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 1.20 by root, Sun Mar 25 21:53:19 2007 UTC vs.
Revision 1.103 by root, Wed Jan 6 08:01:39 2010 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>
9#include <limits.h>
10#include <float.h>
8 11
12#if defined(__BORLANDC__) || defined(_MSC_VER)
13# define snprintf _snprintf // C compilers have this in stdio.h
14#endif
15
16// some old perls do not have this, try to make it work, no
17// guarantees, though. if it breaks, you get to keep the pieces.
18#ifndef UTF8_MAXBYTES
19# define UTF8_MAXBYTES 13
20#endif
21
22// three extra for rounding, sign, and end of string
23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
24
9#define F_ASCII 0x00000001UL 25#define F_ASCII 0x00000001UL
26#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002UL 27#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004UL 28#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008UL 29#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010UL 30#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020UL 31#define F_SPACE_AFTER 0x00000040UL
15#define F_ALLOW_NONREF 0x00000080UL 32#define F_ALLOW_NONREF 0x00000100UL
16#define F_SHRINK 0x00000100UL 33#define F_SHRINK 0x00000200UL
34#define F_ALLOW_BLESSED 0x00000400UL
35#define F_CONV_BLESSED 0x00000800UL
17#define F_MAXDEPTH 0xf8000000UL 36#define F_RELAXED 0x00001000UL
18#define S_MAXDEPTH 27 37#define F_ALLOW_UNKNOWN 0x00002000UL
19 38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
20#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
21
22// F_SELFCONVERT? <=> to_json/toJson
23// F_BLESSED? <=> { $__class__$ => }
24 39
25#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
26#define F_DEFAULT (13UL << S_MAXDEPTH)
27 41
28#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
29#define INDENT_STEP 3 // spaces per indentation level 43#define INDENT_STEP 3 // spaces per indentation level
30 44
31#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
32#define SHORT_STRING_LEN 512 // special-case strings of up to this size 45#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
33 46
34#define SB do { 47#define SB do {
35#define SE } while (0) 48#define SE } while (0)
36 49
50#if __GNUC__ >= 3
51# define expect(expr,value) __builtin_expect ((expr), (value))
52# define INLINE static inline
53#else
54# define expect(expr,value) (expr)
55# define INLINE static
56#endif
57
58#define expect_false(expr) expect ((expr) != 0, 0)
59#define expect_true(expr) expect ((expr) != 0, 1)
60
61#define IN_RANGE_INC(type,val,beg,end) \
62 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
63 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
64
65#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
66
67#ifdef USE_ITHREADS
68# define JSON_SLOW 1
69# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
70#else
71# define JSON_SLOW 0
72# define JSON_STASH json_stash
73#endif
74
37static HV *json_stash; // JSON::XS:: 75static HV *json_stash, *json_boolean_stash; // JSON::XS::
76static SV *json_true, *json_false;
77
78enum {
79 INCR_M_WS = 0, // initial whitespace skipping, must be 0
80 INCR_M_STR, // inside string
81 INCR_M_BS, // inside backslash
82 INCR_M_C0, // inside comment in initial whitespace sequence
83 INCR_M_C1, // inside comment in other places
84 INCR_M_JSON // outside anything, count nesting
85};
86
87#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
88
89typedef struct {
90 U32 flags;
91 U32 max_depth;
92 STRLEN max_size;
93
94 SV *cb_object;
95 HV *cb_sk_object;
96
97 // for the incremental parser
98 SV *incr_text; // the source text so far
99 STRLEN incr_pos; // the current offset into the text
100 int incr_nest; // {[]}-nesting level
101 unsigned char incr_mode;
102} JSON;
103
104INLINE void
105json_init (JSON *json)
106{
107 Zero (json, 1, JSON);
108 json->max_depth = 512;
109}
38 110
39///////////////////////////////////////////////////////////////////////////// 111/////////////////////////////////////////////////////////////////////////////
40// utility functions 112// utility functions
41 113
42static UV * 114INLINE SV *
43SvJSON (SV *sv) 115get_bool (const char *name)
44{ 116{
45 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 117 SV *sv = get_sv (name, 1);
46 croak ("object is not of type JSON::XS");
47 118
48 return &SvUVX (SvRV (sv)); 119 SvREADONLY_on (sv);
49} 120 SvREADONLY_on (SvRV (sv));
50 121
51static void 122 return sv;
123}
124
125INLINE void
52shrink (SV *sv) 126shrink (SV *sv)
53{ 127{
54 sv_utf8_downgrade (sv, 1); 128 sv_utf8_downgrade (sv, 1);
129
55 if (SvLEN (sv) > SvCUR (sv) + 1) 130 if (SvLEN (sv) > SvCUR (sv) + 1)
56 { 131 {
57#ifdef SvPV_shrink_to_cur 132#ifdef SvPV_shrink_to_cur
58 SvPV_shrink_to_cur (sv); 133 SvPV_shrink_to_cur (sv);
59#elif defined (SvPV_renew) 134#elif defined (SvPV_renew)
… …
65// decode an utf-8 character and return it, or (UV)-1 in 140// decode an utf-8 character and return it, or (UV)-1 in
66// case of an error. 141// case of an error.
67// we special-case "safe" characters from U+80 .. U+7FF, 142// we special-case "safe" characters from U+80 .. U+7FF,
68// but use the very good perl function to parse anything else. 143// but use the very good perl function to parse anything else.
69// note that we never call this function for a ascii codepoints 144// note that we never call this function for a ascii codepoints
70static UV 145INLINE UV
71decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 146decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
72{ 147{
73 if (s[0] > 0xdf || s[0] < 0xc2) 148 if (expect_true (len >= 2
74 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 149 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
75 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 150 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
76 { 151 {
77 *clen = 2; 152 *clen = 2;
78 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 153 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
79 } 154 }
80 else 155 else
81 return (UV)-1; 156 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
157}
158
159// likewise for encoding, also never called for ascii codepoints
160// this function takes advantage of this fact, although current gccs
161// seem to optimise the check for >= 0x80 away anyways
162INLINE unsigned char *
163encode_utf8 (unsigned char *s, UV ch)
164{
165 if (expect_false (ch < 0x000080))
166 *s++ = ch;
167 else if (expect_true (ch < 0x000800))
168 *s++ = 0xc0 | ( ch >> 6),
169 *s++ = 0x80 | ( ch & 0x3f);
170 else if ( ch < 0x010000)
171 *s++ = 0xe0 | ( ch >> 12),
172 *s++ = 0x80 | ((ch >> 6) & 0x3f),
173 *s++ = 0x80 | ( ch & 0x3f);
174 else if ( ch < 0x110000)
175 *s++ = 0xf0 | ( ch >> 18),
176 *s++ = 0x80 | ((ch >> 12) & 0x3f),
177 *s++ = 0x80 | ((ch >> 6) & 0x3f),
178 *s++ = 0x80 | ( ch & 0x3f);
179
180 return s;
181}
182
183// convert offset pointer to character index, sv must be string
184static STRLEN
185ptr_to_index (SV *sv, char *offset)
186{
187 return SvUTF8 (sv)
188 ? utf8_distance (offset, SvPVX (sv))
189 : offset - SvPVX (sv);
82} 190}
83 191
84///////////////////////////////////////////////////////////////////////////// 192/////////////////////////////////////////////////////////////////////////////
85// encoder 193// encoder
86 194
… …
88typedef struct 196typedef struct
89{ 197{
90 char *cur; // SvPVX (sv) + current output position 198 char *cur; // SvPVX (sv) + current output position
91 char *end; // SvEND (sv) 199 char *end; // SvEND (sv)
92 SV *sv; // result scalar 200 SV *sv; // result scalar
93 U32 flags; // F_* 201 JSON json;
94 U32 indent; // indentation level 202 U32 indent; // indentation level
95 U32 maxdepth; // max. indentation/recursion level 203 UV limit; // escape character values >= this value when encoding
96} enc_t; 204} enc_t;
97 205
98static void 206INLINE void
99need (enc_t *enc, STRLEN len) 207need (enc_t *enc, STRLEN len)
100{ 208{
101 if (enc->cur + len >= enc->end) 209 if (expect_false (enc->cur + len >= enc->end))
102 { 210 {
103 STRLEN cur = enc->cur - SvPVX (enc->sv); 211 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
104 SvGROW (enc->sv, cur + len + 1); 212 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
105 enc->cur = SvPVX (enc->sv) + cur; 213 enc->cur = SvPVX (enc->sv) + cur;
106 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 214 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
107 } 215 }
108} 216}
109 217
110static void 218INLINE void
111encode_ch (enc_t *enc, char ch) 219encode_ch (enc_t *enc, char ch)
112{ 220{
113 need (enc, 1); 221 need (enc, 1);
114 *enc->cur++ = ch; 222 *enc->cur++ = ch;
115} 223}
… …
123 231
124 while (str < end) 232 while (str < end)
125 { 233 {
126 unsigned char ch = *(unsigned char *)str; 234 unsigned char ch = *(unsigned char *)str;
