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 1.42 by root, Sat Jun 23 22:56:08 2007 UTC vs.
Revision 1.125 by root, Sun Feb 21 15:37:53 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
19 21
22// compatibility with perl <5.18
23#ifndef HvNAMELEN_get
24# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
25#endif
26#ifndef HvNAMELEN
27# define HvNAMELEN(hv) HvNAMELEN_get (hv)
28#endif
29#ifndef HvNAMEUTF8
30# define HvNAMEUTF8(hv) 0
31#endif
32
33// three extra for rounding, sign, and end of string
34#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
35
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
24#define F_CANONICAL 0x00000010UL 40#define F_CANONICAL 0x00000010UL
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
45#define F_ALLOW_BLESSED 0x00000400UL
46#define F_CONV_BLESSED 0x00000800UL
29#define F_MAXDEPTH 0xf8000000UL 47#define F_RELAXED 0x00001000UL
30#define S_MAXDEPTH 27 48#define F_ALLOW_UNKNOWN 0x00002000UL
31 49#define F_ALLOW_TAGS 0x00004000UL
32#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 50#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
33
34// F_SELFCONVERT? <=> to_json/toJson
35// F_BLESSED? <=> { $__class__$ => }
36 51
37#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 52#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
38#define F_DEFAULT (9UL << S_MAXDEPTH)
39 53
40#define INIT_SIZE 32 // initial scalar size to be allocated 54#define INIT_SIZE 32 // initial scalar size to be allocated
41#define INDENT_STEP 3 // spaces per indentation level 55#define INDENT_STEP 3 // spaces per indentation level
42 56
43#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
44 58
59#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
60
45#define SB do { 61#define SB do {
46#define SE } while (0) 62#define SE } while (0)
47 63
48#if __GNUC__ >= 3 64#if __GNUC__ >= 3
49# define expect(expr,value) __builtin_expect ((expr),(value)) 65# define expect(expr,value) __builtin_expect ((expr), (value))
50# define inline inline 66# define INLINE static inline
51#else 67#else
52# define expect(expr,value) (expr) 68# define expect(expr,value) (expr)
53# define inline static 69# define INLINE static
54#endif 70#endif
55 71
56#define expect_false(expr) expect ((expr) != 0, 0) 72#define expect_false(expr) expect ((expr) != 0, 0)
57#define expect_true(expr) expect ((expr) != 0, 1) 73#define expect_true(expr) expect ((expr) != 0, 1)
58 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
59static HV *json_stash; // JSON::XS:: 92static HV *json_stash, *types_boolean_stash; // JSON::XS::
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}
60 127
61///////////////////////////////////////////////////////////////////////////// 128/////////////////////////////////////////////////////////////////////////////
62// utility functions 129// utility functions
63 130
64static UV * 131INLINE SV *
65SvJSON (SV *sv) 132get_bool (const char *name)
66{ 133{
67 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 134 SV *sv = get_sv (name, 1);
68 croak ("object is not of type JSON::XS");
69 135
70 return &SvUVX (SvRV (sv)); 136 SvREADONLY_on (sv);
71} 137 SvREADONLY_on (SvRV (sv));
72 138
73static void 139 return sv;
140}
141
142INLINE void
74shrink (SV *sv) 143shrink (SV *sv)
75{ 144{
76 sv_utf8_downgrade (sv, 1); 145 sv_utf8_downgrade (sv, 1);
146
77 if (SvLEN (sv) > SvCUR (sv) + 1) 147 if (SvLEN (sv) > SvCUR (sv) + 1)
78 { 148 {
79#ifdef SvPV_shrink_to_cur 149#ifdef SvPV_shrink_to_cur
80 SvPV_shrink_to_cur (sv); 150 SvPV_shrink_to_cur (sv);
81#elif defined (SvPV_renew) 151#elif defined (SvPV_renew)
… …
87// 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
88// case of an error. 158// case of an error.
89// we special-case "safe" characters from U+80 .. U+7FF, 159// we special-case "safe" characters from U+80 .. U+7FF,
90// but use the very good perl function to parse anything else. 160// but use the very good perl function to parse anything else.
91// note that we never call this function for a ascii codepoints 161// note that we never call this function for a ascii codepoints
92inline UV 162INLINE UV
93decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 163decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
94{ 164{
95 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 165 if (expect_true (len >= 2
96 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 166 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
97 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 167 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
98 { 168 {
99 *clen = 2; 169 *clen = 2;
100 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 170 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
101 } 171 }
102 else 172 else
103 { 173 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
104 *clen = (STRLEN)-1; 174}
105 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 (;;)
106 } 226 {
107} 227 U8 dig = (U8)*s - '0';
108 228
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}
109///////////////////////////////////////////////////////////////////////////// 304/////////////////////////////////////////////////////////////////////////////
110// encoder 305// encoder
111 306
112// structure used for encoding JSON 307// structure used for encoding JSON
113typedef struct 308typedef struct
114{ 309{
115 char *cur; // SvPVX (sv) + current output position 310 char *cur; // SvPVX (sv) + current output position
116 char *end; // SvEND (sv) 311 char *end; // SvEND (sv)
117 SV *sv; // result scalar 312 SV *sv; // result scalar
118 U32 flags; // F_* 313 JSON json;
119 U32 indent; // indentation level 314 U32 indent; // indentation level
120 U32 maxdepth; // max. indentation/recursion level 315 UV limit; // escape character values >= this value when encoding
121} enc_t; 316} enc_t;
122 317
123inline void 318INLINE void
124need (enc_t *enc, STRLEN len) 319need (enc_t *enc, STRLEN len)
125{ 320{
126 if (expect_false (enc->cur + len >= enc->end)) 321 if (expect_false (enc->cur + len >= enc->end))
127 { 322 {
128 STRLEN cur = enc->cur - SvPVX (enc->sv); 323 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
129 SvGROW (enc->sv, cur + len + 1); 324 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
130 enc->cur = SvPVX (enc->sv) + cur; 325 enc->cur = SvPVX (enc->sv) + cur;
131 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 326 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
132 } 327 }
133} 328}
134 329
135inline void 330INLINE void
136encode_ch (enc_t *enc, char ch) 331encode_ch (enc_t *enc, char ch)
137{ 332{
138 need (enc, 1); 333 need (enc, 1);
139 *enc->cur++ = ch; 334 *enc->cur++ = ch;
140} 335}
… …
194 { 389 {
195 uch = ch; 390 uch = ch;
196 clen = 1; 391 clen = 1;
197 } 392 }
198 393
199 if (uch > 0x10FFFFUL) 394 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
200 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
201
202 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
203 { 395 {
204 if (uch > 0xFFFFUL) 396 if (uch >= 0x10000UL)
205 { 397 {
398 if (uch >= 0x110000UL)
399 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
400
206 need (enc, len += 11); 401 need (enc, len += 11);
207 sprintf (enc->cur, "\\u%04x\\u%04x", 402 sprintf (enc->cur, "\\u%04x\\u%04x",
208 (int)((uch - 0x10000) / 0x400 + 0xD800), 403 (int)((uch - 0x10000) / 0x400 + 0xD800),
209 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 404 (int)((uch - 0x10000) % 0x400 + 0xDC00));
210 enc->cur += 12; 405 enc->cur += 12;
211 } 406 }
212 else 407 else
213 { 408 {
214 static char hexdigit [16] = "0123456789abcdef";
215 need (enc, len += 5); 409 need (enc, len += 5);
216 *enc->cur++ = '\\'; 410 *enc->cur++ = '\\';
217 *enc->cur++ = 'u'; 411 *enc->cur++ = 'u';
218 *enc->cur++ = hexdigit [ uch >> 12 ]; 412 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
219 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 413 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
220 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 414 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
221 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 415 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
222 } 416 }
223 417
224 str += clen; 418 str += clen;
225 } 419 }
226 else if (enc->flags & F_LATIN1) 420 else if (enc->json.flags & F_LATIN1)
