; 580 HE *he;
382 581
383 if (enc->indent >= enc->maxdepth) 582 if (enc->indent >= enc->json.max_depth)
384 croak ("data structure too deep (hit recursion limit)"); 583 croak (ERR_NESTING_EXCEEDED);
385 584
386 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 585 encode_ch (enc, '{');
387 586
388 if ((count = hv_iterinit (hv)))
389 {
390 // for canonical output we have to sort by keys first 587 // for canonical output we have to sort by keys first
391 // actually, this is mostly due to the stupid so-called 588 // actually, this is mostly due to the stupid so-called
392 // security workaround added somewhere in 5.8.x. 589 // security workaround added somewhere in 5.8.x
393 // that randomises hash orderings 590 // that randomises hash orderings
394 if (enc->flags & F_CANONICAL) 591 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
592 {
593 int count = hv_iterinit (hv);
594
595 if (SvMAGICAL (hv))
395 { 596 {
597 // need to count by iterating. could improve by dynamically building the vector below
598 // but I don't care for the speed of this special case.
599 // note also that we will run into undefined behaviour when the two iterations
600 // do not result in the same count, something I might care for in some later release.
601
602 count = 0;
603 while (hv_iternext (hv))
604 ++count;
605
606 hv_iterinit (hv);
607 }
608
609 if (count)
610 {
396 int fast = 1; 611 int i, fast = 1;
397 HE *he; 612 HE *hes_stack [STACK_HES];
398#if defined(__BORLANDC__) || defined(_MSC_VER) 613 HE **hes = hes_stack;
399 HE **hes = _alloca (count * sizeof (HE)); 614
400#else 615 // allocate larger arrays on the heap
401 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 616 if (count > STACK_HES)
402#endif 617 {
618 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
619 hes = (HE **)SvPVX (sv);
620 }
403 621
404 i = 0; 622 i = 0;
405 while ((he = hv_iternext (hv))) 623 while ((he = hv_iternext (hv)))
406 { 624 {
407 hes [i++] = he; 625 hes [i++] = he;
… …
429 647
430 FREETMPS; 648 FREETMPS;
431 LEAVE; 649 LEAVE;
432 } 650 }
433 651
434 for (i = 0; i < count; ++i) 652 encode_nl (enc); ++enc->indent;
653
654 while (count--)
435 { 655 {
436 encode_indent (enc); 656 encode_indent (enc);
657 he = hes [count];
437 encode_he (enc, hes [i]); 658 encode_hk (enc, he);
659 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
438 660
439 if (i < count - 1) 661 if (count)
440 encode_comma (enc); 662 encode_comma (enc);
441 } 663 }
442 664
443 encode_nl (enc); 665 encode_nl (enc); --enc->indent; encode_indent (enc);
444 } 666 }
667 }
445 else 668 else
669 {
670 if (hv_iterinit (hv) || SvMAGICAL (hv))
671 if ((he = hv_iternext (hv)))
446 { 672 {
447 HE *he = hv_iternext (hv); 673 encode_nl (enc); ++enc->indent;
448 674
449 for (;;) 675 for (;;)
450 { 676 {
451 encode_indent (enc); 677 encode_indent (enc);
452 encode_he (enc, he); 678 encode_hk (enc, he);
679 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
453 680
454 if (!(he = hv_iternext (hv))) 681 if (!(he = hv_iternext (hv)))
455 break; 682 break;
456 683
457 encode_comma (enc); 684 encode_comma (enc);
458 } 685 }
459 686
460 encode_nl (enc); 687 encode_nl (enc); --enc->indent; encode_indent (enc);
461 } 688 }
462 } 689 }
463 690
464 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 691 encode_ch (enc, '}');
465} 692}
466 693
467// encode objects, arrays and special \0=false and \1=true values. 694// encode objects, arrays and special \0=false and \1=true values.
468static void 695static void
469encode_rv (enc_t *enc, SV *sv) 696encode_rv (enc_t *enc, SV *sv)
470{ 697{
471 svtype svt; 698 svtype svt;
699 GV *method;
472 700
473 SvGETMAGIC (sv); 701 SvGETMAGIC (sv);
474 svt = SvTYPE (sv); 702 svt = SvTYPE (sv);
475 703
476 if (expect_false (SvOBJECT (sv))) 704 if (expect_false (SvOBJECT (sv)))
477 { 705 {
478 if (SvSTASH (sv) == json_boolean_stash) 706 HV *boolean_stash = !JSON_SLOW || types_boolean_stash
707 ? types_boolean_stash
708 : gv_stashpv ("Types::Serialiser::Boolean", 1);
709 HV *stash = SvSTASH (sv);
710
711 if (stash == boolean_stash)
479 { 712 {
480 if (SvIV (sv) == 0) 713 if (SvIV (sv))
714 encode_str (enc, "true", 4, 0);
715 else
481 encode_str (enc, "false", 5, 0); 716 encode_str (enc, "false", 5, 0);
482 else
483 encode_str (enc, "true", 4, 0);
484 } 717 }
718 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
719 {
720 int count;
721 dSP;
722
723 ENTER; SAVETMPS; PUSHMARK (SP);
724 EXTEND (SP, 2);
725 // we re-bless the reference to get overload and other niceties right
726 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
727 PUSHs (sv_json);
728
729 PUTBACK;
730 count = call_sv ((SV *)GvCV (method), G_ARRAY);
731 SPAGAIN;
732
733 // catch this surprisingly common error
734 if (SvROK (TOPs) && SvRV (TOPs) == sv)
735 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
736
737 encode_ch (enc, '(');
738 encode_ch (enc, '"');
739 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
740 encode_ch (enc, '"');
741 encode_ch (enc, ')');
742 encode_ch (enc, '[');
743
744 while (count)
745 {
746 encode_sv (enc, SP[1 - count--]);
747
748 if (count)
749 encode_ch (enc, ',');
750 }
751
752 encode_ch (enc, ']');
753
754 PUTBACK;
755
756 FREETMPS; LEAVE;
757 }
758 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
759 {
760 dSP;
761
762 ENTER; SAVETMPS; PUSHMARK (SP);
763 // we re-bless the reference to get overload and other niceties right
764 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
765
766 // calling with G_SCALAR ensures that we always get a 1 return value
767 PUTBACK;
768 call_sv ((SV *)GvCV (method), G_SCALAR);
769 SPAGAIN;
770
771 // catch this surprisingly common error
772 if (SvROK (TOPs) && SvRV (TOPs) == sv)
773 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
774
775 sv = POPs;
776 PUTBACK;
777
778 encode_sv (enc, sv);
779
780 FREETMPS; LEAVE;
781 }
782 else if (enc->json.flags & F_ALLOW_BLESSED)
783 encode_str (enc, "null", 4, 0);
485 else 784 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", 785 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)))); 786 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
528 }
529 } 787 }
530 else if (svt == SVt_PVHV) 788 else if (svt == SVt_PVHV)
531 encode_hv (enc, (HV *)sv); 789 encode_hv (enc, (HV *)sv);
532 else if (svt == SVt_PVAV) 790 else if (svt == SVt_PVAV)