127 235
128 if (ch >= 0x20 && ch < 0x80) // most common case 236 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
129 { 237 {
130 if (ch == '"') // but with slow exceptions 238 if (expect_false (ch == '"')) // but with slow exceptions
131 { 239 {
132 need (enc, len += 1); 240 need (enc, len += 1);
133 *enc->cur++ = '\\'; 241 *enc->cur++ = '\\';
134 *enc->cur++ = '"'; 242 *enc->cur++ = '"';
135 } 243 }
136 else if (ch == '\\') 244 else if (expect_false (ch == '\\'))
137 { 245 {
138 need (enc, len += 1); 246 need (enc, len += 1);
139 *enc->cur++ = '\\'; 247 *enc->cur++ = '\\';
140 *enc->cur++ = '\\'; 248 *enc->cur++ = '\\';
141 } 249 }
… …
159 STRLEN clen; 267 STRLEN clen;
160 UV uch; 268 UV uch;
161 269
162 if (is_utf8) 270 if (is_utf8)
163 { 271 {
164 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
165 uch = decode_utf8 (str, end - str, &clen); 272 uch = decode_utf8 (str, end - str, &clen);
166 if (clen == (STRLEN)-1) 273 if (clen == (STRLEN)-1)
167 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 274 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
168 } 275 }
169 else 276 else
170 { 277 {
171 uch = ch; 278 uch = ch;
172 clen = 1; 279 clen = 1;
173 } 280 }
174 281
175 if (uch > 0x10FFFFUL) 282 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
176 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
177
178 if (uch < 0x80 || enc->flags & F_ASCII)
179 { 283 {
180 if (uch > 0xFFFFUL) 284 if (uch >= 0x10000UL)
181 { 285 {
286 if (uch >= 0x110000UL)
287 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
288
182 need (enc, len += 11); 289 need (enc, len += 11);
183 sprintf (enc->cur, "\\u%04x\\u%04x", 290 sprintf (enc->cur, "\\u%04x\\u%04x",
184 (int)((uch - 0x10000) / 0x400 + 0xD800), 291 (int)((uch - 0x10000) / 0x400 + 0xD800),
185 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 292 (int)((uch - 0x10000) % 0x400 + 0xDC00));
186 enc->cur += 12; 293 enc->cur += 12;
187 } 294 }
188 else 295 else
189 { 296 {
190 static char hexdigit [16] = "0123456789abcdef";
191 need (enc, len += 5); 297 need (enc, len += 5);
192 *enc->cur++ = '\\'; 298 *enc->cur++ = '\\';
193 *enc->cur++ = 'u'; 299 *enc->cur++ = 'u';
194 *enc->cur++ = hexdigit [ uch >> 12 ]; 300 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
195 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 301 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
196 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 302 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
197 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 303 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
198 } 304 }
199 305
306 str += clen;
307 }
308 else if (enc->json.flags & F_LATIN1)
309 {
310 *enc->cur++ = uch;
200 str += clen; 311 str += clen;
201 } 312 }
202 else if (is_utf8) 313 else if (is_utf8)
203 { 314 {
204 need (enc, len += clen); 315 need (enc, len += clen);
… …
208 } 319 }
209 while (--clen); 320 while (--clen);
210 } 321 }
211 else 322 else
212 { 323 {
213 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 324 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
214 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 325 enc->cur = encode_utf8 (enc->cur, uch);
215 ++str; 326 ++str;
216 } 327 }
217 } 328 }
218 } 329 }
219 } 330 }
220 331
221 --len; 332 --len;
222 } 333 }
223} 334}
224 335
225static void 336INLINE void
226encode_indent (enc_t *enc) 337encode_indent (enc_t *enc)
227{ 338{
228 if (enc->flags & F_INDENT) 339 if (enc->json.flags & F_INDENT)
229 { 340 {
230 int spaces = enc->indent * INDENT_STEP; 341 int spaces = enc->indent * INDENT_STEP;
231 342
232 need (enc, spaces); 343 need (enc, spaces);
233 memset (enc->cur, ' ', spaces); 344 memset (enc->cur, ' ', spaces);
234 enc->cur += spaces; 345 enc->cur += spaces;
235 } 346 }
236} 347}
237 348
238static void 349INLINE void
239encode_space (enc_t *enc) 350encode_space (enc_t *enc)
240{ 351{
241 need (enc, 1); 352 need (enc, 1);
242 encode_ch (enc, ' '); 353 encode_ch (enc, ' ');
243} 354}
244 355
245static void 356INLINE void
246encode_nl (enc_t *enc) 357encode_nl (enc_t *enc)
247{ 358{
248 if (enc->flags & F_INDENT) 359 if (enc->json.flags & F_INDENT)
249 { 360 {
250 need (enc, 1); 361 need (enc, 1);
251 encode_ch (enc, '\n'); 362 encode_ch (enc, '\n');
252 } 363 }
253} 364}
254 365
255static void 366INLINE void
256encode_comma (enc_t *enc) 367encode_comma (enc_t *enc)
257{ 368{
258 encode_ch (enc, ','); 369 encode_ch (enc, ',');
259 370
260 if (enc->flags & F_INDENT) 371 if (enc->json.flags & F_INDENT)
261 encode_nl (enc); 372 encode_nl (enc);
262 else if (enc->flags & F_SPACE_AFTER) 373 else if (enc->json.flags & F_SPACE_AFTER)
263 encode_space (enc); 374 encode_space (enc);
264} 375}
265 376
266static void encode_sv (enc_t *enc, SV *sv); 377static void encode_sv (enc_t *enc, SV *sv);
267 378
268static void 379static void
269encode_av (enc_t *enc, AV *av) 380encode_av (enc_t *enc, AV *av)
270{ 381{
271 int i, len = av_len (av); 382 int i, len = av_len (av);
272 383
273 encode_ch (enc, '['); encode_nl (enc); 384 if (enc->indent >= enc->json.max_depth)
274 ++enc->indent; 385 croak (ERR_NESTING_EXCEEDED);
275 386
387 encode_ch (enc, '[');
388
389 if (len >= 0)
390 {
391 encode_nl (enc); ++enc->indent;
392
276 for (i = 0; i <= len; ++i) 393 for (i = 0; i <= len; ++i)
277 { 394 {
395 SV **svp = av_fetch (av, i, 0);
396
278 encode_indent (enc); 397 encode_indent (enc);
279 encode_sv (enc, *av_fetch (av, i, 0));
280 398
399 if (svp)
400 encode_sv (enc, *svp);
401 else
402 encode_str (enc, "null", 4, 0);
403
281 if (i < len) 404 if (i < len)
282 encode_comma (enc); 405 encode_comma (enc);
283 } 406 }
284 407
408 encode_nl (enc); --enc->indent; encode_indent (enc);
409 }
410
285 encode_nl (enc); 411 encode_ch (enc, ']');
286
287 --enc->indent;
288 encode_indent (enc); encode_ch (enc, ']');
289} 412}
290 413
291static void 414static void
292encode_he (enc_t *enc, HE *he) 415encode_hk (enc_t *enc, HE *he)
293{ 416{
294 encode_ch (enc, '"'); 417 encode_ch (enc, '"');
295 418
296 if (HeKLEN (he) == HEf_SVKEY) 419 if (HeKLEN (he) == HEf_SVKEY)
297 { 420 {
… …
307 else 430 else
308 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 431 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
309 432
310 encode_ch (enc, '"'); 433 encode_ch (enc, '"');
311 434
312 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 435 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
313 encode_ch (enc, ':'); 436 encode_ch (enc, ':');
314 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 437 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
315 encode_sv (enc, HeVAL (he));
316} 438}
317 439
318// compare hash entries, used when all keys are bytestrings 440// compare hash entries, used when all keys are bytestrings
319static int 441static int
320he_cmp_fast (const void *a_, const void *b_) 442he_cmp_fast (const void *a_, const void *b_)
… …
325 HE *b = *(HE **)b_; 447 HE *b = *(HE **)b_;
326 448
327 STRLEN la = HeKLEN (a); 449 STRLEN la = HeKLEN (a);
328 STRLEN lb = HeKLEN (b); 450 STRLEN lb = HeKLEN (b);
329 451
330 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 452 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
331 cmp = la - lb; 453 cmp = lb - la;
332 454
333 return cmp; 455 return cmp;
334} 456}
335 457
336// compare hash entries, used when some keys are sv's or utf-x 458// compare hash entries, used when some keys are sv's or utf-x
337static int 459static int
338he_cmp_slow (const void *a, const void *b) 460he_cmp_slow (const void *a, const void *b)
339{ 461{
340 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 462 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
341} 463}
342 464
343static void 465static void
344encode_hv (enc_t *enc, HV *hv) 466encode_hv (enc_t *enc, HV *hv)
345{ 467{
346 int count, i; 468 HE *he;
347 469
348 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 470 if (enc->indent >= enc->json.max_depth)
471 croak (ERR_NESTING_EXCEEDED);
349 472
350 if ((count = hv_iterinit (hv))) 473 encode_ch (enc, '{');
351 { 474
352 // for canonical output we have to sort by keys first 475 // for canonical output we have to sort by keys first
353 // actually, this is mostly due to the stupid so-called 476 // actually, this is mostly due to the stupid so-called
354 // security workaround added somewhere in 5.8.x. 477 // security workaround added somewhere in 5.8.x
355 // that randomises hash orderings 478 // that randomises hash orderings
356 if (enc->flags & F_CANONICAL) 479 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
480 {
481 int count = hv_iterinit (hv);
482
483 if (SvMAGICAL (hv))
357 { 484 {
485 // need to count by iterating. could improve by dynamically building the vector below
486 // but I don't care for the speed of this special case.