227 { 421 {
228 *enc->cur++ = uch; 422 *enc->cur++ = uch;
229 str += clen; 423 str += clen;
230 } 424 }
231 else if (is_utf8) 425 else if (is_utf8)
… …
238 while (--clen); 432 while (--clen);
239 } 433 }
240 else 434 else
241 { 435 {
242 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
243 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 437 enc->cur = encode_utf8 (enc->cur, uch);
244 ++str; 438 ++str;
245 } 439 }
246 } 440 }
247 } 441 }
248 } 442 }
249 443
250 --len; 444 --len;
251 } 445 }
252} 446}
253 447
254inline void 448INLINE void
255encode_indent (enc_t *enc) 449encode_indent (enc_t *enc)
256{ 450{
257 if (enc->flags & F_INDENT) 451 if (enc->json.flags & F_INDENT)
258 { 452 {
259 int spaces = enc->indent * INDENT_STEP; 453 int spaces = enc->indent * INDENT_STEP;
260 454
261 need (enc, spaces); 455 need (enc, spaces);
262 memset (enc->cur, ' ', spaces); 456 memset (enc->cur, ' ', spaces);
263 enc->cur += spaces; 457 enc->cur += spaces;
264 } 458 }
265} 459}
266 460
267inline void 461INLINE void
268encode_space (enc_t *enc) 462encode_space (enc_t *enc)
269{ 463{
270 need (enc, 1); 464 need (enc, 1);
271 encode_ch (enc, ' '); 465 encode_ch (enc, ' ');
272} 466}
273 467
274inline void 468INLINE void
275encode_nl (enc_t *enc) 469encode_nl (enc_t *enc)
276{ 470{
277 if (enc->flags & F_INDENT) 471 if (enc->json.flags & F_INDENT)
278 { 472 {
279 need (enc, 1); 473 need (enc, 1);
280 encode_ch (enc, '\n'); 474 encode_ch (enc, '\n');
281 } 475 }
282} 476}
283 477
284inline void 478INLINE void
285encode_comma (enc_t *enc) 479encode_comma (enc_t *enc)
286{ 480{
287 encode_ch (enc, ','); 481 encode_ch (enc, ',');
288 482
289 if (enc->flags & F_INDENT) 483 if (enc->json.flags & F_INDENT)
290 encode_nl (enc); 484 encode_nl (enc);
291 else if (enc->flags & F_SPACE_AFTER) 485 else if (enc->json.flags & F_SPACE_AFTER)
292 encode_space (enc); 486 encode_space (enc);
293} 487}
294 488
295static void encode_sv (enc_t *enc, SV *sv); 489static void encode_sv (enc_t *enc, SV *sv);
296 490
297static void 491static void
298encode_av (enc_t *enc, AV *av) 492encode_av (enc_t *enc, AV *av)
299{ 493{
300 int i, len = av_len (av); 494 int i, len = av_len (av);
301 495
302 if (enc->indent >= enc->maxdepth) 496 if (enc->indent >= enc->json.max_depth)
303 croak ("data structure too deep (hit recursion limit)"); 497 croak (ERR_NESTING_EXCEEDED);
304 498
305 encode_ch (enc, '['); encode_nl (enc); 499 encode_ch (enc, '[');
306 ++enc->indent;
307 500
501 if (len >= 0)
502 {
503 encode_nl (enc); ++enc->indent;
504
308 for (i = 0; i <= len; ++i) 505 for (i = 0; i <= len; ++i)
309 { 506 {
507 SV **svp = av_fetch (av, i, 0);
508
310 encode_indent (enc); 509 encode_indent (enc);
311 encode_sv (enc, *av_fetch (av, i, 0));
312 510
511 if (svp)
512 encode_sv (enc, *svp);
513 else
514 encode_str (enc, "null", 4, 0);
515
313 if (i < len) 516 if (i < len)
314 encode_comma (enc); 517 encode_comma (enc);
315 } 518 }
316 519
520 encode_nl (enc); --enc->indent; encode_indent (enc);
521 }
522
317 encode_nl (enc); 523 encode_ch (enc, ']');
318
319 --enc->indent;
320 encode_indent (enc); encode_ch (enc, ']');
321} 524}
322 525
323static void 526static void
324encode_he (enc_t *enc, HE *he) 527encode_hk (enc_t *enc, HE *he)
325{ 528{
326 encode_ch (enc, '"'); 529 encode_ch (enc, '"');
327 530
328 if (HeKLEN (he) == HEf_SVKEY) 531 if (HeKLEN (he) == HEf_SVKEY)
329 { 532 {
330 SV *sv = HeSVKEY (he); 533 SV *sv = HeSVKEY (he);
331 STRLEN len; 534 STRLEN len;
332 char *str; 535 char *str;
333 536
334 SvGETMAGIC (sv); 537 SvGETMAGIC (sv);
335 str = SvPV (sv, len); 538 str = SvPV (sv, len);
336 539
337 encode_str (enc, str, len, SvUTF8 (sv)); 540 encode_str (enc, str, len, SvUTF8 (sv));
338 } 541 }
339 else 542 else
340 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 543 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
341 544
342 encode_ch (enc, '"'); 545 encode_ch (enc, '"');
343 546
344 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 547 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
345 encode_ch (enc, ':'); 548 encode_ch (enc, ':');
346 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 549 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
347 encode_sv (enc, HeVAL (he));
348} 550}
349 551
350// compare hash entries, used when all keys are bytestrings 552// compare hash entries, used when all keys are bytestrings
351static int 553static int
352he_cmp_fast (const void *a_, const void *b_) 554he_cmp_fast (const void *a_, const void *b_)
… …
357 HE *b = *(HE **)b_; 559 HE *b = *(HE **)b_;
358 560
359 STRLEN la = HeKLEN (a); 561 STRLEN la = HeKLEN (a);
360 STRLEN lb = HeKLEN (b); 562 STRLEN lb = HeKLEN (b);
361 563
362 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 564 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
363 cmp = la - lb; 565 cmp = lb - la;
364 566
365 return cmp; 567 return cmp;
366} 568}
367 569
368// 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
369static int 571static int
370he_cmp_slow (const void *a, const void *b) 572he_cmp_slow (const void *a, const void *b)
371{ 573{
372 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 574 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
373} 575}
374 576
375static void 577static void
376encode_hv (enc_t *enc, HV *hv) 578encode_hv (enc_t *enc, HV *hv)
377{ 579{
378 int count, i; 580 HE *he;
379 581
380 if (enc->indent >= enc->maxdepth) 582 if (enc->indent >= enc->json.max_depth)
381 croak ("data structure too deep (hit recursion limit)"); 583 croak (ERR_NESTING_EXCEEDED);
382 584
383 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 585 encode_ch (enc, '{');
384 586
385 if ((count = hv_iterinit (hv)))
386 {
387 // for canonical output we have to sort by keys first 587 // for canonical output we have to sort by keys first
388 // actually, this is mostly due to the stupid so-called 588 // actually, this is mostly due to the stupid so-called
389 // security workaround added somewhere in 5.8.x. 589 // security workaround added somewhere in 5.8.x
390 // that randomises hash orderings 590 // that randomises hash orderings
391 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))
392 { 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 {
393 int fast = 1; 611 int i, fast = 1;
394 HE *he; 612 HE *hes_stack [STACK_HES];
395#if defined(__BORLANDC__) || defined(_MSC_VER) 613 HE **hes = hes_stack;
396 HE **hes = _alloca (count * sizeof (HE)); 614
397#else 615 // allocate larger arrays on the heap
398 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 616 if (count > STACK_HES)
399#endif 617 {
618 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
619 hes = (HE **)SvPVX (sv);
620 }
400 621
401 i = 0; 622 i = 0;
402 while ((he = hv_iternext (hv))) 623 while ((he = hv_iternext (hv)))
403 { 624 {
404 hes [i++] = he; 625 hes [i++] = he;
… …
426 647
427 FREETMPS; 648 FREETMPS;
428 LEAVE; 649 LEAVE;
429 } 650 }
430 651
431 for (i = 0; i < count; ++i) 652 encode_nl (enc); ++enc->indent;
653
654 while (count--)
432 { 655 {
433 encode_indent (enc); 656 encode_indent (enc);
657 he = hes [count];
434 encode_he (enc, hes [i]); 658 encode_hk (enc, he);
659 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
435 660
436 if (i < count - 1) 661 if (count)
437 encode_comma (enc); 662 encode_comma (enc);
438 } 663 }
439 664
440 encode_nl (enc); 665 encode_nl (enc); --enc->indent; encode_indent (enc);
441 } 666 }
667 }
442 else 668 else
669 {
670 if (hv_iterinit (hv) || SvMAGICAL (hv))
671 if ((he = hv_iternext (hv)))
443 { 672 {
444 HE *he = hv_iternext (hv); 673 encode_nl (enc); ++enc->indent;
445 674
446 for (;;) 675 for (;;)
447 { 676 {
448 encode_indent (enc); 677 encode_indent (enc);
449 encode_he (enc, he); 678 encode_hk (enc, he);
679 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
450 680
451 if (!(he = hv_iternext (hv))) 681 if (!(he = hv_iternext (hv)))
452 break; 682 break;
453 683
454 encode_comma (enc); 684 encode_comma (enc);
455 } 685 }
456 686
457 encode_nl (enc); 687 encode_nl (enc); --enc->indent; encode_indent (enc);
458 } 688 }
459 } 689 }
460 690
461 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 691 encode_ch (enc, '}');
462} 692}
463 693
464// encode objects, arrays and special \0=false and \1=true values. 694// encode objects, arrays and special \0=false and \1=true values.