533 encode_av (enc, (AV *)sv); 791 encode_av (enc, (AV *)sv);
534 else if (svt < SVt_PVAV) 792 else if (svt < SVt_PVAV)
535 { 793 {
536 if (SvNIOK (sv) && SvIV (sv) == 0) 794 STRLEN len = 0;
795 char *pv = svt ? SvPV (sv, len) : 0;
796
797 if (len == 1 && *pv == '1')
798 encode_str (enc, "true", 4, 0);
799 else if (len == 1 && *pv == '0')
537 encode_str (enc, "false", 5, 0); 800 encode_str (enc, "false", 5, 0);
538 else if (SvNIOK (sv) && SvIV (sv) == 1) 801 else if (enc->json.flags & F_ALLOW_UNKNOWN)
539 encode_str (enc, "true", 4, 0); 802 encode_str (enc, "null", 4, 0);
540 else 803 else
541 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 804 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
542 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 805 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
543 } 806 }
807 else if (enc->json.flags & F_ALLOW_UNKNOWN)
808 encode_str (enc, "null", 4, 0);
544 else 809 else
545 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 810 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
546 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 811 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
547} 812}
548 813
… …
566 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 831 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
567 enc->cur += strlen (enc->cur); 832 enc->cur += strlen (enc->cur);
568 } 833 }
569 else if (SvIOKp (sv)) 834 else if (SvIOKp (sv))
570 { 835 {
571 // we assume we can always read an IV as a UV 836 // we assume we can always read an IV as a UV and vice versa
572 if (SvUV (sv) & ~(UV)0x7fff) 837 // we assume two's complement
573 { 838 // we assume no aliasing issues in the union
574 // large integer, use the (rather slow) snprintf way. 839 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
575 need (enc, sizeof (UV) * 3); 840 : 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 { 841 {
583 // optimise the "small number case" 842 // optimise the "small number case"
584 // code will likely be branchless and use only a single multiplication 843 // code will likely be branchless and use only a single multiplication
844 // works for numbers up to 59074
585 I32 i = SvIV (sv); 845 I32 i = SvIVX (sv);
586 U32 u; 846 U32 u;
587 char digit, nz = 0; 847 char digit, nz = 0;
588 848
589 need (enc, 6); 849 need (enc, 6);
590 850
… …
596 856
597 // now output digit by digit, each time masking out the integer part 857 // 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, 858 // and multiplying by 5 while moving the decimal point one to the right,
599 // resulting in a net multiplication by 10. 859 // resulting in a net multiplication by 10.
600 // we always write the digit to memory but conditionally increment 860 // we always write the digit to memory but conditionally increment
601 // the pointer, to ease the usage of conditional move instructions. 861 // 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; 862 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; 863 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; 864 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; 865 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' 866 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
607 } 867 }
868 else
869 {
870 // large integer, use the (rather slow) snprintf way.
871 need (enc, IVUV_MAXCHARS);
872 enc->cur +=
873 SvIsUV(sv)
874 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
875 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
876 }
608 } 877 }
609 else if (SvROK (sv)) 878 else if (SvROK (sv))
610 encode_rv (enc, SvRV (sv)); 879 encode_rv (enc, SvRV (sv));
611 else if (!SvOK (sv)) 880 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
612 encode_str (enc, "null", 4, 0); 881 encode_str (enc, "null", 4, 0);
613 else 882 else
614 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 883 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
615 SvPV_nolen (sv), SvFLAGS (sv)); 884 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
616} 885}
617 886
618static SV * 887static SV *
619encode_json (SV *scalar, U32 flags) 888encode_json (SV *scalar, JSON *json)
620{ 889{
621 enc_t enc; 890 enc_t enc;
622 891
623 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 892 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
624 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 893 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
625 894
626 enc.flags = flags; 895 enc.json = *json;
627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 896 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
628 enc.cur = SvPVX (enc.sv); 897 enc.cur = SvPVX (enc.sv);
629 enc.end = SvEND (enc.sv); 898 enc.end = SvEND (enc.sv);
630 enc.indent = 0; 899 enc.indent = 0;
631 enc.maxdepth = DEC_DEPTH (flags); 900 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
901 : enc.json.flags & F_LATIN1 ? 0x000100UL
902 : 0x110000UL;
632 903
633 SvPOK_only (enc.sv); 904 SvPOK_only (enc.sv);
634 encode_sv (&enc, scalar); 905 encode_sv (&enc, scalar);
906 encode_nl (&enc);
635 907
636 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 908 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
637 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 909 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
638 910
639 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 911 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
640 SvUTF8_on (enc.sv); 912 SvUTF8_on (enc.sv);
641 913
642 if (enc.flags & F_SHRINK) 914 if (enc.json.flags & F_SHRINK)
643 shrink (enc.sv); 915 shrink (enc.sv);
644 916
645 return enc.sv; 917 return enc.sv;
646} 918}
647 919
… …
652typedef struct 924typedef struct
653{ 925{
654 char *cur; // current parser pointer 926 char *cur; // current parser pointer
655 char *end; // end of input string 927 char *end; // end of input string
656 const char *err; // parse error, if != 0 928 const char *err; // parse error, if != 0
657 U32 flags; // F_* 929 JSON json;
658 U32 depth; // recursion depth 930 U32 depth; // recursion depth
659 U32 maxdepth; // recursion depth limit 931 U32 maxdepth; // recursion depth limit
660} dec_t; 932} dec_t;
661 933
662inline void 934INLINE void
935decode_comment (dec_t *dec)
936{
937 // only '#'-style comments allowed a.t.m.