487 // note also that we will run into undefined behaviour when the two iterations
488 // do not result in the same count, something I might care for in some later release.
489
490 count = 0;
491 while (hv_iternext (hv))
492 ++count;
493
494 hv_iterinit (hv);
495 }
496
497 if (count)
498 {
499 int i, fast = 1;
500#if defined(__BORLANDC__) || defined(_MSC_VER)
501 HE **hes = _alloca (count * sizeof (HE));
502#else
358 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode 503 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
359 int fast = 1; 504#endif
360 505
361 i = 0; 506 i = 0;
362 while ((he = hv_iternext (hv))) 507 while ((he = hv_iternext (hv)))
363 { 508 {
364 hes [i++] = he; 509 hes [i++] = he;
… …
386 531
387 FREETMPS; 532 FREETMPS;
388 LEAVE; 533 LEAVE;
389 } 534 }
390 535
391 for (i = 0; i < count; ++i) 536 encode_nl (enc); ++enc->indent;
537
538 while (count--)
392 { 539 {
393 encode_indent (enc); 540 encode_indent (enc);
541 he = hes [count];
394 encode_he (enc, hes [i]); 542 encode_hk (enc, he);
543 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
395 544
396 if (i < count - 1) 545 if (count)
397 encode_comma (enc); 546 encode_comma (enc);
398 } 547 }
399 548
549 encode_nl (enc); --enc->indent; encode_indent (enc);
550 }
551 }
552 else
553 {
554 if (hv_iterinit (hv) || SvMAGICAL (hv))
555 if ((he = hv_iternext (hv)))
556 {
557 encode_nl (enc); ++enc->indent;
558
559 for (;;)
560 {
400 encode_nl (enc); 561 encode_indent (enc);
562 encode_hk (enc, he);
563 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
564
565 if (!(he = hv_iternext (hv)))
566 break;
567
568 encode_comma (enc);
569 }
570
571 encode_nl (enc); --enc->indent; encode_indent (enc);
572 }
573 }
574
575 encode_ch (enc, '}');
576}
577
578// encode objects, arrays and special \0=false and \1=true values.
579static void
580encode_rv (enc_t *enc, SV *sv)
581{
582 svtype svt;
583
584 SvGETMAGIC (sv);
585 svt = SvTYPE (sv);
586
587 if (expect_false (SvOBJECT (sv)))
588 {
589 HV *stash = !JSON_SLOW || json_boolean_stash
590 ? json_boolean_stash
591 : gv_stashpv ("JSON::XS::Boolean", 1);
592
593 if (SvSTASH (sv) == stash)
594 {
595 if (SvIV (sv))
596 encode_str (enc, "true", 4, 0);
597 else
598 encode_str (enc, "false", 5, 0);
401 } 599 }
402 else 600 else
403 { 601 {
404 SV *sv; 602#if 0
405 HE *he = hv_iternext (hv); 603 if (0 && sv_derived_from (rv, "JSON::Literal"))
406
407 for (;;)
408 { 604 {
409 encode_indent (enc); 605 // not yet
410 encode_he (enc, he);
411
412 if (!(he = hv_iternext (hv)))
413 break;
414
415 encode_comma (enc);
416 } 606 }
607#endif
608 if (enc->json.flags & F_CONV_BLESSED)
609 {
610 // we re-bless the reference to get overload and other niceties right
611 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
417 612
613 if (to_json)
614 {
615 dSP;
616
617 ENTER; SAVETMPS; PUSHMARK (SP);
618 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
619
620 // calling with G_SCALAR ensures that we always get a 1 return value
621 PUTBACK;
622 call_sv ((SV *)GvCV (to_json), G_SCALAR);
623 SPAGAIN;
624
625 // catch this surprisingly common error
626 if (SvROK (TOPs) && SvRV (TOPs) == sv)
627 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
628
629 sv = POPs;
630 PUTBACK;
631
418 encode_nl (enc); 632 encode_sv (enc, sv);
633
634 FREETMPS; LEAVE;
635 }
636 else if (enc->json.flags & F_ALLOW_BLESSED)
637 encode_str (enc, "null", 4, 0);
638 else
639 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
640 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
641 }
642 else if (enc->json.flags & F_ALLOW_BLESSED)
643 encode_str (enc, "null", 4, 0);
644 else
645 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
646 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
419 } 647 }
420 } 648 }
649 else if (svt == SVt_PVHV)
650 encode_hv (enc, (HV *)sv);
651 else if (svt == SVt_PVAV)
652 encode_av (enc, (AV *)sv);
653 else if (svt < SVt_PVAV)
654 {
655 STRLEN len = 0;
656 char *pv = svt ? SvPV (sv, len) : 0;
421 657
422 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 658 if (len == 1 && *pv == '1')
659 encode_str (enc, "true", 4, 0);
660 else if (len == 1 && *pv == '0')
661 encode_str (enc, "false", 5, 0);
662 else if (enc->json.flags & F_ALLOW_UNKNOWN)
663 encode_str (enc, "null", 4, 0);
664 else
665 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
666 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
667 }
668 else if (enc->json.flags & F_ALLOW_UNKNOWN)
669 encode_str (enc, "null", 4, 0);
670 else
671 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
672 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
423} 673}
424 674
425static void 675static void
426encode_sv (enc_t *enc, SV *sv) 676encode_sv (enc_t *enc, SV *sv)
427{ 677{
… …
435 encode_str (enc, str, len, SvUTF8 (sv)); 685 encode_str (enc, str, len, SvUTF8 (sv));
436 encode_ch (enc, '"'); 686 encode_ch (enc, '"');
437 } 687 }
438 else if (SvNOKp (sv)) 688 else if (SvNOKp (sv))
439 { 689 {
690 // trust that perl will do the right thing w.r.t. JSON syntax.
440 need (enc, NV_DIG + 32); 691 need (enc, NV_DIG + 32);
441 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 692 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
442 enc->cur += strlen (enc->cur); 693 enc->cur += strlen (enc->cur);
443 } 694 }
444 else if (SvIOKp (sv)) 695 else if (SvIOKp (sv))
445 { 696 {
697 // we assume we can always read an IV as a UV and vice versa
698 // we assume two's complement
699 // we assume no aliasing issues in the union
700 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
701 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
702 {
703 // optimise the "small number case"
704 // code will likely be branchless and use only a single multiplication
705 // works for numbers up to 59074
706 I32 i = SvIVX (sv);
707 U32 u;
708 char digit, nz = 0;
709
446 need (enc, 64); 710 need (enc, 6);
711
712 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
713 u = i < 0 ? -i : i;
714
715 // convert to 4.28 fixed-point representation
716 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
717
718 // now output digit by digit, each time masking out the integer part
719 // and multiplying by 5 while moving the decimal point one to the right,
720 // resulting in a net multiplication by 10.