465static void 695static void
466encode_rv (enc_t *enc, SV *sv) 696encode_rv (enc_t *enc, SV *sv)
467{ 697{
468 svtype svt; 698 svtype svt;
699 GV *method;
469 700
470 SvGETMAGIC (sv); 701 SvGETMAGIC (sv);
471 svt = SvTYPE (sv); 702 svt = SvTYPE (sv);
472 703
704 if (expect_false (SvOBJECT (sv)))
705 {
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)
712 {
713 if (SvIV (sv))
714 encode_str (enc, "true", 4, 0);
715 else
716 encode_str (enc, "false", 5, 0);
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;
724 SAVESTACK_POS ();
725 PUSHMARK (SP);
726 EXTEND (SP, 2);
727 // we re-bless the reference to get overload and other niceties right
728 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
729 PUSHs (sv_json);
730
731 PUTBACK;
732 count = call_sv ((SV *)GvCV (method), G_ARRAY);
733 SPAGAIN;
734
735 // catch this surprisingly common error
736 if (SvROK (TOPs) && SvRV (TOPs) == sv)
737 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
738
739 encode_ch (enc, '(');
740 encode_ch (enc, '"');
741 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
742 encode_ch (enc, '"');
743 encode_ch (enc, ')');
744 encode_ch (enc, '[');
745
746 while (count)
747 {
748 encode_sv (enc, SP[1 - count--]);
749
750 if (count)
751 encode_ch (enc, ',');
752 }
753
754 encode_ch (enc, ']');
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;
763 PUSHMARK (SP);
764 // we re-bless the reference to get overload and other niceties right
765 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
766
767 // calling with G_SCALAR ensures that we always get a 1 return value
768 PUTBACK;
769 call_sv ((SV *)GvCV (method), G_SCALAR);
770 SPAGAIN;
771
772 // catch this surprisingly common error
773 if (SvROK (TOPs) && SvRV (TOPs) == sv)
774 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
775
776 sv = POPs;
777 PUTBACK;
778
779 encode_sv (enc, sv);
780
781 FREETMPS; LEAVE;
782 }
783 else if (enc->json.flags & F_ALLOW_BLESSED)
784 encode_str (enc, "null", 4, 0);
785 else
786 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
787 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
788 }
473 if (svt == SVt_PVHV) 789 else if (svt == SVt_PVHV)
474 encode_hv (enc, (HV *)sv); 790 encode_hv (enc, (HV *)sv);
475 else if (svt == SVt_PVAV) 791 else if (svt == SVt_PVAV)
476 encode_av (enc, (AV *)sv); 792 encode_av (enc, (AV *)sv);
477 else if (svt < SVt_PVAV) 793 else if (svt < SVt_PVAV)
478 { 794 {
479 if (SvNIOK (sv) && SvIV (sv) == 0) 795 STRLEN len = 0;
796 char *pv = svt ? SvPV (sv, len) : 0;
797
798 if (len == 1 && *pv == '1')
799 encode_str (enc, "true", 4, 0);
800 else if (len == 1 && *pv == '0')
480 encode_str (enc, "false", 5, 0); 801 encode_str (enc, "false", 5, 0);
481 else if (SvNIOK (sv) && SvIV (sv) == 1) 802 else if (enc->json.flags & F_ALLOW_UNKNOWN)
482 encode_str (enc, "true", 4, 0); 803 encode_str (enc, "null", 4, 0);
483 else 804 else
484 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 805 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
485 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 806 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
486 } 807 }
808 else if (enc->json.flags & F_ALLOW_UNKNOWN)
809 encode_str (enc, "null", 4, 0);
487 else 810 else
488 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 811 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
489 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 812 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
490} 813}
491 814
… …
509 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 832 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
510 enc->cur += strlen (enc->cur); 833 enc->cur += strlen (enc->cur);
511 } 834 }
512 else if (SvIOKp (sv)) 835 else if (SvIOKp (sv))
513 { 836 {
514 // we assume we can always read an IV as a UV 837 // we assume we can always read an IV as a UV and vice versa
515 if (SvUV (sv) & ~(UV)0x7fff) 838 // we assume two's complement
516 { 839 // we assume no aliasing issues in the union
517 // large integer, use the (rather slow) snprintf way. 840 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
518 need (enc, sizeof (UV) * 3); 841 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
519 enc->cur +=
520 SvIsUV(sv)
521 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
522 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
523 }
524 else
525 { 842 {
526 // optimise the "small number case" 843 // optimise the "small number case"
527 // code will likely be branchless and use only a single multiplication 844 // code will likely be branchless and use only a single multiplication
845 // works for numbers up to 59074
528 I32 i = SvIV (sv); 846 I32 i = SvIVX (sv);
529 U32 u; 847 U32 u;
530 char digit, nz = 0; 848 char digit, nz = 0;
531 849
532 need (enc, 6); 850 need (enc, 6);
533 851
… …
539 857
540 // now output digit by digit, each time masking out the integer part 858 // now output digit by digit, each time masking out the integer part
541 // and multiplying by 5 while moving the decimal point one to the right, 859 // and multiplying by 5 while moving the decimal point one to the right,
542 // resulting in a net multiplication by 10. 860 // resulting in a net multiplication by 10.
543 // we always write the digit to memory but conditionally increment 861 // we always write the digit to memory but conditionally increment
544 // the pointer, to ease the usage of conditional move instructions. 862 // the pointer, to enable the use of conditional move instructions.
545 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 863 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
546 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 864 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
547 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 865 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
548 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 866 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
549 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 867 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
550 } 868 }
869 else
870 {
871 // large integer, use the (rather slow) snprintf way.