938
939 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
940 ++dec->cur;
941}
942
943INLINE void
663decode_ws (dec_t *dec) 944decode_ws (dec_t *dec)
664{ 945{
665 for (;;) 946 for (;;)
666 { 947 {
667 char ch = *dec->cur; 948 char ch = *dec->cur;
668 949
669 if (ch > 0x20 950 if (ch > 0x20)
951 {
952 if (expect_false (ch == '#'))
953 {
954 if (dec->json.flags & F_RELAXED)
955 decode_comment (dec);
956 else
957 break;
958 }
959 else
960 break;
961 }
670 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 962 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
671 break; 963 break; // parse error, but let higher level handle it, gives better error messages
672 964
673 ++dec->cur; 965 ++dec->cur;
674 } 966 }
675} 967}
676 968
… …
680 if (*dec->cur != ch) \ 972 if (*dec->cur != ch) \
681 ERR (# ch " expected"); \ 973 ERR (# ch " expected"); \
682 ++dec->cur; \ 974 ++dec->cur; \
683 SE 975 SE
684 976
685#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 977#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
686#define DEC_DEC_DEPTH --dec->depth 978#define DEC_DEC_DEPTH --dec->depth
687 979
688static SV *decode_sv (dec_t *dec); 980static SV *decode_sv (dec_t *dec);
689 981
690static signed char decode_hexdigit[256]; 982static signed char decode_hexdigit[256];
… …
782 1074
783 if (hi >= 0x80) 1075 if (hi >= 0x80)
784 { 1076 {
785 utf8 = 1; 1077 utf8 = 1;
786 1078
787 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1079 cur = encode_utf8 (cur, hi);
788 } 1080 }
789 else 1081 else
790 *cur++ = hi; 1082 *cur++ = hi;
791 } 1083 }
792 break; 1084 break;
… …
794 default: 1086 default:
795 --dec_cur; 1087 --dec_cur;
796 ERR ("illegal backslash escape sequence in string"); 1088 ERR ("illegal backslash escape sequence in string");
797 } 1089 }
798 } 1090 }
799 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1091 else if (expect_true (ch >= 0x20 && ch < 0x80))
800 *cur++ = ch; 1092 *cur++ = ch;
801 else if (ch >= 0x80) 1093 else if (ch >= 0x80)
802 { 1094 {
803 STRLEN clen; 1095 STRLEN clen;
804 UV uch;
805 1096
806 --dec_cur; 1097 --dec_cur;
807 1098
808 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1099 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
809 if (clen == (STRLEN)-1) 1100 if (clen == (STRLEN)-1)
810 ERR ("malformed UTF-8 character in JSON string"); 1101 ERR ("malformed UTF-8 character in JSON string");
811 1102
812 do 1103 do
813 *cur++ = *dec_cur++; 1104 *cur++ = *dec_cur++;
814 while (--clen); 1105 while (--clen);
815 1106
816 utf8 = 1; 1107 utf8 = 1;
817 } 1108 }
1109 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1110 *cur++ = ch;
818 else 1111 else
819 { 1112 {
820 --dec_cur; 1113 --dec_cur;
821 1114
822 if (!ch) 1115 if (!ch)
… …
830 { 1123 {
831 STRLEN len = cur - buf; 1124 STRLEN len = cur - buf;
832 1125
833 if (sv) 1126 if (sv)
834 { 1127 {
835 SvGROW (sv, SvCUR (sv) + len + 1); 1128 STRLEN cur = SvCUR (sv);
1129
1130 if (SvLEN (sv) <= cur + len)
1131 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1132
836 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1133 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
837 SvCUR_set (sv, SvCUR (sv) + len); 1134 SvCUR_set (sv, SvCUR (sv) + len);
838 } 1135 }
839 else 1136 else
840 sv = newSVpvn (buf, len); 1137 sv = newSVpvn (buf, len);
… …
925 is_nv = 1; 1222 is_nv = 1;
926 } 1223 }
927 1224
928 if (!is_nv) 1225 if (!is_nv)
929 { 1226 {
1227 int len = dec->cur - start;
1228
930 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1229 // special case the rather common 1..5-digit-int case
931 if (*start == '-') 1230 if (*start == '-')
932 switch (dec->cur - start) 1231 switch (len)
933 { 1232 {
934 case 2: return newSViv (-( start [1] - '0' * 1)); 1233 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
935 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1234 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)); 1235 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)); 1236 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1237 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
938 } 1238 }
939 else 1239 else
940 switch (dec->cur - start) 1240 switch (len)
941 { 1241 {
942 case 1: return newSViv ( start [0] - '0' * 1); 1242 case 1: return newSViv ( start [0] - '0' * 1);
943 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1243 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); 1244 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); 1245 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1246 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
946 } 1247 }
947 1248
948 { 1249 {
949 UV uv; 1250 UV uv;