721 // we always write the digit to memory but conditionally increment
722 // the pointer, to enable the use of conditional move instructions.
723 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
724 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
725 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
726 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
727 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
728 }
729 else
730 {
731 // large integer, use the (rather slow) snprintf way.
732 need (enc, IVUV_MAXCHARS);
447 enc->cur += 733 enc->cur +=
448 SvIsUV(sv) 734 SvIsUV(sv)
449 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 735 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
450 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 736 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
737 }
451 } 738 }
452 else if (SvROK (sv)) 739 else if (SvROK (sv))
453 { 740 encode_rv (enc, SvRV (sv));
454 SV *rv = SvRV (sv); 741 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
455
456 if (enc->indent >= enc->maxdepth)
457 croak ("data structure too deep (hit recursion limit)");
458
459 switch (SvTYPE (rv))
460 {
461 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
462 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
463
464 default:
465 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
466 SvPV_nolen (sv));
467 }
468 }
469 else if (!SvOK (sv))
470 encode_str (enc, "null", 4, 0); 742 encode_str (enc, "null", 4, 0);
471 else 743 else
472 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 744 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
473 SvPV_nolen (sv), SvFLAGS (sv)); 745 SvPV_nolen (sv), SvFLAGS (sv));
474} 746}
475 747
476static SV * 748static SV *
477encode_json (SV *scalar, U32 flags) 749encode_json (SV *scalar, JSON *json)
478{ 750{
751 enc_t enc;
752
479 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 753 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
480 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 754 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
481 755
482 enc_t enc; 756 enc.json = *json;
483 enc.flags = flags;
484 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 757 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
485 enc.cur = SvPVX (enc.sv); 758 enc.cur = SvPVX (enc.sv);
486 enc.end = SvEND (enc.sv); 759 enc.end = SvEND (enc.sv);
487 enc.indent = 0; 760 enc.indent = 0;
488 enc.maxdepth = DEC_DEPTH (flags); 761 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
762 : enc.json.flags & F_LATIN1 ? 0x000100UL
763 : 0x110000UL;
489 764
490 SvPOK_only (enc.sv); 765 SvPOK_only (enc.sv);
491 encode_sv (&enc, scalar); 766 encode_sv (&enc, scalar);
767 encode_nl (&enc);
492 768
769 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
770 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
771
493 if (!(flags & (F_ASCII | F_UTF8))) 772 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
494 SvUTF8_on (enc.sv); 773 SvUTF8_on (enc.sv);
495 774
496 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
497
498 if (enc.flags & F_SHRINK) 775 if (enc.json.flags & F_SHRINK)
499 shrink (enc.sv); 776 shrink (enc.sv);
500 777
501 return enc.sv; 778 return enc.sv;
502} 779}
503 780
… …
508typedef struct 785typedef struct
509{ 786{
510 char *cur; // current parser pointer 787 char *cur; // current parser pointer
511 char *end; // end of input string 788 char *end; // end of input string
512 const char *err; // parse error, if != 0 789 const char *err; // parse error, if != 0
513 U32 flags; // F_* 790 JSON json;
514 U32 depth; // recursion depth 791 U32 depth; // recursion depth
515 U32 maxdepth; // recursion depth limit 792 U32 maxdepth; // recursion depth limit
516} dec_t; 793} dec_t;
517 794
518static void 795INLINE void
796decode_comment (dec_t *dec)
797{
798 // only '#'-style comments allowed a.t.m.
799
800 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
801 ++dec->cur;
802}
803
804INLINE void
519decode_ws (dec_t *dec) 805decode_ws (dec_t *dec)
520{ 806{
521 for (;;) 807 for (;;)
522 { 808 {
523 char ch = *dec->cur; 809 char ch = *dec->cur;
524 810
525 if (ch > 0x20 811 if (ch > 0x20)
812 {
813 if (expect_false (ch == '#'))
814 {
815 if (dec->json.flags & F_RELAXED)
816 decode_comment (dec);
817 else
818 break;
819 }
820 else
821 break;
822 }
526 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 823 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
527 break; 824 break; // parse error, but let higher level handle it, gives better error messages
528 825
529 ++dec->cur; 826 ++dec->cur;
530 } 827 }
531} 828}
532 829
… …
536 if (*dec->cur != ch) \ 833 if (*dec->cur != ch) \
537 ERR (# ch " expected"); \ 834 ERR (# ch " expected"); \
538 ++dec->cur; \ 835 ++dec->cur; \
539 SE 836 SE
540 837
541#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 838#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
542#define DEC_DEC_DEPTH --dec->depth 839#define DEC_DEC_DEPTH --dec->depth
543 840
544static SV *decode_sv (dec_t *dec); 841static SV *decode_sv (dec_t *dec);
545 842
546static signed char decode_hexdigit[256]; 843static signed char decode_hexdigit[256];
… …
549decode_4hex (dec_t *dec) 846decode_4hex (dec_t *dec)
550{ 847{
551 signed char d1, d2, d3, d4; 848 signed char d1, d2, d3, d4;
552 unsigned char *cur = (unsigned char *)dec->cur; 849 unsigned char *cur = (unsigned char *)dec->cur;
553 850
554 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 851 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
555 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 852 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
556 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 853 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
557 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 854 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
558 855
559 dec->cur += 4; 856 dec->cur += 4;
560 857
561 return ((UV)d1) << 12 858 return ((UV)d1) << 12
562 | ((UV)d2) << 8 859 | ((UV)d2) << 8
… …
570static SV * 867static SV *
571decode_str (dec_t *dec) 868decode_str (dec_t *dec)
572{ 869{
573 SV *sv = 0; 870 SV *sv = 0;
574 int utf8 = 0; 871 int utf8 = 0;
872 char *dec_cur = dec->cur;
575 873
576 do 874 do
577 { 875 {
578 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 876 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
579 char *cur = buf; 877 char *cur = buf;
580 878
581 do 879 do
582 { 880 {
583 unsigned char ch = *(unsigned char *)dec->cur++; 881 unsigned char ch = *(unsigned char *)dec_cur++;
584 882
585 if (ch == '"') 883 if (expect_false (ch == '"'))
586 { 884 {
587 --dec->cur; 885 --dec_cur;
588 break; 886 break;
589 } 887 }
590 else if (ch == '\\') 888 else if (expect_false (ch == '\\'))
591 { 889 {
592 switch (*dec->cur) 890 switch (*dec_cur)
593 { 891 {
594 case '\\': 892 case '\\':
595 case '/': 893 case '/':
596 case '"': *cur++ = *dec->cur++; break; 894 case '"': *cur++ = *dec_cur++; break;
597 895
598 case 'b': ++dec->cur; *cur++ = '\010'; break; 896 case 'b': ++dec_cur; *cur++ = '\010'; break;
599 case 't': ++dec->cur; *cur++ = '\011'; break; 897 case 't': ++dec_cur; *cur++ = '\011'; break;
600 case 'n': ++dec->cur; *cur++ = '\012'; break; 898 case 'n': ++dec_cur; *cur++ = '\012'; break;