872 need (enc, IVUV_MAXCHARS);
873 enc->cur +=
874 SvIsUV(sv)
875 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
876 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
877 }
551 } 878 }
552 else if (SvROK (sv)) 879 else if (SvROK (sv))
553 encode_rv (enc, SvRV (sv)); 880 encode_rv (enc, SvRV (sv));
554 else if (!SvOK (sv)) 881 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
555 encode_str (enc, "null", 4, 0); 882 encode_str (enc, "null", 4, 0);
556 else 883 else
557 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 884 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
558 SvPV_nolen (sv), SvFLAGS (sv)); 885 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
886}
887
888static int
889json_scalar (SV *scalar)
890{
891 return 0;//D
892 if (!SvROK (scalar))
893 return 1;
559} 894}
560 895
561static SV * 896static SV *
562encode_json (SV *scalar, U32 flags) 897encode_json (SV *scalar, JSON *json)
563{ 898{
564 enc_t enc; 899 enc_t enc;
565 900
566 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 901 if (!(json->flags & F_ALLOW_NONREF) && json_scalar (scalar))
567 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 902 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
568 903
569 enc.flags = flags; 904 enc.json = *json;
570 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 905 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
571 enc.cur = SvPVX (enc.sv); 906 enc.cur = SvPVX (enc.sv);
572 enc.end = SvEND (enc.sv); 907 enc.end = SvEND (enc.sv);
573 enc.indent = 0; 908 enc.indent = 0;
574 enc.maxdepth = DEC_DEPTH (flags); 909 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
910 : enc.json.flags & F_LATIN1 ? 0x000100UL
911 : 0x110000UL;
575 912
576 SvPOK_only (enc.sv); 913 SvPOK_only (enc.sv);
577 encode_sv (&enc, scalar); 914 encode_sv (&enc, scalar);
915 encode_nl (&enc);
578 916
579 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 917 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
580 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 918 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
581 919
582 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 920 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
583 SvUTF8_on (enc.sv); 921 SvUTF8_on (enc.sv);
584 922
585 if (enc.flags & F_SHRINK) 923 if (enc.json.flags & F_SHRINK)
586 shrink (enc.sv); 924 shrink (enc.sv);
587 925
588 return enc.sv; 926 return enc.sv;
589} 927}
590 928
… …
595typedef struct 933typedef struct
596{ 934{
597 char *cur; // current parser pointer 935 char *cur; // current parser pointer
598 char *end; // end of input string 936 char *end; // end of input string
599 const char *err; // parse error, if != 0 937 const char *err; // parse error, if != 0
600 U32 flags; // F_* 938 JSON json;
601 U32 depth; // recursion depth 939 U32 depth; // recursion depth
602 U32 maxdepth; // recursion depth limit 940 U32 maxdepth; // recursion depth limit
603} dec_t; 941} dec_t;
604 942
605inline void 943INLINE void
944decode_comment (dec_t *dec)
945{
946 // only '#'-style comments allowed a.t.m.
947
948 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
949 ++dec->cur;
950}
951
952INLINE void
606decode_ws (dec_t *dec) 953decode_ws (dec_t *dec)
607{ 954{
608 for (;;) 955 for (;;)
609 { 956 {
610 char ch = *dec->cur; 957 char ch = *dec->cur;
611 958
612 if (ch > 0x20 959 if (ch > 0x20)
960 {
961 if (expect_false (ch == '#'))
962 {
963 if (dec->json.flags & F_RELAXED)
964 decode_comment (dec);
965 else
966 break;
967 }
968 else
969 break;
970 }
613 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 971 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
614 break; 972 break; // parse error, but let higher level handle it, gives better error messages
615 973
616 ++dec->cur; 974 ++dec->cur;
617 } 975 }
618} 976}
619 977
… …
623 if (*dec->cur != ch) \ 981 if (*dec->cur != ch) \
624 ERR (# ch " expected"); \ 982 ERR (# ch " expected"); \
625 ++dec->cur; \ 983 ++dec->cur; \
626 SE 984 SE
627 985
628#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 986#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
629#define DEC_DEC_DEPTH --dec->depth 987#define DEC_DEC_DEPTH --dec->depth
630 988
631static SV *decode_sv (dec_t *dec); 989static SV *decode_sv (dec_t *dec);
632 990
633static signed char decode_hexdigit[256]; 991static signed char decode_hexdigit[256];
… …
725 1083
726 if (hi >= 0x80) 1084 if (hi >= 0x80)
727 { 1085 {
728 utf8 = 1; 1086 utf8 = 1;
729 1087
730 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1088 cur = encode_utf8 (cur, hi);
731 } 1089 }
732 else 1090 else
733 *cur++ = hi; 1091 *cur++ = hi;
734 } 1092 }
735 break; 1093 break;
… …
737 default: 1095 default:
738 --dec_cur; 1096 --dec_cur;
739 ERR ("illegal backslash escape sequence in string"); 1097 ERR ("illegal backslash escape sequence in string");
740 } 1098 }
741 } 1099 }
742 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1100 else if (expect_true (ch >= 0x20 && ch < 0x80))
743 *cur++ = ch; 1101 *cur++ = ch;
744 else if (ch >= 0x80) 1102 else if (ch >= 0x80)
745 { 1103 {
746 STRLEN clen; 1104 STRLEN clen;
747 UV uch;
748 1105
749 --dec_cur; 1106 --dec_cur;
750 1107
751 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1108 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
752 if (clen == (STRLEN)-1) 1109 if (clen == (STRLEN)-1)
753 ERR ("malformed UTF-8 character in JSON string"); 1110 ERR ("malformed UTF-8 character in JSON string");
754 1111
755 do 1112 do
756 *cur++ = *dec_cur++; 1113 *cur++ = *dec_cur++;
757 while (--clen); 1114 while (--clen);
758 1115
759 utf8 = 1; 1116 utf8 = 1;
760 } 1117 }
1118 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1119 *cur++ = ch;
761 else 1120 else
762 { 1121 {
763 --dec_cur; 1122 --dec_cur;
764 1123
765 if (!ch) 1124 if (!ch)
… …
773 { 1132 {
774 STRLEN len = cur - buf; 1133 STRLEN len = cur - buf;
775 1134
776 if (sv) 1135 if (sv)
777 { 1136 {
778 SvGROW (sv, SvCUR (sv) + len + 1); 1137 STRLEN cur = SvCUR (sv);
1138
1139 if (SvLEN (sv) <= cur + len)
1140 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1141
779 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1142 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
780 SvCUR_set (sv, SvCUR (sv) + len); 1143 SvCUR_set (sv, SvCUR (sv) + len);
781 } 1144 }
782 else 1145 else
783 sv = newSVpvn (buf, len); 1146 sv = newSVpvn (buf, len);
… …
868 is_nv = 1; 1231 is_nv = 1;
869 } 1232 }
870 1233
871 if (!is_nv) 1234 if (!is_nv)
872 { 1235 {
1236 int len = dec->cur - start;
1237
873 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1238 // special case the rather common 1..5-digit-int case
874 if (*start == '-') 1239 if (*start == '-')
875 switch (dec->cur - start) 1240 switch (len)
876 { 1241 {
877 case 2: return newSViv (-( start [1] - '0' * 1)); 1242 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
878 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1243 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
879 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1244 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
880 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1245 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1246 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
881 } 1247 }
882 else 1248 else
883 switch (dec->cur - start) 1249 switch (len)
884 { 1250 {
885 case 1: return newSViv ( start [0] - '0' * 1); 1251 case 1: return newSViv ( start [0] - '0' * 1);
886 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1252 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
887 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1253 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
888 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1254 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1255 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
889 } 1256 }
890 1257