950 int numtype = grok_number (start, dec->cur - start, &uv); 1251 int numtype = grok_number (start, len, &uv);
951 if (numtype & IS_NUMBER_IN_UV) 1252 if (numtype & IS_NUMBER_IN_UV)
952 if (numtype & IS_NUMBER_NEG) 1253 if (numtype & IS_NUMBER_NEG)
953 { 1254 {
954 if (uv < (UV)IV_MIN) 1255 if (uv < (UV)IV_MIN)
955 return newSViv (-(IV)uv); 1256 return newSViv (-(IV)uv);
956 } 1257 }
957 else 1258 else
958 return newSVuv (uv); 1259 return newSVuv (uv);
959
960 // here would likely be the place for bigint support
961 } 1260 }
962 }
963 1261
964 // if we ever support bigint or bigfloat, this is the place for bigfloat 1262 len -= *start == '-' ? 1 : 0;
1263
1264 // does not fit into IV or UV, try NV
1265 if (len <= NV_DIG)
1266 // fits into NV without loss of precision
1267 return newSVnv (json_atof (start));
1268
1269 // everything else fails, convert it to a string
1270 return newSVpvn (start, dec->cur - start);
1271 }
1272
1273 // loss of precision here
965 return newSVnv (Atof (start)); 1274 return newSVnv (json_atof (start));
966 1275
967fail: 1276fail:
968 return 0; 1277 return 0;
969} 1278}
970 1279
… …
994 if (*dec->cur == ']') 1303 if (*dec->cur == ']')
995 { 1304 {
996 ++dec->cur; 1305 ++dec->cur;
997 break; 1306 break;
998 } 1307 }
999 1308
1000 if (*dec->cur != ',') 1309 if (*dec->cur != ',')
1001 ERR (", or ] expected while parsing array"); 1310 ERR (", or ] expected while parsing array");
1002 1311
1003 ++dec->cur; 1312 ++dec->cur;
1313
1314 decode_ws (dec);
1315
1316 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1317 {
1318 ++dec->cur;
1319 break;
1320 }
1004 } 1321 }
1005 1322
1006 DEC_DEC_DEPTH; 1323 DEC_DEC_DEPTH;
1007 return newRV_noinc ((SV *)av); 1324 return newRV_noinc ((SV *)av);
1008 1325
… …
1013} 1330}
1014 1331
1015static SV * 1332static SV *
1016decode_hv (dec_t *dec) 1333decode_hv (dec_t *dec)
1017{ 1334{
1335 SV *sv;
1018 HV *hv = newHV (); 1336 HV *hv = newHV ();
1019 1337
1020 DEC_INC_DEPTH; 1338 DEC_INC_DEPTH;
1021 decode_ws (dec); 1339 decode_ws (dec);
1022 1340
1023 if (*dec->cur == '}') 1341 if (*dec->cur == '}')
1024 ++dec->cur; 1342 ++dec->cur;
1025 else 1343 else
1026 for (;;) 1344 for (;;)
1027 { 1345 {
1346 EXPECT_CH ('"');
1347
1348 // heuristic: assume that
1349 // a) decode_str + hv_store_ent are abysmally slow.
1350 // b) most hash keys are short, simple ascii text.
1351 // => try to "fast-match" such strings to avoid
1352 // the overhead of decode_str + hv_store_ent.
1353 {
1028 SV *key, *value; 1354 SV *value;
1355 char *p = dec->cur;
1356 char *e = p + 24; // only try up to 24 bytes
1029 1357
1030 decode_ws (dec); EXPECT_CH ('"'); 1358 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 { 1359 {
1360 // the >= 0x80 is false on most architectures
1361 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1362 {
1363 // slow path, back up and use decode_str
1364 SV *key = decode_str (dec);
1365 if (!key)
1366 goto fail;
1367
1368 decode_ws (dec); EXPECT_CH (':');
1369
1370 decode_ws (dec);
1371 value = decode_sv (dec);
1372 if (!value)
1373 {
1374 SvREFCNT_dec (key);
1375 goto fail;
1376 }
1377
1378 hv_store_ent (hv, key, value, 0);
1041 SvREFCNT_dec (key); 1379 SvREFCNT_dec (key);
1380
1381 break;
1382 }
1383 else if (*p == '"')
1384 {
1385 // fast path, got a simple key
1386 char *key = dec->cur;
1387 int len = p - key;
1388 dec->cur = p + 1;
1389
1390 decode_ws (dec); EXPECT_CH (':');
1391
1392 decode_ws (dec);
1393 value = decode_sv (dec);
1394 if (!value)
1042 goto fail; 1395 goto fail;
1396
1397 hv_store (hv, key, len, value, 0);
1398
1399 break;
1400 }
1401
1402 ++p;
1043 } 1403 }
1044 1404 }
1045 hv_store_ent (hv, key, value, 0);
1046 SvREFCNT_dec (key);
1047 1405
1048 decode_ws (dec); 1406 decode_ws (dec);
1049 1407
1050 if (*dec->cur == '}') 1408 if (*dec->cur == '}')
1051 { 1409 {
… …
1055 1413
1056 if (*dec->cur != ',') 1414 if (*dec->cur != ',')
1057 ERR (", or } expected while parsing object/hash"); 1415 ERR (", or } expected while parsing object/hash");
1058 1416
1059 ++dec->cur; 1417 ++dec->cur;
1418
1419 decode_ws (dec);
1420
1421 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1422 {
1423 ++dec->cur;
1424 break;
1425 }
1060 } 1426 }
1061 1427
1062 DEC_DEC_DEPTH; 1428 DEC_DEC_DEPTH;
1063 return newRV_noinc ((SV *)hv); 1429 sv = newRV_noinc ((SV *)hv);
1430
1431 // check filter callbacks
1432 if (dec->json.flags & F_HOOK)
1433 {
1434 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1435 {
1436 HE *cb, *he;
1437
1438 hv_iterinit (hv);
1439 he = hv_iternext (hv);
1440 hv_iterinit (hv);
1441
1442 // the next line creates a mortal sv each time its called.
1443 // might want to optimise this for common cases.