601 case 'f': ++dec->cur; *cur++ = '\014'; break; 899 case 'f': ++dec_cur; *cur++ = '\014'; break;
602 case 'r': ++dec->cur; *cur++ = '\015'; break; 900 case 'r': ++dec_cur; *cur++ = '\015'; break;
603 901
604 case 'u': 902 case 'u':
605 { 903 {
606 UV lo, hi; 904 UV lo, hi;
607 ++dec->cur; 905 ++dec_cur;
608 906
907 dec->cur = dec_cur;
609 hi = decode_4hex (dec); 908 hi = decode_4hex (dec);
909 dec_cur = dec->cur;
610 if (hi == (UV)-1) 910 if (hi == (UV)-1)
611 goto fail; 911 goto fail;
612 912
613 // possibly a surrogate pair 913 // possibly a surrogate pair
614 if (hi >= 0xd800) 914 if (hi >= 0xd800)
615 if (hi < 0xdc00) 915 if (hi < 0xdc00)
616 { 916 {
617 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 917 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
618 ERR ("missing low surrogate character in surrogate pair"); 918 ERR ("missing low surrogate character in surrogate pair");
619 919
620 dec->cur += 2; 920 dec_cur += 2;
621 921
922 dec->cur = dec_cur;
622 lo = decode_4hex (dec); 923 lo = decode_4hex (dec);
924 dec_cur = dec->cur;
623 if (lo == (UV)-1) 925 if (lo == (UV)-1)
624 goto fail; 926 goto fail;
625 927
626 if (lo < 0xdc00 || lo >= 0xe000) 928 if (lo < 0xdc00 || lo >= 0xe000)
627 ERR ("surrogate pair expected"); 929 ERR ("surrogate pair expected");
… …
633 935
634 if (hi >= 0x80) 936 if (hi >= 0x80)
635 { 937 {
636 utf8 = 1; 938 utf8 = 1;
637 939
638 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 940 cur = encode_utf8 (cur, hi);
639 } 941 }
640 else 942 else
641 *cur++ = hi; 943 *cur++ = hi;
642 } 944 }
643 break; 945 break;
644 946
645 default: 947 default:
646 --dec->cur; 948 --dec_cur;
647 ERR ("illegal backslash escape sequence in string"); 949 ERR ("illegal backslash escape sequence in string");
648 } 950 }
649 } 951 }
650 else if (ch >= 0x20 && ch <= 0x7f) 952 else if (expect_true (ch >= 0x20 && ch < 0x80))
651 *cur++ = ch; 953 *cur++ = ch;
652 else if (ch >= 0x80) 954 else if (ch >= 0x80)
653 { 955 {
654 --dec->cur;
655
656 STRLEN clen; 956 STRLEN clen;
957
958 --dec_cur;
959
657 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 960 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
658 if (clen == (STRLEN)-1) 961 if (clen == (STRLEN)-1)
659 ERR ("malformed UTF-8 character in JSON string"); 962 ERR ("malformed UTF-8 character in JSON string");
660 963
661 do 964 do
662 {
663 *cur++ = *dec->cur++; 965 *cur++ = *dec_cur++;
664 }
665 while (--clen); 966 while (--clen);
666 967
667 utf8 = 1; 968 utf8 = 1;
668 } 969 }
669 else if (!ch)
670 ERR ("unexpected end of string while parsing json string");
671 else 970 else
971 {
972 --dec_cur;
973
974 if (!ch)
975 ERR ("unexpected end of string while parsing JSON string");
976 else
672 ERR ("invalid character encountered"); 977 ERR ("invalid character encountered while parsing JSON string");
673 978 }
674 } 979 }
675 while (cur < buf + SHORT_STRING_LEN); 980 while (cur < buf + SHORT_STRING_LEN);
676 981
982 {
677 STRLEN len = cur - buf; 983 STRLEN len = cur - buf;
678 984
679 if (sv) 985 if (sv)
680 { 986 {
681 SvGROW (sv, SvCUR (sv) + len + 1); 987 STRLEN cur = SvCUR (sv);
988
989 if (SvLEN (sv) <= cur + len)
990 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
991
682 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 992 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
683 SvCUR_set (sv, SvCUR (sv) + len); 993 SvCUR_set (sv, SvCUR (sv) + len);
684 } 994 }
685 else 995 else
686 sv = newSVpvn (buf, len); 996 sv = newSVpvn (buf, len);
687 } 997 }
998 }
688 while (*dec->cur != '"'); 999 while (*dec_cur != '"');
689 1000
690 ++dec->cur; 1001 ++dec_cur;
691 1002
692 if (sv) 1003 if (sv)
693 { 1004 {
694 SvPOK_only (sv); 1005 SvPOK_only (sv);
695 *SvEND (sv) = 0; 1006 *SvEND (sv) = 0;
… …
698 SvUTF8_on (sv); 1009 SvUTF8_on (sv);
699 } 1010 }
700 else 1011 else
701 sv = newSVpvn ("", 0); 1012 sv = newSVpvn ("", 0);
702 1013
1014 dec->cur = dec_cur;
703 return sv; 1015 return sv;
704 1016
705fail: 1017fail:
1018 dec->cur = dec_cur;
706 return 0; 1019 return 0;
707} 1020}
708 1021
709static SV * 1022static SV *
710decode_num (dec_t *dec) 1023decode_num (dec_t *dec)
… …
768 is_nv = 1; 1081 is_nv = 1;
769 } 1082 }
770 1083
771 if (!is_nv) 1084 if (!is_nv)
772 { 1085 {
773 UV uv; 1086 int len = dec->cur - start;
774 int numtype = grok_number (start, dec->cur - start, &uv); 1087
775 if (numtype & IS_NUMBER_IN_UV) 1088 // special case the rather common 1..5-digit-int case
776 if (numtype & IS_NUMBER_NEG) 1089 if (*start == '-')
1090 switch (len)
777 { 1091 {
778 if (uv < (UV)IV_MIN) 1092 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
779 return newSViv (-(IV)uv); 1093 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1094 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1095 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1096 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
780 } 1097 }
1098 else
1099 switch (len)
1100 {
1101 case 1: return newSViv ( start [0] - '0' * 1);
1102 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1103 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1104 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1105 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1106 }
1107
1108 {
1109 UV uv;
1110 int numtype = grok_number (start, len, &uv);
1111 if (numtype & IS_NUMBER_IN_UV)
1112 if (numtype & IS_NUMBER_NEG)
1113 {
1114 if (uv < (UV)IV_MIN)
1115 return newSViv (-(IV)uv);
1116 }
781 else 1117 else
782 return newSVuv (uv); 1118 return newSVuv (uv);
783 } 1119 }
784 1120
1121 len -= *start == '-' ? 1 : 0;
1122
1123 // does not fit into IV or UV, try NV
1124 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1125 #if defined (LDBL_DIG)
1126 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1127 #endif
1128 )
1129 // fits into NV without loss of precision
1130 return newSVnv (Atof (start));
1131
1132 // everything else fails, convert it to a string
1133 return newSVpvn (start, dec->cur - start);
1134 }
1135
1136 // loss of precision here
785 return newSVnv (Atof (start)); 1137 return newSVnv (Atof (start));
786 1138
787fail: 1139fail:
788 return 0; 1140 return 0;
789} 1141}
… …
819 1171
820 if (*dec->cur != ',') 1172 if (*dec->cur != ',')
821 ERR (", or ] expected while parsing array"); 1173 ERR (", or ] expected while parsing array");
822 1174
823 ++dec->cur; 1175 ++dec->cur;
1176
1177 decode_ws (dec);
1178
1179 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1180 {
1181 ++dec->cur;
1182 break;
1183 }
824 } 1184 }
825 1185
826 DEC_DEC_DEPTH; 1186 DEC_DEC_DEPTH;
827 return newRV_noinc ((SV *)av); 1187 return newRV_noinc ((SV *)av);
828 1188
… …
833} 1193}
834 1194
835static SV * 1195static SV *
836decode_hv (dec_t *dec) 1196decode_hv (dec_t *dec)
837{ 1197{
1198 SV *sv;
838 HV *hv = newHV (); 1199 HV *hv = newHV ();
839 1200
840 DEC_INC_DEPTH; 1201 DEC_INC_DEPTH;
841 decode_ws (dec); 1202 decode_ws (dec);
842 1203
843 if (*dec->cur == '}') 1204 if (*dec->cur == '}')
844 ++dec->cur; 1205 ++dec->cur;
845 else 1206 else
846 for (;;) 1207 for (;;)
847 { 1208 {
1209 EXPECT_CH ('"');
1210
1211 // heuristic: assume that
1212 // a) decode_str + hv_store_ent are abysmally slow.
1213 // b) most hash keys are short, simple ascii text.
1214 // => try to "fast-match" such strings to avoid
1215 // the overhead of decode_str + hv_store_ent.