891 { 1258 {
892 UV uv; 1259 UV uv;
893 int numtype = grok_number (start, dec->cur - start, &uv); 1260 int numtype = grok_number (start, len, &uv);
894 if (numtype & IS_NUMBER_IN_UV) 1261 if (numtype & IS_NUMBER_IN_UV)
895 if (numtype & IS_NUMBER_NEG) 1262 if (numtype & IS_NUMBER_NEG)
896 { 1263 {
897 if (uv < (UV)IV_MIN) 1264 if (uv < (UV)IV_MIN)
898 return newSViv (-(IV)uv); 1265 return newSViv (-(IV)uv);
899 } 1266 }
900 else 1267 else
901 return newSVuv (uv); 1268 return newSVuv (uv);
902
903 // here would likely be the place for bigint support
904 } 1269 }
905 }
906 1270
907 // if we ever support bigint or bigfloat, this is the place for bigfloat 1271 len -= *start == '-' ? 1 : 0;
1272
1273 // does not fit into IV or UV, try NV
1274 if (len <= NV_DIG)
1275 // fits into NV without loss of precision
1276 return newSVnv (json_atof (start));
1277
1278 // everything else fails, convert it to a string
1279 return newSVpvn (start, dec->cur - start);
1280 }
1281
1282 // loss of precision here
908 return newSVnv (Atof (start)); 1283 return newSVnv (json_atof (start));
909 1284
910fail: 1285fail:
911 return 0; 1286 return 0;
912} 1287}
913 1288
… …
937 if (*dec->cur == ']') 1312 if (*dec->cur == ']')
938 { 1313 {
939 ++dec->cur; 1314 ++dec->cur;
940 break; 1315 break;
941 } 1316 }
942 1317
943 if (*dec->cur != ',') 1318 if (*dec->cur != ',')
944 ERR (", or ] expected while parsing array"); 1319 ERR (", or ] expected while parsing array");
945 1320
946 ++dec->cur; 1321 ++dec->cur;
1322
1323 decode_ws (dec);
1324
1325 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1326 {
1327 ++dec->cur;
1328 break;
1329 }
947 } 1330 }
948 1331
949 DEC_DEC_DEPTH; 1332 DEC_DEC_DEPTH;
950 return newRV_noinc ((SV *)av); 1333 return newRV_noinc ((SV *)av);
951 1334
… …
956} 1339}
957 1340
958static SV * 1341static SV *
959decode_hv (dec_t *dec) 1342decode_hv (dec_t *dec)
960{ 1343{
1344 SV *sv;
961 HV *hv = newHV (); 1345 HV *hv = newHV ();
962 1346
963 DEC_INC_DEPTH; 1347 DEC_INC_DEPTH;
964 decode_ws (dec); 1348 decode_ws (dec);
965 1349
966 if (*dec->cur == '}') 1350 if (*dec->cur == '}')
967 ++dec->cur; 1351 ++dec->cur;
968 else 1352 else
969 for (;;) 1353 for (;;)
970 { 1354 {
1355 EXPECT_CH ('"');
1356
1357 // heuristic: assume that
1358 // a) decode_str + hv_store_ent are abysmally slow.
1359 // b) most hash keys are short, simple ascii text.
1360 // => try to "fast-match" such strings to avoid
1361 // the overhead of decode_str + hv_store_ent.
1362 {
971 SV *key, *value; 1363 SV *value;
1364 char *p = dec->cur;
1365 char *e = p + 24; // only try up to 24 bytes
972 1366
973 decode_ws (dec); EXPECT_CH ('"'); 1367 for (;;)
974
975 key = decode_str (dec);
976 if (!key)
977 goto fail;
978
979 decode_ws (dec); EXPECT_CH (':');
980
981 value = decode_sv (dec);
982 if (!value)
983 { 1368 {
1369 // the >= 0x80 is false on most architectures
1370 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1371 {
1372 // slow path, back up and use decode_str
1373 SV *key = decode_str (dec);
1374 if (!key)
1375 goto fail;
1376
1377 decode_ws (dec); EXPECT_CH (':');
1378
1379 decode_ws (dec);
1380 value = decode_sv (dec);
1381 if (!value)
1382 {
1383 SvREFCNT_dec (key);
1384 goto fail;
1385 }
1386
1387 hv_store_ent (hv, key, value, 0);
984 SvREFCNT_dec (key); 1388 SvREFCNT_dec (key);
1389
1390 break;
1391 }
1392 else if (*p == '"')
1393 {
1394 // fast path, got a simple key
1395 char *key = dec->cur;
1396 int len = p - key;
1397 dec->cur = p + 1;
1398
1399 decode_ws (dec); EXPECT_CH (':');
1400
1401 decode_ws (dec);
1402 value = decode_sv (dec);
1403 if (!value)
985 goto fail; 1404 goto fail;
1405
1406 hv_store (hv, key, len, value, 0);
1407
1408 break;
1409 }
1410
1411 ++p;
986 } 1412 }
987 1413 }
988 hv_store_ent (hv, key, value, 0);
989 SvREFCNT_dec (key);
990 1414
991 decode_ws (dec); 1415 decode_ws (dec);
992 1416
993 if (*dec->cur == '}') 1417 if (*dec->cur == '}')
994 { 1418 {
… …
998 1422
999 if (*dec->cur != ',') 1423 if (*dec->cur != ',')
1000 ERR (", or } expected while parsing object/hash"); 1424 ERR (", or } expected while parsing object/hash");
1001 1425
1002 ++dec->cur; 1426 ++dec->cur;
1427
1428 decode_ws (dec);
1429
1430 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1431 {
1432 ++dec->cur;
1433 break;
1434 }
1003 } 1435 }
1004 1436
1005 DEC_DEC_DEPTH; 1437 DEC_DEC_DEPTH;
1006 return newRV_noinc ((SV *)hv); 1438 sv = newRV_noinc ((SV *)hv);
1439
1440 // check filter callbacks
1441 if (dec->json.flags & F_HOOK)
1442 {
1443 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1444 {
1445 HE *cb, *he;
1446
1447 hv_iterinit (hv);
1448 he = hv_iternext (hv);
1449 hv_iterinit (hv);
1450
1451 // the next line creates a mortal sv each time it's called.
1452 // might want to optimise this for common cases.
1453 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1454
1455 if (cb)
1456 {
1457 dSP;
1458 int count;
1459
1460 ENTER; SAVETMPS;
1461 SAVESTACK_POS ();
1462 PUSHMARK (SP);
1463 XPUSHs (HeVAL (he));
1464 sv_2mortal (sv);
1465
1466 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1467
1468 if (count == 1)
1469 {
1470 sv = newSVsv (POPs);
1471 FREETMPS; LEAVE;
1472 return sv;
1473 }
1474
1475 SvREFCNT_inc (sv);
1476 FREETMPS; LEAVE;
1477 }
1478 }
1479
1480 if (dec->json.cb_object)
1481 {
1482 dSP;
1483 int count;
1484
1485 ENTER; SAVETMPS;
1486 SAVESTACK_POS ();
1487 PUSHMARK (SP);
1488 XPUSHs (sv_2mortal (sv));
1489
1490 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1491
1492 if (count == 1)
1493 {
1494 sv = newSVsv (POPs);
1495 FREETMPS; LEAVE;
1496 return sv;
1497 }
1498
1499 SvREFCNT_inc (sv);
1500 FREETMPS; LEAVE;
1501 }
1502 }
1503
1504 return sv;
1007 1505
1008fail: 1506fail:
1009 SvREFCNT_dec (hv); 1507 SvREFCNT_dec (hv);
1010 DEC_DEC_DEPTH; 1508 DEC_DEC_DEPTH;
1011 return 0; 1509 return 0;
1012} 1510}
1013 1511
1014static SV * 1512static SV *
1513decode_tag (dec_t *dec)
1514{
1515 SV *tag = 0;
1516 SV *val = 0;
1517
1518 if (!(dec->json.flags & F_ALLOW_TAGS))
1519 ERR ("malformed JSON string, neither array, object, number, string or atom");
1520
1521 ++dec->cur;
1522
1523 decode_ws (dec);
1524
1525 tag = decode_sv (dec);
1526 if (!tag)
1527 goto fail;
1528
1529 if (!SvPOK (tag))
1530 ERR ("malformed JSON string, (tag) must be a string");
1531
1532 decode_ws (dec);
1533
1534 if (*dec->cur != ')')
1535 ERR (") expected after tag");
1536
1537 ++dec->cur;
1538
1539 decode_ws (dec);
1540
1541 val = decode_sv (dec);
1542 if (!val)
1543 goto fail;
1544
1545 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1546 ERR ("malformed JSON string, tag value must be an array");
1547
1548 {
1549 AV *av = (AV *)SvRV (val);
1550 int i, len = av_len (av) + 1;
1551 HV *stash = gv_stashsv (tag, 0);
1552 SV *sv;
1553
1554 if (!stash)
1555 ERR ("cannot decode perl-object (package does not exist)");
1556
1557 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1558
1559 if (!method)
1560 ERR ("cannot decode perl-object (package does not have a THAW method)");
1561
1562 dSP;
1563
1564 ENTER; SAVETMPS;
1565 PUSHMARK (SP);
1566 EXTEND (SP, len + 2);
1567 // we re-bless the reference to get overload and other niceties right
1568 PUSHs (tag);
1569 PUSHs (sv_json);
1570
1571 for (i = 0; i < len; ++i)
1572 PUSHs (*av_fetch (av, i, 1));
1573
1574 PUTBACK;
1575 call_sv ((SV *)GvCV (method), G_SCALAR);
1576 SPAGAIN;
1577
1578 SvREFCNT_dec (tag);
1579 SvREFCNT_dec (val);
1580 sv = SvREFCNT_inc (POPs);
1581
1582 PUTBACK;
1583
1584 FREETMPS; LEAVE;
1585
1586 return sv;
1587 }
1588
1589fail:
1590 SvREFCNT_dec (tag);
1591 SvREFCNT_dec (val);
1592 return 0;
1593}
1594
1595static SV *
1015decode_sv (dec_t *dec) 1596decode_sv (dec_t *dec)
1016{ 1597{
1017 decode_ws (dec);
1018
1019 // the beauty of JSON: you need exactly one character lookahead 1598 // the beauty of JSON: you need exactly one character lookahead
1020 // to parse anything. 1599 // to parse everything.