1444 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1445
1446 if (cb)
1447 {
1448 dSP;
1449 int count;
1450
1451 ENTER; SAVETMPS; PUSHMARK (SP);
1452 XPUSHs (HeVAL (he));
1453 sv_2mortal (sv);
1454
1455 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1456
1457 if (count == 1)
1458 {
1459 sv = newSVsv (POPs);
1460 FREETMPS; LEAVE;
1461 return sv;
1462 }
1463
1464 SvREFCNT_inc (sv);
1465 FREETMPS; LEAVE;
1466 }
1467 }
1468
1469 if (dec->json.cb_object)
1470 {
1471 dSP;
1472 int count;
1473
1474 ENTER; SAVETMPS; PUSHMARK (SP);
1475 XPUSHs (sv_2mortal (sv));
1476
1477 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1478
1479 if (count == 1)
1480 {
1481 sv = newSVsv (POPs);
1482 FREETMPS; LEAVE;
1483 return sv;
1484 }
1485
1486 SvREFCNT_inc (sv);
1487 FREETMPS; LEAVE;
1488 }
1489 }
1490
1491 return sv;
1064 1492
1065fail: 1493fail:
1066 SvREFCNT_dec (hv); 1494 SvREFCNT_dec (hv);
1067 DEC_DEC_DEPTH; 1495 DEC_DEC_DEPTH;
1068 return 0; 1496 return 0;
1069} 1497}
1070 1498
1071static SV * 1499static SV *
1500decode_tag (dec_t *dec)
1501{
1502 SV *tag = 0;
1503 SV *val = 0;
1504
1505 if (!(dec->json.flags & F_ALLOW_TAGS))
1506 ERR ("malformed JSON string, neither array, object, number, string or atom");
1507
1508 ++dec->cur;
1509
1510 decode_ws (dec);
1511
1512 tag = decode_sv (dec);
1513 if (!tag)
1514 goto fail;
1515
1516 if (!SvPOK (tag))
1517 ERR ("malformed JSON string, (tag) must be a string");
1518
1519 decode_ws (dec);
1520
1521 if (*dec->cur != ')')
1522 ERR (") expected after tag");
1523
1524 ++dec->cur;
1525
1526 decode_ws (dec);
1527
1528 val = decode_sv (dec);
1529 if (!val)
1530 goto fail;
1531
1532 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1533 ERR ("malformed JSON string, tag value must be an array");
1534
1535 {
1536 AV *av = (AV *)SvRV (val);
1537 int i, len = av_len (av) + 1;
1538 HV *stash = gv_stashsv (tag, 0);
1539 SV *sv;
1540
1541 if (!stash)
1542 ERR ("cannot decode perl-object (package does not exist)");
1543
1544 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1545
1546 if (!method)
1547 ERR ("cannot decode perl-object (package does not have a THAW method)");
1548
1549 dSP;
1550
1551 ENTER; SAVETMPS; PUSHMARK (SP);
1552 EXTEND (SP, len + 2);
1553 // we re-bless the reference to get overload and other niceties right
1554 PUSHs (tag);
1555 PUSHs (sv_json);
1556
1557 for (i = 0; i < len; ++i)
1558 PUSHs (*av_fetch (av, i, 1));
1559
1560 PUTBACK;
1561 call_sv ((SV *)GvCV (method), G_SCALAR);
1562 SPAGAIN;
1563
1564 SvREFCNT_dec (tag);
1565 SvREFCNT_dec (val);
1566 sv = SvREFCNT_inc (POPs);
1567
1568 PUTBACK;
1569
1570 FREETMPS; LEAVE;
1571
1572 return sv;
1573 }
1574
1575fail:
1576 SvREFCNT_dec (tag);
1577 SvREFCNT_dec (val);
1578 return 0;
1579}
1580
1581static SV *
1072decode_sv (dec_t *dec) 1582decode_sv (dec_t *dec)
1073{ 1583{
1074 decode_ws (dec);
1075
1076 // the beauty of JSON: you need exactly one character lookahead 1584 // the beauty of JSON: you need exactly one character lookahead
1077 // to parse anything. 1585 // to parse everything.
1078 switch (*dec->cur) 1586 switch (*dec->cur)
1079 { 1587 {
1080 case '"': ++dec->cur; return decode_str (dec); 1588 case '"': ++dec->cur; return decode_str (dec);
1081 case '[': ++dec->cur; return decode_av (dec); 1589 case '[': ++dec->cur; return decode_av (dec);
1082 case '{': ++dec->cur; return decode_hv (dec); 1590 case '{': ++dec->cur; return decode_hv (dec);
1591 case '(': return decode_tag (dec);
1083 1592
1084 case '-': 1593 case '-':
1085 case '0': case '1': case '2': case '3': case '4': 1594 case '0': case '1': case '2': case '3': case '4':
1086 case '5': case '6': case '7': case '8': case '9': 1595 case '5': case '6': case '7': case '8': case '9':
1087 return decode_num (dec); 1596 return decode_num (dec);
1088 1597
1089 case 't': 1598 case 't':
1090 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1599 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1091 { 1600 {
1092 dec->cur += 4; 1601 dec->cur += 4;
1093 return SvREFCNT_inc (json_true); 1602#if JSON_SLOW
1603 types_true = get_bool ("Types::Serialiser::true");
1604#endif
1605 return newSVsv (types_true);
1094 } 1606 }
1095 else 1607 else
1096 ERR ("'true' expected"); 1608 ERR ("'true' expected");
1097 1609
1098 break; 1610 break;
1099 1611
1100 case 'f': 1612 case 'f':
1101 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1613 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1102 { 1614 {
1103 dec->cur += 5; 1615 dec->cur += 5;
1104 return SvREFCNT_inc (json_false); 1616#if JSON_SLOW
1617 types_false = get_bool ("Types::Serialiser::false");
1618#endif
1619 return newSVsv (types_false);
1105 } 1620 }
1106 else 1621 else
1107 ERR ("'false' expected"); 1622 ERR ("'false' expected");
1108 1623
1109 break; 1624 break;
… …
1118 ERR ("'null' expected"); 1633 ERR ("'null' expected");
1119 1634
1120 break; 1635 break;
1121 1636
1122 default: 1637 default:
1123 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1638 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1124 break; 1639 break;
1125 } 1640 }
1126 1641
1127fail: 1642fail:
1128 return 0; 1643 return 0;
1129} 1644}
1130 1645
1131static SV * 1646static SV *
1132decode_json (SV *string, U32 flags, UV *offset_return) 1647decode_json (SV *string, JSON *json, char **offset_return)
1133{ 1648{
1134 dec_t dec; 1649 dec_t dec;
1135 UV offset;
1136 SV *sv; 1650 SV *sv;
1137 1651
1652 /* work around bugs in 5.10 where manipulating magic values
1653 * makes perl ignore the magic in subsequent accesses.