1216 {
848 SV *key, *value; 1217 SV *value;
1218 char *p = dec->cur;
1219 char *e = p + 24; // only try up to 24 bytes
849 1220
850 decode_ws (dec); EXPECT_CH ('"'); 1221 for (;;)
851
852 key = decode_str (dec);
853 if (!key)
854 goto fail;
855
856 decode_ws (dec); EXPECT_CH (':');
857
858 value = decode_sv (dec);
859 if (!value)
860 { 1222 {
1223 // the >= 0x80 is false on most architectures
1224 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1225 {
1226 // slow path, back up and use decode_str
1227 SV *key = decode_str (dec);
1228 if (!key)
1229 goto fail;
1230
1231 decode_ws (dec); EXPECT_CH (':');
1232
1233 decode_ws (dec);
1234 value = decode_sv (dec);
1235 if (!value)
1236 {
1237 SvREFCNT_dec (key);
1238 goto fail;
1239 }
1240
1241 hv_store_ent (hv, key, value, 0);
861 SvREFCNT_dec (key); 1242 SvREFCNT_dec (key);
1243
1244 break;
1245 }
1246 else if (*p == '"')
1247 {
1248 // fast path, got a simple key
1249 char *key = dec->cur;
1250 int len = p - key;
1251 dec->cur = p + 1;
1252
1253 decode_ws (dec); EXPECT_CH (':');
1254
1255 decode_ws (dec);
1256 value = decode_sv (dec);
1257 if (!value)
862 goto fail; 1258 goto fail;
1259
1260 hv_store (hv, key, len, value, 0);
1261
1262 break;
1263 }
1264
1265 ++p;
863 } 1266 }
864 1267 }
865 hv_store_ent (hv, key, value, 0);
866 SvREFCNT_dec (key);
867 1268
868 decode_ws (dec); 1269 decode_ws (dec);
869 1270
870 if (*dec->cur == '}') 1271 if (*dec->cur == '}')
871 { 1272 {
… …
875 1276
876 if (*dec->cur != ',') 1277 if (*dec->cur != ',')
877 ERR (", or } expected while parsing object/hash"); 1278 ERR (", or } expected while parsing object/hash");
878 1279
879 ++dec->cur; 1280 ++dec->cur;
1281
1282 decode_ws (dec);
1283
1284 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1285 {
1286 ++dec->cur;
1287 break;
1288 }
880 } 1289 }
881 1290
882 DEC_DEC_DEPTH; 1291 DEC_DEC_DEPTH;
883 return newRV_noinc ((SV *)hv); 1292 sv = newRV_noinc ((SV *)hv);
1293
1294 // check filter callbacks
1295 if (dec->json.flags & F_HOOK)
1296 {
1297 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1298 {
1299 HE *cb, *he;
1300
1301 hv_iterinit (hv);
1302 he = hv_iternext (hv);
1303 hv_iterinit (hv);
1304
1305 // the next line creates a mortal sv each time its called.
1306 // might want to optimise this for common cases.
1307 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1308
1309 if (cb)
1310 {
1311 dSP;
1312 int count;
1313
1314 ENTER; SAVETMPS; PUSHMARK (SP);
1315 XPUSHs (HeVAL (he));
1316
1317 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1318
1319 if (count == 1)
1320 {
1321 sv = newSVsv (POPs);
1322 FREETMPS; LEAVE;
1323 return sv;
1324 }
1325
1326 FREETMPS; LEAVE;
1327 }
1328 }
1329
1330 if (dec->json.cb_object)
1331 {
1332 dSP;
1333 int count;
1334
1335 ENTER; SAVETMPS; PUSHMARK (SP);
1336 XPUSHs (sv_2mortal (sv));
1337
1338 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1339
1340 if (count == 1)
1341 {
1342 sv = newSVsv (POPs);
1343 FREETMPS; LEAVE;
1344 return sv;
1345 }
1346
1347 SvREFCNT_inc (sv);
1348 FREETMPS; LEAVE;
1349 }
1350 }
1351
1352 return sv;
884 1353
885fail: 1354fail:
886 SvREFCNT_dec (hv); 1355 SvREFCNT_dec (hv);
887 DEC_DEC_DEPTH; 1356 DEC_DEC_DEPTH;
888 return 0; 1357 return 0;
889} 1358}
890 1359
891static SV * 1360static SV *
892decode_sv (dec_t *dec) 1361decode_sv (dec_t *dec)
893{ 1362{
894 decode_ws (dec); 1363 // the beauty of JSON: you need exactly one character lookahead
1364 // to parse everything.
895 switch (*dec->cur) 1365 switch (*dec->cur)
896 { 1366 {
897 case '"': ++dec->cur; return decode_str (dec); 1367 case '"': ++dec->cur; return decode_str (dec);
898 case '[': ++dec->cur; return decode_av (dec); 1368 case '[': ++dec->cur; return decode_av (dec);
899 case '{': ++dec->cur; return decode_hv (dec); 1369 case '{': ++dec->cur; return decode_hv (dec);
900 1370
901 case '-': 1371 case '-':
902 case '0': case '1': case '2': case '3': case '4': 1372 case '0': case '1': case '2': case '3': case '4':
903 case '5': case '6': case '7': case '8': case '9': 1373 case '5': case '6': case '7': case '8': case '9':
904 return decode_num (dec); 1374 return decode_num (dec);
905 1375
906 case 't': 1376 case 't':
907 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1377 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
908 { 1378 {
909 dec->cur += 4; 1379 dec->cur += 4;
1380#if JSON_SLOW
1381 json_true = get_bool ("JSON::XS::true");
1382#endif
910 return newSViv (1); 1383 return newSVsv (json_true);
911 } 1384 }
912 else 1385 else
913 ERR ("'true' expected"); 1386 ERR ("'true' expected");
914 1387
915 break; 1388 break;
916 1389
917 case 'f': 1390 case 'f':
918 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1391 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
919 { 1392 {
920 dec->cur += 5; 1393 dec->cur += 5;
1394#if JSON_SLOW
1395 json_false = get_bool ("JSON::XS::false");
1396#endif
921 return newSViv (0); 1397 return newSVsv (json_false);
922 } 1398 }
923 else 1399 else
924 ERR ("'false' expected"); 1400 ERR ("'false' expected");
925 1401
926 break; 1402 break;
… …
935 ERR ("'null' expected"); 1411 ERR ("'null' expected");
936 1412
937 break; 1413 break;
938 1414
939 default: 1415 default:
940 ERR ("malformed json string, neither array, object, number, string or atom"); 1416 ERR ("malformed JSON string, neither array, object, number, string or atom");
941 break; 1417 break;
942 } 1418 }
943 1419
944fail: 1420fail:
945 return 0; 1421 return 0;
946} 1422}
947 1423
948static SV * 1424static SV *
949decode_json (SV *string, U32 flags) 1425decode_json (SV *string, JSON *json, char **offset_return)
950{ 1426{
1427 dec_t dec;
951 SV *sv; 1428 SV *sv;
952 1429
1430 /* work around bugs in 5.10 where manipulating magic values
1431 * will perl ignore the magic in subsequent accesses
1432 */
1433 /*SvGETMAGIC (string);*/
1434 if (SvMAGICAL (string))
1435 string = sv_2mortal (newSVsv (string));
1436
1437 SvUPGRADE (string, SVt_PV);
1438
1439 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1440 * assertion with -DDEBUGGING, although SvCUR is documented to
1441 * return the xpv_cur field which certainly exists after upgrading.
1442 * according to nicholas clark, calling SvPOK fixes this.
1443 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1444 * and hope for the best.
1445 * Damnit, SvPV_nolen still trips over yet another assertion. This
1446 * assertion business is seriously broken, try yet another workaround
1447 * for the broken -DDEBUGGING.