1021 switch (*dec->cur) 1600 switch (*dec->cur)
1022 { 1601 {
1023 case '"': ++dec->cur; return decode_str (dec); 1602 case '"': ++dec->cur; return decode_str (dec);
1024 case '[': ++dec->cur; return decode_av (dec); 1603 case '[': ++dec->cur; return decode_av (dec);
1025 case '{': ++dec->cur; return decode_hv (dec); 1604 case '{': ++dec->cur; return decode_hv (dec);
1605 case '(': return decode_tag (dec);
1026 1606
1027 case '-': 1607 case '-':
1028 case '0': case '1': case '2': case '3': case '4': 1608 case '0': case '1': case '2': case '3': case '4':
1029 case '5': case '6': case '7': case '8': case '9': 1609 case '5': case '6': case '7': case '8': case '9':
1030 return decode_num (dec); 1610 return decode_num (dec);
1031 1611
1032 case 't': 1612 case 't':
1033 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1613 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1034 { 1614 {
1035 dec->cur += 4; 1615 dec->cur += 4;
1616#if JSON_SLOW
1617 types_true = get_bool ("Types::Serialiser::true");
1618#endif
1036 return newSViv (1); 1619 return newSVsv (types_true);
1037 } 1620 }
1038 else 1621 else
1039 ERR ("'true' expected"); 1622 ERR ("'true' expected");
1040 1623
1041 break; 1624 break;
1042 1625
1043 case 'f': 1626 case 'f':
1044 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1627 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1045 { 1628 {
1046 dec->cur += 5; 1629 dec->cur += 5;
1630#if JSON_SLOW
1631 types_false = get_bool ("Types::Serialiser::false");
1632#endif
1047 return newSViv (0); 1633 return newSVsv (types_false);
1048 } 1634 }
1049 else 1635 else
1050 ERR ("'false' expected"); 1636 ERR ("'false' expected");
1051 1637
1052 break; 1638 break;
… …
1061 ERR ("'null' expected"); 1647 ERR ("'null' expected");
1062 1648
1063 break; 1649 break;
1064 1650
1065 default: 1651 default:
1066 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1652 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1067 break; 1653 break;
1068 } 1654 }
1069 1655
1070fail: 1656fail:
1071 return 0; 1657 return 0;
1072} 1658}
1073 1659
1074static SV * 1660static SV *
1075decode_json (SV *string, U32 flags, UV *offset_return) 1661decode_json (SV *string, JSON *json, char **offset_return)
1076{ 1662{
1077 dec_t dec; 1663 dec_t dec;
1078 UV offset;
1079 SV *sv; 1664 SV *sv;
1080 1665
1666 /* work around bugs in 5.10 where manipulating magic values
1667 * makes perl ignore the magic in subsequent accesses.
1668 * also make a copy of non-PV values, to get them into a clean
1669 * state (SvPV should do that, but it's buggy, see below).
1670 */
1081 SvGETMAGIC (string); 1671 /*SvGETMAGIC (string);*/
1672 if (SvMAGICAL (string) || !SvPOK (string))
1673 string = sv_2mortal (newSVsv (string));
1674
1082 SvUPGRADE (string, SVt_PV); 1675 SvUPGRADE (string, SVt_PV);
1083 1676
1084 if (flags & F_UTF8) 1677 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1678 * assertion with -DDEBUGGING, although SvCUR is documented to
1679 * return the xpv_cur field which certainly exists after upgrading.
1680 * according to nicholas clark, calling SvPOK fixes this.
1681 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1682 * and hope for the best.
1683 * Damnit, SvPV_nolen still trips over yet another assertion. This
1684 * assertion business is seriously broken, try yet another workaround
1685 * for the broken -DDEBUGGING.
1686 */
1687 {
1688#ifdef DEBUGGING
1689 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1690#else
1691 STRLEN offset = SvCUR (string);
1692#endif
1693
1694 if (offset > json->max_size && json->max_size)
1695 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1696 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1697 }
1698
1699 if (DECODE_WANTS_OCTETS (json))
1085 sv_utf8_downgrade (string, 0); 1700 sv_utf8_downgrade (string, 0);
1086 else 1701 else
1087 sv_utf8_upgrade (string); 1702 sv_utf8_upgrade (string);
1088 1703
1089 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1704 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1090 1705
1091 dec.flags = flags; 1706 dec.json = *json;
1092 dec.cur = SvPVX (string); 1707 dec.cur = SvPVX (string);
1093 dec.end = SvEND (string); 1708 dec.end = SvEND (string);
1094 dec.err = 0; 1709 dec.err = 0;
1095 dec.depth = 0; 1710 dec.depth = 0;
1096 dec.maxdepth = DEC_DEPTH (dec.flags); 1711
1712 if (dec.json.cb_object || dec.json.cb_sk_object)
1713 dec.json.flags |= F_HOOK;
1097 1714
1098 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1715 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1716
1717 decode_ws (&dec);
1099 sv = decode_sv (&dec); 1718 sv = decode_sv (&dec);
1719
1720 if (offset_return)
1721 *offset_return = dec.cur;
1100 1722
1101 if (!(offset_return || !sv)) 1723 if (!(offset_return || !sv))
1102 { 1724 {
1103 // check for trailing garbage 1725 // check for trailing garbage
1104 decode_ws (&dec); 1726 decode_ws (&dec);
… …
1107 { 1729 {
1108 dec.err = "garbage after JSON object"; 1730 dec.err = "garbage after JSON object";
1109 SvREFCNT_dec (sv); 1731 SvREFCNT_dec (sv);
1110 sv = 0; 1732 sv = 0;
1111 } 1733 }
1112 }
1113
1114 if (offset_return || !sv)
1115 {
1116 offset = dec.flags & F_UTF8
1117 ? dec.cur - SvPVX (string)
1118 : utf8_distance (dec.cur, SvPVX (string));
1119
1120 if (offset_return)
1121 *offset_return = offset;
1122 } 1734 }
1123 1735
1124 if (!sv) 1736 if (!sv)
1125 { 1737 {
1126 SV *uni = sv_newmortal (); 1738 SV *uni = sv_newmortal ();
… …
1132 SAVEVPTR (PL_curcop); 1744 SAVEVPTR (PL_curcop);
1133 PL_curcop = &cop; 1745 PL_curcop = &cop;
1134 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1746 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1135 LEAVE; 1747 LEAVE;
1136 1748
1137 croak ("%s, at character offset %d [\"%s\"]", 1749 croak ("%s, at character offset %d (before \"%s\")",
1138 dec.err, 1750 dec.err,
1139 (int)offset, 1751 (int)ptr_to_index (string, dec.cur),
1140 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1752 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1141 } 1753 }
1142 1754
1143 sv = sv_2mortal (sv); 1755 sv = sv_2mortal (sv);
1144 1756
1145 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1757 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1146 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1758 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1147 1759
1148 return sv; 1760 return sv;
1761}
1762
1763/////////////////////////////////////////////////////////////////////////////
1764// incremental parser
1765
1766static void
1767incr_parse (JSON *self)
1768{
1769 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1770
1771 // the state machine here is a bit convoluted and could be simplified a lot
1772 // but this would make it slower, so...