1654 * also make a copy of non-PV values, to get them into a clean
1655 * state (SvPV should do that, but it's buggy, see below).
1656 */
1138 SvGETMAGIC (string); 1657 /*SvGETMAGIC (string);*/
1658 if (SvMAGICAL (string) || !SvPOK (string))
1659 string = sv_2mortal (newSVsv (string));
1660
1139 SvUPGRADE (string, SVt_PV); 1661 SvUPGRADE (string, SVt_PV);
1140 1662
1141 if (flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (flags)) 1663 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1664 * assertion with -DDEBUGGING, although SvCUR is documented to
1665 * return the xpv_cur field which certainly exists after upgrading.
1666 * according to nicholas clark, calling SvPOK fixes this.
1667 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1668 * and hope for the best.
1669 * Damnit, SvPV_nolen still trips over yet another assertion. This
1670 * assertion business is seriously broken, try yet another workaround
1671 * for the broken -DDEBUGGING.
1672 */
1673 {
1674#ifdef DEBUGGING
1675 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1676#else
1677 STRLEN offset = SvCUR (string);
1678#endif
1679
1680 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", 1681 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)); 1682 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1683 }
1144 1684
1145 if (flags & F_UTF8) 1685 if (DECODE_WANTS_OCTETS (json))
1146 sv_utf8_downgrade (string, 0); 1686 sv_utf8_downgrade (string, 0);
1147 else 1687 else
1148 sv_utf8_upgrade (string); 1688 sv_utf8_upgrade (string);
1149 1689
1150 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1690 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1151 1691
1152 dec.flags = flags; 1692 dec.json = *json;
1153 dec.cur = SvPVX (string); 1693 dec.cur = SvPVX (string);
1154 dec.end = SvEND (string); 1694 dec.end = SvEND (string);
1155 dec.err = 0; 1695 dec.err = 0;
1156 dec.depth = 0; 1696 dec.depth = 0;
1157 dec.maxdepth = DEC_DEPTH (dec.flags); 1697
1698 if (dec.json.cb_object || dec.json.cb_sk_object)
1699 dec.json.flags |= F_HOOK;
1158 1700
1159 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1701 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1702
1703 decode_ws (&dec);
1160 sv = decode_sv (&dec); 1704 sv = decode_sv (&dec);
1705
1706 if (offset_return)
1707 *offset_return = dec.cur;
1161 1708
1162 if (!(offset_return || !sv)) 1709 if (!(offset_return || !sv))
1163 { 1710 {
1164 // check for trailing garbage 1711 // check for trailing garbage
1165 decode_ws (&dec); 1712 decode_ws (&dec);
… …
1168 { 1715 {
1169 dec.err = "garbage after JSON object"; 1716 dec.err = "garbage after JSON object";
1170 SvREFCNT_dec (sv); 1717 SvREFCNT_dec (sv);
1171 sv = 0; 1718 sv = 0;
1172 } 1719 }
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 } 1720 }
1184 1721
1185 if (!sv) 1722 if (!sv)
1186 { 1723 {
1187 SV *uni = sv_newmortal (); 1724 SV *uni = sv_newmortal ();
… …
1193 SAVEVPTR (PL_curcop); 1730 SAVEVPTR (PL_curcop);
1194 PL_curcop = &cop; 1731 PL_curcop = &cop;
1195 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1732 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1196 LEAVE; 1733 LEAVE;
1197 1734
1198 croak ("%s, at character offset %d [\"%s\"]", 1735 croak ("%s, at character offset %d (before \"%s\")",
1199 dec.err, 1736 dec.err,
1200 (int)offset, 1737 (int)ptr_to_index (string, dec.cur),
1201 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1738 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1202 } 1739 }
1203 1740
1204 sv = sv_2mortal (sv); 1741 sv = sv_2mortal (sv);
1205 1742
1206 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1743 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (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)"); 1744 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1208 1745
1209 return sv; 1746 return sv;
1747}
1748
1749/////////////////////////////////////////////////////////////////////////////
1750// incremental parser
1751
1752static void
1753incr_parse (JSON *self)
1754{
1755 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1756
1757 // the state machine here is a bit convoluted and could be simplified a lot
1758 // but this would make it slower, so...