1448 */
1449 {
1450#ifdef DEBUGGING
1451 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1452#else
1453 STRLEN offset = SvCUR (string);
1454#endif
1455
1456 if (offset > json->max_size && json->max_size)
1457 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1458 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1459 }
1460
953 if (flags & F_UTF8) 1461 if (json->flags & F_UTF8)
954 sv_utf8_downgrade (string, 0); 1462 sv_utf8_downgrade (string, 0);
955 else 1463 else
956 sv_utf8_upgrade (string); 1464 sv_utf8_upgrade (string);
957 1465
958 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1466 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
959 1467
960 dec_t dec; 1468 dec.json = *json;
961 dec.flags = flags;
962 dec.cur = SvPVX (string); 1469 dec.cur = SvPVX (string);
963 dec.end = SvEND (string); 1470 dec.end = SvEND (string);
964 dec.err = 0; 1471 dec.err = 0;
965 dec.depth = 0; 1472 dec.depth = 0;
966 dec.maxdepth = DEC_DEPTH (dec.flags);
967 1473
1474 if (dec.json.cb_object || dec.json.cb_sk_object)
1475 dec.json.flags |= F_HOOK;
1476
968 *dec.end = 0; // this should basically be a nop, too, but make sure its there 1477 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1478
1479 decode_ws (&dec);
969 sv = decode_sv (&dec); 1480 sv = decode_sv (&dec);
970 1481
1482 if (offset_return)
1483 *offset_return = dec.cur;
1484
1485 if (!(offset_return || !sv))
1486 {
1487 // check for trailing garbage
1488 decode_ws (&dec);
1489
1490 if (*dec.cur)
1491 {
1492 dec.err = "garbage after JSON object";
1493 SvREFCNT_dec (sv);
1494 sv = 0;
1495 }
1496 }
1497
971 if (!sv) 1498 if (!sv)
972 { 1499 {
973 IV offset = dec.flags & F_UTF8
974 ? dec.cur - SvPVX (string)
975 : utf8_distance (dec.cur, SvPVX (string));
976 SV *uni = sv_newmortal (); 1500 SV *uni = sv_newmortal ();
977 1501
978 // horrible hack to silence warning inside pv_uni_display 1502 // horrible hack to silence warning inside pv_uni_display
979 COP cop = *PL_curcop; 1503 COP cop = *PL_curcop;
980 cop.cop_warnings = pWARN_NONE; 1504 cop.cop_warnings = pWARN_NONE;
… …
982 SAVEVPTR (PL_curcop); 1506 SAVEVPTR (PL_curcop);
983 PL_curcop = &cop; 1507 PL_curcop = &cop;
984 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1508 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
985 LEAVE; 1509 LEAVE;
986 1510
987 croak ("%s, at character offset %d (%s)", 1511 croak ("%s, at character offset %d (before \"%s\")",
988 dec.err, 1512 dec.err,
989 (int)offset, 1513 ptr_to_index (string, dec.cur),
990 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1514 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
991 } 1515 }
992 1516
993 sv = sv_2mortal (sv); 1517 sv = sv_2mortal (sv);
994 1518
995 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1519 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
996 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1520 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
997 1521
998 return sv; 1522 return sv;
1523}
1524
1525/////////////////////////////////////////////////////////////////////////////
1526// incremental parser
1527
1528static void
1529incr_parse (JSON *self)
1530{
1531 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1532
1533 // the state machine here is a bit convoluted and could be simplified a lot
1534 // but this would make it slower, so...
1535
1536 for (;;)
1537 {
1538 //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
1539 switch (self->incr_mode)
1540 {
1541 // only used for initial whitespace skipping
1542 case INCR_M_WS:
1543 for (;;)
1544 {
1545 if (*p > 0x20)
1546 {
1547 if (*p == '#')
1548 {
1549 self->incr_mode = INCR_M_C0;
1550 goto incr_m_c;
1551 }
1552 else
1553 {
1554 self->incr_mode = INCR_M_JSON;
1555 goto incr_m_json;
1556 }
1557 }
1558 else if (!*p)
1559 goto interrupt;
1560
1561 ++p;
1562 }
1563
1564 // skip a single char inside a string (for \\-processing)
1565 case INCR_M_BS:
1566 if (!*p)
1567 goto interrupt;
1568
1569 ++p;
1570 self->incr_mode = INCR_M_STR;
1571 goto incr_m_str;
1572
1573 // inside #-style comments
1574 case INCR_M_C0:
1575 case INCR_M_C1:
1576 incr_m_c:
1577 for (;;)
1578 {
1579 if (*p == '\n')
1580 {
1581 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1582 break;
1583 }
1584 else if (!*p)
1585 goto interrupt;
1586
1587 ++p;
1588 }
1589
1590 break;
1591
1592 // inside a string
1593 case INCR_M_STR:
1594 incr_m_str:
1595 for (;;)
1596 {
1597 if (*p == '"')
1598 {
1599 ++p;
1600 self->incr_mode = INCR_M_JSON;
1601
1602 if (!self->incr_nest)
1603 goto interrupt;
1604
1605 goto incr_m_json;
1606 }
1607 else if (*p == '\\')
1608 {
1609 ++p; // "virtually" consumes character after \
1610
1611 if (!*p) // if at end of string we have to switch modes
1612 {
1613 self->incr_mode = INCR_M_BS;
1614 goto interrupt;
1615 }
1616 }
1617 else if (!*p)
1618 goto interrupt;
1619
1620 ++p;
1621 }
1622
1623 // after initial ws, outside string
1624 case INCR_M_JSON:
1625 incr_m_json:
1626 for (;;)
1627 {
1628 switch (*p++)
1629 {
1630 case 0:
1631 --p;
1632 goto interrupt;
1633
1634 case 0x09:
1635 case 0x0a:
1636 case 0x0d:
1637 case 0x20:
1638 if (!self->incr_nest)
1639 {
1640 --p; // do not eat the whitespace, let the next round do it
1641 goto interrupt;
1642 }
1643 break;
1644
1645 case '"':
1646 self->incr_mode = INCR_M_STR;
1647 goto incr_m_str;
1648
1649 case '[':
1650 case '{':
1651 if (++self->incr_nest > self->max_depth)
1652 croak (ERR_NESTING_EXCEEDED);
1653 break;
1654
1655 case ']':
1656 case '}':
1657 if (--self->incr_nest <= 0)
1658 goto interrupt;
1659 break;
1660
1661 case '#':
1662 self->incr_mode = INCR_M_C1;
1663 goto incr_m_c;
1664 }
1665 }
1666 }
1667
1668 modechange:
1669 ;
1670 }
1671
1672interrupt:
1673 self->incr_pos = p - SvPVX (self->incr_text);
1674 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1675 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
999} 1676}
1000 1677
1001///////////////////////////////////////////////////////////////////////////// 1678/////////////////////////////////////////////////////////////////////////////
1002// XS interface functions 1679// XS interface functions
1003 1680
1004MODULE = JSON::XS PACKAGE = JSON::XS 1681MODULE = JSON::XS PACKAGE = JSON::XS
1005 1682
1006BOOT: 1683BOOT:
1007{ 1684{
1008 int i; 1685 int i;
1009
1010 memset (decode_hexdigit, 0xff, 256);
1011 1686
1012 for (i = 0; i < 256; ++i) 1687 for (i = 0; i < 256; ++i)
1013 decode_hexdigit [i] = 1688 decode_hexdigit [i] =
1014 i >= '0' && i <= '9' ? i - '0' 1689 i >= '0' && i <= '9' ? i - '0'
1015 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1690 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1016 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1691 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1017 : -1; 1692 : -1;
1018 1693
1019 json_stash = gv_stashpv ("JSON::XS", 1); 1694 json_stash = gv_stashpv ("JSON::XS" , 1);
1695 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1696
1697 json_true = get_bool ("JSON::XS::true");
1698 json_false = get_bool ("JSON::XS::false");
1699
1700 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1020} 1701}
1021 1702
1022PROTOTYPES: DISABLE 1703PROTOTYPES: DISABLE
1023 1704
1024SV *new (char *dummy) 1705void CLONE (...)