1773
1774 for (;;)
1775 {
1776 //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
1777 switch (self->incr_mode)
1778 {
1779 // only used for initial whitespace skipping
1780 case INCR_M_WS:
1781 for (;;)
1782 {
1783 if (*p > 0x20)
1784 {
1785 if (*p == '#')
1786 {
1787 self->incr_mode = INCR_M_C0;
1788 goto incr_m_c;
1789 }
1790 else
1791 {
1792 self->incr_mode = INCR_M_JSON;
1793 goto incr_m_json;
1794 }
1795 }
1796 else if (!*p)
1797 goto interrupt;
1798
1799 ++p;
1800 }
1801
1802 // skip a single char inside a string (for \\-processing)
1803 case INCR_M_BS:
1804 if (!*p)
1805 goto interrupt;
1806
1807 ++p;
1808 self->incr_mode = INCR_M_STR;
1809 goto incr_m_str;
1810
1811 // inside #-style comments
1812 case INCR_M_C0:
1813 case INCR_M_C1:
1814 incr_m_c:
1815 for (;;)
1816 {
1817 if (*p == '\n')
1818 {
1819 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1820 break;
1821 }
1822 else if (!*p)
1823 goto interrupt;
1824
1825 ++p;
1826 }
1827
1828 break;
1829
1830 // inside a string
1831 case INCR_M_STR:
1832 incr_m_str:
1833 for (;;)
1834 {
1835 if (*p == '"')
1836 {
1837 ++p;
1838 self->incr_mode = INCR_M_JSON;
1839
1840 if (!self->incr_nest)
1841 goto interrupt;
1842
1843 goto incr_m_json;
1844 }
1845 else if (*p == '\\')
1846 {
1847 ++p; // "virtually" consumes character after \
1848
1849 if (!*p) // if at end of string we have to switch modes
1850 {
1851 self->incr_mode = INCR_M_BS;
1852 goto interrupt;
1853 }
1854 }
1855 else if (!*p)
1856 goto interrupt;
1857
1858 ++p;
1859 }
1860
1861 // after initial ws, outside string
1862 case INCR_M_JSON:
1863 incr_m_json:
1864 for (;;)
1865 {
1866 switch (*p++)
1867 {
1868 case 0:
1869 --p;
1870 goto interrupt;
1871
1872 case 0x09:
1873 case 0x0a:
1874 case 0x0d:
1875 case 0x20:
1876 if (!self->incr_nest)
1877 {
1878 --p; // do not eat the whitespace, let the next round do it
1879 goto interrupt;
1880 }
1881 break;
1882
1883 case '"':
1884 self->incr_mode = INCR_M_STR;
1885 goto incr_m_str;
1886
1887 case '[':
1888 case '{':
1889 case '(':
1890 if (++self->incr_nest > self->max_depth)
1891 croak (ERR_NESTING_EXCEEDED);
1892 break;
1893
1894 case ']':
1895 case '}':
1896 if (--self->incr_nest <= 0)
1897 goto interrupt;
1898 break;
1899
1900 case ')':
1901 --self->incr_nest;
1902 break;
1903
1904 case '#':
1905 self->incr_mode = INCR_M_C1;
1906 goto incr_m_c;
1907 }
1908 }
1909 }
1910
1911 modechange:
1912 ;
1913 }
1914
1915interrupt:
1916 self->incr_pos = p - SvPVX (self->incr_text);
1917 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1918 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1149} 1919}
1150 1920
1151///////////////////////////////////////////////////////////////////////////// 1921/////////////////////////////////////////////////////////////////////////////
1152// XS interface functions 1922// XS interface functions
1153 1923
… …
1162 i >= '0' && i <= '9' ? i - '0' 1932 i >= '0' && i <= '9' ? i - '0'
1163 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1933 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1164 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1934 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1165 : -1; 1935 : -1;
1166 1936
1167 json_stash = gv_stashpv ("JSON::XS", 1); 1937 json_stash = gv_stashpv ("JSON::XS" , 1);
1938 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1939
1940 types_true = get_bool ("Types::Serialiser::true");
1941 types_false = get_bool ("Types::Serialiser::false");
1942
1943 sv_json = newSVpv ("JSON", 0);
1944 SvREADONLY_on (sv_json);
1945
1946 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1168} 1947}
1169 1948
1170PROTOTYPES: DISABLE 1949PROTOTYPES: DISABLE
1171 1950
1172SV *new (char *dummy) 1951void CLONE (...)