1759
1760 for (;;)
1761 {
1762 //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
1763 switch (self->incr_mode)
1764 {
1765 // only used for initial whitespace skipping
1766 case INCR_M_WS:
1767 for (;;)
1768 {
1769 if (*p > 0x20)
1770 {
1771 if (*p == '#')
1772 {
1773 self->incr_mode = INCR_M_C0;
1774 goto incr_m_c;
1775 }
1776 else
1777 {
1778 self->incr_mode = INCR_M_JSON;
1779 goto incr_m_json;
1780 }
1781 }
1782 else if (!*p)
1783 goto interrupt;
1784
1785 ++p;
1786 }
1787
1788 // skip a single char inside a string (for \\-processing)
1789 case INCR_M_BS:
1790 if (!*p)
1791 goto interrupt;
1792
1793 ++p;
1794 self->incr_mode = INCR_M_STR;
1795 goto incr_m_str;
1796
1797 // inside #-style comments
1798 case INCR_M_C0:
1799 case INCR_M_C1:
1800 incr_m_c:
1801 for (;;)
1802 {
1803 if (*p == '\n')
1804 {
1805 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1806 break;
1807 }
1808 else if (!*p)
1809 goto interrupt;
1810
1811 ++p;
1812 }
1813
1814 break;
1815
1816 // inside a string
1817 case INCR_M_STR:
1818 incr_m_str:
1819 for (;;)
1820 {
1821 if (*p == '"')
1822 {
1823 ++p;
1824 self->incr_mode = INCR_M_JSON;
1825
1826 if (!self->incr_nest)
1827 goto interrupt;
1828
1829 goto incr_m_json;
1830 }
1831 else if (*p == '\\')
1832 {
1833 ++p; // "virtually" consumes character after \
1834
1835 if (!*p) // if at end of string we have to switch modes
1836 {
1837 self->incr_mode = INCR_M_BS;
1838 goto interrupt;
1839 }
1840 }
1841 else if (!*p)
1842 goto interrupt;
1843
1844 ++p;
1845 }
1846
1847 // after initial ws, outside string
1848 case INCR_M_JSON:
1849 incr_m_json:
1850 for (;;)
1851 {
1852 switch (*p++)
1853 {
1854 case 0:
1855 --p;
1856 goto interrupt;
1857
1858 case 0x09:
1859 case 0x0a:
1860 case 0x0d:
1861 case 0x20:
1862 if (!self->incr_nest)
1863 {
1864 --p; // do not eat the whitespace, let the next round do it
1865 goto interrupt;
1866 }
1867 break;
1868
1869 case '"':
1870 self->incr_mode = INCR_M_STR;
1871 goto incr_m_str;
1872
1873 case '[':
1874 case '{':
1875 case '(':
1876 if (++self->incr_nest > self->max_depth)
1877 croak (ERR_NESTING_EXCEEDED);
1878 break;
1879
1880 case ']':
1881 case '}':
1882 if (--self->incr_nest <= 0)
1883 goto interrupt;
1884 break;
1885
1886 case ')':
1887 --self->incr_nest;
1888 break;
1889
1890 case '#':
1891 self->incr_mode = INCR_M_C1;
1892 goto incr_m_c;
1893 }
1894 }
1895 }
1896
1897 modechange:
1898 ;
1899 }
1900
1901interrupt:
1902 self->incr_pos = p - SvPVX (self->incr_text);
1903 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1904 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1210} 1905}
1211 1906
1212///////////////////////////////////////////////////////////////////////////// 1907/////////////////////////////////////////////////////////////////////////////
1213// XS interface functions 1908// XS interface functions
1214 1909
… …
1223 i >= '0' && i <= '9' ? i - '0' 1918 i >= '0' && i <= '9' ? i - '0'
1224 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1919 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1225 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1920 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1226 : -1; 1921 : -1;
1227 1922
1228 json_stash = gv_stashpv ("JSON::XS" , 1); 1923 json_stash = gv_stashpv ("JSON::XS" , 1);
1229 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1924 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1230 1925
1231 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1926 types_true = get_bool ("Types::Serialiser::true");
1232 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1927 types_false = get_bool ("Types::Serialiser::false");
1928
1929 sv_json = newSVpv ("JSON", 0);
1930 SvREADONLY_on (sv_json);
1931
1932 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1233} 1933}
1234 1934
1235PROTOTYPES: DISABLE 1935PROTOTYPES: DISABLE
1236 1936
1237SV *new (char *dummy) 1937void CLONE (...)
1238 CODE: 1938 CODE:
1239 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1939 json_stash = 0;
1240 OUTPUT: 1940 types_boolean_stash = 0;
1241 RETVAL
1242 1941
1942void new (char *klass)
1943 PPCODE:
1944{
1945 SV *pv = NEWSV (0, sizeof (JSON));
1946 SvPOK_only (pv);
1947 json_init ((JSON *)SvPVX (pv));
1948 XPUSHs (sv_2mortal (sv_bless (
1949 newRV_noinc (pv),
1950 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1951 )));
1952}
1953
1243SV *ascii (SV *self, int enable = 1) 1954void ascii (JSON *self, int enable = 1)
1244 ALIAS: 1955 ALIAS:
1245 ascii = F_ASCII 1956 ascii = F_ASCII
1246 latin1 = F_LATIN1 1957 latin1 = F_LATIN1
1247 utf8 = F_UTF8 1958 utf8 = F_UTF8
1248 indent = F_INDENT 1959 indent = F_INDENT
… …
1252 pretty = F_PRETTY 1963 pretty = F_PRETTY
1253 allow_nonref = F_ALLOW_NONREF 1964 allow_nonref = F_ALLOW_NONREF
1254 shrink = F_SHRINK 1965 shrink = F_SHRINK
1255 allow_blessed = F_ALLOW_BLESSED 1966 allow_blessed = F_ALLOW_BLESSED
1256 convert_blessed = F_CONV_BLESSED 1967 convert_blessed = F_CONV_BLESSED
1968 relaxed = F_RELAXED
1969 allow_unknown = F_ALLOW_UNKNOWN
1970 allow_tags = F_ALLOW_TAGS
1971 PPCODE:
1972{
1973 if (enable)
1974 self->flags |= ix;
1975 else
1976 self->flags &= ~ix;
1977
1978 XPUSHs (ST (0));
1979}
1980
1981void get_ascii (JSON *self)
1982 ALIAS:
1983 get_ascii = F_ASCII
1984 get_latin1 = F_LATIN1
1985 get_utf8 = F_UTF8
1986 get_indent = F_INDENT
1987 get_canonical = F_CANONICAL
1988 get_space_before = F_SPACE_BEFORE
1989 get_space_after = F_SPACE_AFTER
1990 get_allow_nonref = F_ALLOW_NONREF
1991 get_shrink = F_SHRINK
1992 get_allow_blessed = F_ALLOW_BLESSED
1993 get_convert_blessed = F_CONV_BLESSED
1994 get_relaxed = F_RELAXED
1995 get_allow_unknown = F_ALLOW_UNKNOWN
1996 get_allow_tags = F_ALLOW_TAGS
1997 PPCODE:
1998 XPUSHs (boolSV (self->flags & ix));
1999
2000void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
2001 PPCODE:
2002 self->max_depth = max_depth;
2003 XPUSHs (ST (0));
2004
2005U32 get_max_depth (JSON *self)
1257 CODE: 2006 CODE:
1258{ 2007 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: 2008 OUTPUT:
1268 RETVAL 2009 RETVAL
1269 2010
1270SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 2011void max_size (JSON *self, U32 max_size = 0)
2012 PPCODE:
2013 self->max_size = max_size;
2014 XPUSHs (ST (0));
2015
2016int get_max_size (JSON *self)
1271 CODE: 2017 CODE:
1272{ 2018 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: 2019 OUTPUT:
1286 RETVAL 2020 RETVAL
1287 2021
1288SV *max_size (SV *self, UV max_size = 0) 2022void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
2023 PPCODE:
2024{
2025 SvREFCNT_dec (self->cb_object);
2026 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
2027
2028 XPUSHs (ST (0));
2029}
2030
2031void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
2032 PPCODE:
2033{
2034 if (!self->cb_sk_object)
2035 self->cb_sk_object = newHV ();
2036
2037 if (SvOK (cb))
2038 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
2039 else
2040 {
2041 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
2042
2043 if (!HvKEYS (self->cb_sk_object))
2044 {
2045 SvREFCNT_dec (self->cb_sk_object);
2046 self->cb_sk_object = 0;
2047 }
2048 }
2049
2050 XPUSHs (ST (0));
2051}
2052
2053void encode (JSON *self, SV *scalar)
2054 PPCODE:
2055 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2056 XPUSHs (scalar);
2057
2058void decode (JSON *self, SV *jsonstr)
2059 PPCODE:
2060 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2061 XPUSHs (jsonstr);
2062
2063void decode_prefix (JSON *self, SV *jsonstr)
2064 PPCODE:
2065{
2066 SV *sv;
2067 char *offset;
2068 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
2069 EXTEND (SP, 2);
2070 PUSHs (sv);
2071 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
2072}
2073
2074void incr_parse (JSON *self, SV *jsonstr = 0)
2075 PPCODE:
2076{
2077 if (!self->incr_text)
2078 self->incr_text = newSVpvn ("", 0);
2079
2080 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2081 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2082 if (DECODE_WANTS_OCTETS (self))
2083 {
2084 if (self->incr_pos)
2085 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2086 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2087
2088 sv_utf8_downgrade (self->incr_text, 0);
2089 }
2090 else
2091 {
2092 sv_utf8_upgrade (self->incr_text);
2093
2094 if (self->incr_pos)
2095 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2096 - (U8 *)SvPVX (self->incr_text);
2097 }
2098
2099 // append data, if any
2100 if (jsonstr)
2101 {
2102 /* make sure both strings have same encoding */
2103 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2104 if (SvUTF8 (jsonstr))
2105 sv_utf8_downgrade (jsonstr, 0);
2106 else
2107 sv_utf8_upgrade (jsonstr);
2108
2109 /* and then just blindly append */
2110 {
2111 STRLEN len;
2112 const char *str = SvPV (jsonstr, len);
2113 STRLEN cur = SvCUR (self->incr_text);
2114
2115 if (SvLEN (self->incr_text) <= cur + len)
2116 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
2117
2118 Move (str, SvEND (self->incr_text), len, char);
2119 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2120 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2121 }
2122 }
2123
2124 if (GIMME_V != G_VOID)
2125 do
2126 {
2127 SV *sv;
2128 char *offset;
2129
2130 if (!INCR_DONE (self))
2131 {
2132 incr_parse (self);
2133
2134 if (self->incr_pos > self->max_size && self->max_size)
2135 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2136 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2137
2138 if (!INCR_DONE (self))
2139 {
2140 // as an optimisation, do not accumulate white space in the incr buffer
2141 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2142 {
2143 self->incr_pos = 0;
2144 SvCUR_set (self->incr_text, 0);
2145 }
2146
2147 break;
2148 }
2149 }
2150
2151 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2152 XPUSHs (sv);
2153
2154 self->incr_pos -= offset - SvPVX (self->incr_text);
2155 self->incr_nest = 0;
2156 self->incr_mode = 0;
2157
2158 sv_chop (self->incr_text, offset);
2159 }
2160 while (GIMME_V == G_ARRAY);
2161}
2162
2163SV *incr_text (JSON *self)
2164 ATTRS: lvalue
1289 CODE: 2165 CODE:
1290{ 2166{
1291 UV *uv = SvJSON (self); 2167 if (self->incr_pos)
1292 UV log2 = 0; 2168 croak ("incr_text can not be called when the incremental parser already started parsing");
1293 2169
1294 if (max_size > 0x80000000UL) max_size = 0x80000000UL; 2170 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} 2171}
1304 OUTPUT: 2172 OUTPUT:
1305 RETVAL 2173 RETVAL
1306 2174
2175void incr_skip (JSON *self)
2176 CODE:
2177{
2178 if (self->incr_pos)
2179 {
2180 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2181 self->incr_pos = 0;
2182 self->incr_nest = 0;
2183 self->incr_mode = 0;
2184 }
2185}
2186
2187void incr_reset (JSON *self)
2188 CODE:
2189{
2190 SvREFCNT_dec (self->incr_text);
2191 self->incr_text = 0;
2192 self->incr_pos = 0;
2193 self->incr_nest = 0;
2194 self->incr_mode = 0;
2195}
2196
2197void DESTROY (JSON *self)
2198 CODE:
2199 SvREFCNT_dec (self->cb_sk_object);
2200 SvREFCNT_dec (self->cb_object);
2201 SvREFCNT_dec (self->incr_text);
2202
2203PROTOTYPES: ENABLE
2204
1307void encode (SV *self, SV *scalar) 2205void encode_json (SV *scalar)
1308 PPCODE: 2206 PPCODE:
1309 XPUSHs (encode_json (scalar, *SvJSON (self))); 2207{
2208 JSON json;
2209 json_init (&json);
2210 json.flags |= F_UTF8;
2211 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2212 XPUSHs (scalar);
2213}
1310 2214
1311void decode (SV *self, SV *jsonstr) 2215void decode_json (SV *jsonstr)
1312 PPCODE: 2216 PPCODE:
1313 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1314
1315void decode_prefix (SV *self, SV *jsonstr)
1316 PPCODE:
1317{ 2217{
1318 UV offset; 2218 JSON json;
1319 EXTEND (SP, 2); 2219 json_init (&json);
1320 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 2220 json.flags |= F_UTF8;
1321 PUSHs (sv_2mortal (newSVuv (offset))); 2221 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2222 XPUSHs (jsonstr);
1322} 2223}
1323 2224
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