1025 CODE: 1706 CODE:
1026 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1707 json_stash = 0;
1708 json_boolean_stash = 0;
1709
1710void new (char *klass)
1711 PPCODE:
1712{
1713 SV *pv = NEWSV (0, sizeof (JSON));
1714 SvPOK_only (pv);
1715 json_init ((JSON *)SvPVX (pv));
1716 XPUSHs (sv_2mortal (sv_bless (
1717 newRV_noinc (pv),
1718 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1719 )));
1720}
1721
1722void ascii (JSON *self, int enable = 1)
1723 ALIAS:
1724 ascii = F_ASCII
1725 latin1 = F_LATIN1
1726 utf8 = F_UTF8
1727 indent = F_INDENT
1728 canonical = F_CANONICAL
1729 space_before = F_SPACE_BEFORE
1730 space_after = F_SPACE_AFTER
1731 pretty = F_PRETTY
1732 allow_nonref = F_ALLOW_NONREF
1733 shrink = F_SHRINK
1734 allow_blessed = F_ALLOW_BLESSED
1735 convert_blessed = F_CONV_BLESSED
1736 relaxed = F_RELAXED
1737 allow_unknown = F_ALLOW_UNKNOWN
1738 PPCODE:
1739{
1740 if (enable)
1741 self->flags |= ix;
1742 else
1743 self->flags &= ~ix;
1744
1745 XPUSHs (ST (0));
1746}
1747
1748void get_ascii (JSON *self)
1749 ALIAS:
1750 get_ascii = F_ASCII
1751 get_latin1 = F_LATIN1
1752 get_utf8 = F_UTF8
1753 get_indent = F_INDENT
1754 get_canonical = F_CANONICAL
1755 get_space_before = F_SPACE_BEFORE
1756 get_space_after = F_SPACE_AFTER
1757 get_allow_nonref = F_ALLOW_NONREF
1758 get_shrink = F_SHRINK
1759 get_allow_blessed = F_ALLOW_BLESSED
1760 get_convert_blessed = F_CONV_BLESSED
1761 get_relaxed = F_RELAXED
1762 get_allow_unknown = F_ALLOW_UNKNOWN
1763 PPCODE:
1764 XPUSHs (boolSV (self->flags & ix));
1765
1766void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1767 PPCODE:
1768 self->max_depth = max_depth;
1769 XPUSHs (ST (0));
1770
1771U32 get_max_depth (JSON *self)
1772 CODE:
1773 RETVAL = self->max_depth;
1027 OUTPUT: 1774 OUTPUT:
1028 RETVAL 1775 RETVAL
1029 1776
1030SV *ascii (SV *self, int enable = 1) 1777void max_size (JSON *self, U32 max_size = 0)
1031 ALIAS: 1778 PPCODE:
1032 ascii = F_ASCII 1779 self->max_size = max_size;
1033 utf8 = F_UTF8 1780 XPUSHs (ST (0));
1034 indent = F_INDENT 1781
1035 canonical = F_CANONICAL 1782int get_max_size (JSON *self)
1036 space_before = F_SPACE_BEFORE
1037 space_after = F_SPACE_AFTER
1038 pretty = F_PRETTY
1039 allow_nonref = F_ALLOW_NONREF
1040 shrink = F_SHRINK
1041 CODE: 1783 CODE:
1042{ 1784 RETVAL = self->max_size;
1043 UV *uv = SvJSON (self);
1044 if (enable)
1045 *uv |= ix;
1046 else
1047 *uv &= ~ix;
1048
1049 RETVAL = newSVsv (self);
1050}
1051 OUTPUT: 1785 OUTPUT:
1052 RETVAL 1786 RETVAL
1053 1787
1054SV *max_depth (SV *self, int max_depth = 0x80000000UL) 1788void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1789 PPCODE:
1790{
1791 SvREFCNT_dec (self->cb_object);
1792 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1793
1794 XPUSHs (ST (0));
1795}
1796
1797void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1798 PPCODE:
1799{
1800 if (!self->cb_sk_object)
1801 self->cb_sk_object = newHV ();
1802
1803 if (SvOK (cb))
1804 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1805 else
1806 {
1807 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1808
1809 if (!HvKEYS (self->cb_sk_object))
1810 {
1811 SvREFCNT_dec (self->cb_sk_object);
1812 self->cb_sk_object = 0;
1813 }
1814 }
1815
1816 XPUSHs (ST (0));
1817}
1818
1819void encode (JSON *self, SV *scalar)
1820 PPCODE:
1821 XPUSHs (encode_json (scalar, self));
1822
1823void decode (JSON *self, SV *jsonstr)
1824 PPCODE:
1825 XPUSHs (decode_json (jsonstr, self, 0));
1826
1827void decode_prefix (JSON *self, SV *jsonstr)
1828 PPCODE:
1829{
1830 char *offset;
1831 EXTEND (SP, 2);
1832 PUSHs (decode_json (jsonstr, self, &offset));
1833 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1834}
1835
1836void incr_parse (JSON *self, SV *jsonstr = 0)
1837 PPCODE:
1838{
1839 if (!self->incr_text)
1840 self->incr_text = newSVpvn ("", 0);
1841
1842 // append data, if any
1843 if (jsonstr)
1844 {
1845 if (SvUTF8 (jsonstr))
1846 {
1847 if (!SvUTF8 (self->incr_text))
1848 {
1849 /* utf-8-ness differs, need to upgrade */
1850 sv_utf8_upgrade (self->incr_text);
1851
1852 if (self->incr_pos)
1853 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1854 - (U8 *)SvPVX (self->incr_text);
1855 }
1856 }
1857 else if (SvUTF8 (self->incr_text))
1858 sv_utf8_upgrade (jsonstr);
1859
1860 {
1861 STRLEN len;
1862 const char *str = SvPV (jsonstr, len);
1863 STRLEN cur = SvCUR (self->incr_text);
1864
1865 if (SvLEN (self->incr_text) <= cur + len)
1866 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1867
1868 Move (str, SvEND (self->incr_text), len, char);
1869 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1870 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1871 }
1872 }
1873
1874 if (GIMME_V != G_VOID)
1875 do
1876 {
1877 char *offset;
1878
1879 if (!INCR_DONE (self))
1880 {
1881 incr_parse (self);
1882
1883 if (self->incr_pos > self->max_size && self->max_size)
1884 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1885 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1886
1887 if (!INCR_DONE (self))
1888 break;
1889 }
1890
1891 XPUSHs (decode_json (self->incr_text, self, &offset));
1892
1893 self->incr_pos -= offset - SvPVX (self->incr_text);
1894 self->incr_nest = 0;
1895 self->incr_mode = 0;
1896
1897 sv_chop (self->incr_text, offset);
1898 }
1899 while (GIMME_V == G_ARRAY);
1900}
1901
1902SV *incr_text (JSON *self)
1903 ATTRS: lvalue
1055 CODE: 1904 CODE:
1056{ 1905{
1057 UV *uv = SvJSON (self); 1906 if (self->incr_pos)
1058 UV log2 = 0; 1907 croak ("incr_text can not be called when the incremental parser already started parsing");
1059 1908
1060 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 1909 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1061
1062 while ((1UL << log2) < max_depth)
1063 ++log2;
1064
1065 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1066
1067 RETVAL = newSVsv (self);
1068} 1910}
1069 OUTPUT: 1911 OUTPUT:
1070 RETVAL 1912 RETVAL
1071 1913
1914void incr_skip (JSON *self)
1915 CODE:
1916{
1917 if (self->incr_pos)
1918 {
1919 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1920 self->incr_pos = 0;
1921 self->incr_nest = 0;
1922 self->incr_mode = 0;
1923 }
1924}
1925
1926void incr_reset (JSON *self)
1927 CODE:
1928{
1929 SvREFCNT_dec (self->incr_text);
1930 self->incr_text = 0;
1931 self->incr_pos = 0;
1932 self->incr_nest = 0;
1933 self->incr_mode = 0;
1934}
1935
1936void DESTROY (JSON *self)
1937 CODE:
1938 SvREFCNT_dec (self->cb_sk_object);
1939 SvREFCNT_dec (self->cb_object);
1940 SvREFCNT_dec (self->incr_text);
1941
1942PROTOTYPES: ENABLE
1943
1072void encode (SV *self, SV *scalar) 1944void encode_json (SV *scalar)
1945 ALIAS:
1946 to_json_ = 0
1947 encode_json = F_UTF8
1073 PPCODE: 1948 PPCODE:
1949{
1950 JSON json;
1951 json_init (&json);
1952 json.flags |= ix;
1074 XPUSHs (encode_json (scalar, *SvJSON (self))); 1953 XPUSHs (encode_json (scalar, &json));
1954}
1075 1955
1076void decode (SV *self, SV *jsonstr) 1956void decode_json (SV *jsonstr)
1957 ALIAS:
1958 from_json_ = 0
1959 decode_json = F_UTF8
1077 PPCODE: 1960 PPCODE:
1961{
1962 JSON json;
1963 json_init (&json);
1964 json.flags |= ix;
1078 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1965 XPUSHs (decode_json (jsonstr, &json, 0));
1966}
1079 1967
1080PROTOTYPES: ENABLE
1081 1968
1082void to_json (SV *scalar)
1083 ALIAS:
1084 objToJson = 0
1085 PPCODE:
1086 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1087
1088void from_json (SV *jsonstr)
1089 ALIAS:
1090 jsonToObj = 0
1091 PPCODE:
1092 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8));
1093

Diff Legend

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