1173 CODE: 1952 CODE:
1174 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1953 json_stash = 0;
1954 types_boolean_stash = 0;
1955
1956void new (char *klass)
1957 PPCODE:
1958{
1959 SV *pv = NEWSV (0, sizeof (JSON));
1960 SvPOK_only (pv);
1961 json_init ((JSON *)SvPVX (pv));
1962 XPUSHs (sv_2mortal (sv_bless (
1963 newRV_noinc (pv),
1964 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1965 )));
1966}
1967
1968void ascii (JSON *self, int enable = 1)
1969 ALIAS:
1970 ascii = F_ASCII
1971 latin1 = F_LATIN1
1972 utf8 = F_UTF8
1973 indent = F_INDENT
1974 canonical = F_CANONICAL
1975 space_before = F_SPACE_BEFORE
1976 space_after = F_SPACE_AFTER
1977 pretty = F_PRETTY
1978 allow_nonref = F_ALLOW_NONREF
1979 shrink = F_SHRINK
1980 allow_blessed = F_ALLOW_BLESSED
1981 convert_blessed = F_CONV_BLESSED
1982 relaxed = F_RELAXED
1983 allow_unknown = F_ALLOW_UNKNOWN
1984 allow_tags = F_ALLOW_TAGS
1985 PPCODE:
1986{
1987 if (enable)
1988 self->flags |= ix;
1989 else
1990 self->flags &= ~ix;
1991
1992 XPUSHs (ST (0));
1993}
1994
1995void get_ascii (JSON *self)
1996 ALIAS:
1997 get_ascii = F_ASCII
1998 get_latin1 = F_LATIN1
1999 get_utf8 = F_UTF8
2000 get_indent = F_INDENT
2001 get_canonical = F_CANONICAL
2002 get_space_before = F_SPACE_BEFORE
2003 get_space_after = F_SPACE_AFTER
2004 get_allow_nonref = F_ALLOW_NONREF
2005 get_shrink = F_SHRINK
2006 get_allow_blessed = F_ALLOW_BLESSED
2007 get_convert_blessed = F_CONV_BLESSED
2008 get_relaxed = F_RELAXED
2009 get_allow_unknown = F_ALLOW_UNKNOWN
2010 get_allow_tags = F_ALLOW_TAGS
2011 PPCODE:
2012 XPUSHs (boolSV (self->flags & ix));
2013
2014void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
2015 PPCODE:
2016 self->max_depth = max_depth;
2017 XPUSHs (ST (0));
2018
2019U32 get_max_depth (JSON *self)
2020 CODE:
2021 RETVAL = self->max_depth;
1175 OUTPUT: 2022 OUTPUT:
1176 RETVAL 2023 RETVAL
1177 2024
1178SV *ascii (SV *self, int enable = 1) 2025void max_size (JSON *self, U32 max_size = 0)
1179 ALIAS: 2026 PPCODE:
1180 ascii = F_ASCII 2027 self->max_size = max_size;
1181 latin1 = F_LATIN1 2028 XPUSHs (ST (0));
1182 utf8 = F_UTF8 2029
1183 indent = F_INDENT 2030int get_max_size (JSON *self)
1184 canonical = F_CANONICAL
1185 space_before = F_SPACE_BEFORE
1186 space_after = F_SPACE_AFTER
1187 pretty = F_PRETTY
1188 allow_nonref = F_ALLOW_NONREF
1189 shrink = F_SHRINK
1190 CODE: 2031 CODE:
1191{ 2032 RETVAL = self->max_size;
1192 UV *uv = SvJSON (self);
1193 if (enable)
1194 *uv |= ix;
1195 else
1196 *uv &= ~ix;
1197
1198 RETVAL = newSVsv (self);
1199}
1200 OUTPUT: 2033 OUTPUT:
1201 RETVAL 2034 RETVAL
1202 2035
1203SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 2036void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
2037 PPCODE:
2038{
2039 SvREFCNT_dec (self->cb_object);
2040 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
2041
2042 XPUSHs (ST (0));
2043}
2044
2045void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
2046 PPCODE:
2047{
2048 if (!self->cb_sk_object)
2049 self->cb_sk_object = newHV ();
2050
2051 if (SvOK (cb))
2052 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
2053 else
2054 {
2055 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
2056
2057 if (!HvKEYS (self->cb_sk_object))
2058 {
2059 SvREFCNT_dec (self->cb_sk_object);
2060 self->cb_sk_object = 0;
2061 }
2062 }
2063
2064 XPUSHs (ST (0));
2065}
2066
2067void encode (JSON *self, SV *scalar)
2068 PPCODE:
2069 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2070 XPUSHs (scalar);
2071
2072void decode (JSON *self, SV *jsonstr)
2073 PPCODE:
2074 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2075 XPUSHs (jsonstr);
2076
2077void decode_prefix (JSON *self, SV *jsonstr)
2078 PPCODE:
2079{
2080 SV *sv;
2081 char *offset;
2082 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
2083 EXTEND (SP, 2);
2084 PUSHs (sv);
2085 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
2086}
2087
2088void incr_parse (JSON *self, SV *jsonstr = 0)
2089 PPCODE:
2090{
2091 if (!self->incr_text)
2092 self->incr_text = newSVpvn ("", 0);
2093
2094 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2095 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2096 if (DECODE_WANTS_OCTETS (self))
2097 {
2098 if (self->incr_pos)
2099 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2100 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2101
2102 sv_utf8_downgrade (self->incr_text, 0);
2103 }
2104 else
2105 {
2106 sv_utf8_upgrade (self->incr_text);
2107
2108 if (self->incr_pos)
2109 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2110 - (U8 *)SvPVX (self->incr_text);
2111 }
2112
2113 // append data, if any
2114 if (jsonstr)
2115 {
2116 /* make sure both strings have same encoding */
2117 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2118 if (SvUTF8 (jsonstr))
2119 sv_utf8_downgrade (jsonstr, 0);
2120 else
2121 sv_utf8_upgrade (jsonstr);
2122
2123 /* and then just blindly append */
2124 {
2125 STRLEN len;
2126 const char *str = SvPV (jsonstr, len);
2127 STRLEN cur = SvCUR (self->incr_text);
2128
2129 if (SvLEN (self->incr_text) <= cur + len)
2130 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
2131
2132 Move (str, SvEND (self->incr_text), len, char);
2133 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2134 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2135 }
2136 }
2137
2138 if (GIMME_V != G_VOID)
2139 do
2140 {
2141 SV *sv;
2142 char *offset;
2143
2144 if (!INCR_DONE (self))
2145 {
2146 incr_parse (self);
2147
2148 if (self->incr_pos > self->max_size && self->max_size)
2149 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2150 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2151
2152 if (!INCR_DONE (self))
2153 {
2154 // as an optimisation, do not accumulate white space in the incr buffer
2155 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2156 {
2157 self->incr_pos = 0;
2158 SvCUR_set (self->incr_text, 0);
2159 }
2160
2161 break;
2162 }
2163 }
2164
2165 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2166 XPUSHs (sv);
2167
2168 self->incr_pos -= offset - SvPVX (self->incr_text);
2169 self->incr_nest = 0;
2170 self->incr_mode = 0;
2171
2172 sv_chop (self->incr_text, offset);
2173 }
2174 while (GIMME_V == G_ARRAY);
2175}
2176
2177SV *incr_text (JSON *self)
2178 ATTRS: lvalue
1204 CODE: 2179 CODE:
1205{ 2180{
1206 UV *uv = SvJSON (self); 2181 if (self->incr_pos)
1207 UV log2 = 0; 2182 croak ("incr_text can not be called when the incremental parser already started parsing");
1208 2183
1209 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 2184 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1210
1211 while ((1UL << log2) < max_depth)
1212 ++log2;
1213
1214 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1215
1216 RETVAL = newSVsv (self);
1217} 2185}
1218 OUTPUT: 2186 OUTPUT:
1219 RETVAL 2187 RETVAL
1220 2188
2189void incr_skip (JSON *self)
2190 CODE:
2191{
2192 if (self->incr_pos)
2193 {
2194 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2195 self->incr_pos = 0;
2196 self->incr_nest = 0;
2197 self->incr_mode = 0;
2198 }
2199}
2200
2201void incr_reset (JSON *self)
2202 CODE:
2203{
2204 SvREFCNT_dec (self->incr_text);
2205 self->incr_text = 0;
2206 self->incr_pos = 0;
2207 self->incr_nest = 0;
2208 self->incr_mode = 0;
2209}
2210
2211void DESTROY (JSON *self)
2212 CODE:
2213 SvREFCNT_dec (self->cb_sk_object);
2214 SvREFCNT_dec (self->cb_object);
2215 SvREFCNT_dec (self->incr_text);
2216
2217PROTOTYPES: ENABLE
2218
1221void encode (SV *self, SV *scalar) 2219void encode_json (SV *scalar)
1222 PPCODE: 2220 PPCODE:
1223 XPUSHs (encode_json (scalar, *SvJSON (self))); 2221{
2222 JSON json;
2223 json_init (&json);
2224 json.flags |= F_UTF8;
2225 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2226 XPUSHs (scalar);
2227}
1224 2228
1225void decode (SV *self, SV *jsonstr) 2229void decode_json (SV *jsonstr)
1226 PPCODE: 2230 PPCODE:
1227 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1228
1229void decode_prefix (SV *self, SV *jsonstr)
1230 PPCODE:
1231{ 2231{
1232 UV offset; 2232 JSON json;
1233 EXTEND (SP, 2); 2233 json_init (&json);
1234 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 2234 json.flags |= F_UTF8;
1235 PUSHs (sv_2mortal (newSVuv (offset))); 2235 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2236 XPUSHs (jsonstr);
1236} 2237}
1237 2238
1238PROTOTYPES: ENABLE
1239
1240void to_json (SV *scalar)
1241 ALIAS:
1242 objToJson = 0
1243 PPCODE:
1244 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1245
1246void from_json (SV *jsonstr)
1247 ALIAS:
1248 jsonToObj = 0
1249 PPCODE:
1250 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1251

Diff Legend

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