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 1.5 by root, Thu Mar 22 23:24:18 2007 UTC vs.
Revision 1.119 by root, Mon Oct 28 23:19:54 2013 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 0x00000001 25#define F_ASCII 0x00000001UL
26#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 27#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 28#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 29#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 30#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 31#define F_SPACE_AFTER 0x00000040UL
15#define F_JSON_RPC 0x00000040
16#define F_ALLOW_NONREF 0x00000080 32#define F_ALLOW_NONREF 0x00000100UL
33#define F_SHRINK 0x00000200UL
34#define F_ALLOW_BLESSED 0x00000400UL
35#define F_CONV_BLESSED 0x00000800UL
36#define F_RELAXED 0x00001000UL
37#define F_ALLOW_UNKNOWN 0x00002000UL
38#define F_ALLOW_TAGS 0x00004000UL
39#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
17 40
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 41#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0
20 42
21#define INIT_SIZE 32 // initial scalar size to be allocated 43#define INIT_SIZE 32 // initial scalar size to be allocated
44#define INDENT_STEP 3 // spaces per indentation level
45
46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
47
48#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
22 49
23#define SB do { 50#define SB do {
24#define SE } while (0) 51#define SE } while (0)
25 52
26static HV *json_stash; 53#if __GNUC__ >= 3
54# define expect(expr,value) __builtin_expect ((expr), (value))
55# define INLINE static inline
56#else
57# define expect(expr,value) (expr)
58# define INLINE static
59#endif
60
61#define expect_false(expr) expect ((expr) != 0, 0)
62#define expect_true(expr) expect ((expr) != 0, 1)
63
64#define IN_RANGE_INC(type,val,beg,end) \
65 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
66 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
67
68#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
69
70#ifdef USE_ITHREADS
71# define JSON_SLOW 1
72# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
73#else
74# define JSON_SLOW 0
75# define JSON_STASH json_stash
76#endif
77
78// the amount of HEs to allocate on the stack, when sorting keys
79#define STACK_HES 64
80
81static HV *json_stash, *types_boolean_stash; // JSON::XS::
82static SV *types_true, *types_false, *sv_json;
83
84enum {
85 INCR_M_WS = 0, // initial whitespace skipping, must be 0
86 INCR_M_STR, // inside string
87 INCR_M_BS, // inside backslash
88 INCR_M_C0, // inside comment in initial whitespace sequence
89 INCR_M_C1, // inside comment in other places
90 INCR_M_JSON // outside anything, count nesting
91};
92
93#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
94
95typedef struct {
96 U32 flags;
97 U32 max_depth;
98 STRLEN max_size;
99
100 SV *cb_object;
101 HV *cb_sk_object;
102
103 // for the incremental parser
104 SV *incr_text; // the source text so far
105 STRLEN incr_pos; // the current offset into the text
106 int incr_nest; // {[]}-nesting level
107 unsigned char incr_mode;
108} JSON;
109
110INLINE void
111json_init (JSON *json)
112{
113 Zero (json, 1, JSON);
114 json->max_depth = 512;
115}
116
117/////////////////////////////////////////////////////////////////////////////
118// utility functions
119
120INLINE SV *
121get_bool (const char *name)
122{
123 SV *sv = get_sv (name, 1);
124
125 SvREADONLY_on (sv);
126 SvREADONLY_on (SvRV (sv));
127
128 return sv;
129}
130
131INLINE void
132shrink (SV *sv)
133{
134 sv_utf8_downgrade (sv, 1);
135
136 if (SvLEN (sv) > SvCUR (sv) + 1)
137 {
138#ifdef SvPV_shrink_to_cur
139 SvPV_shrink_to_cur (sv);
140#elif defined (SvPV_renew)
141 SvPV_renew (sv, SvCUR (sv) + 1);
142#endif
143 }
144}
145
146// decode an utf-8 character and return it, or (UV)-1 in
147// case of an error.
148// we special-case "safe" characters from U+80 .. U+7FF,
149// but use the very good perl function to parse anything else.
150// note that we never call this function for a ascii codepoints
151INLINE UV
152decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
153{
154 if (expect_true (len >= 2
155 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
156 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
157 {
158 *clen = 2;
159 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
160 }
161 else
162 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
163}
164
165// likewise for encoding, also never called for ascii codepoints
166// this function takes advantage of this fact, although current gccs
167// seem to optimise the check for >= 0x80 away anyways
168INLINE unsigned char *
169encode_utf8 (unsigned char *s, UV ch)
170{
171 if (expect_false (ch < 0x000080))
172 *s++ = ch;
173 else if (expect_true (ch < 0x000800))
174 *s++ = 0xc0 | ( ch >> 6),
175 *s++ = 0x80 | ( ch & 0x3f);
176 else if ( ch < 0x010000)
177 *s++ = 0xe0 | ( ch >> 12),
178 *s++ = 0x80 | ((ch >> 6) & 0x3f),
179 *s++ = 0x80 | ( ch & 0x3f);
180 else if ( ch < 0x110000)
181 *s++ = 0xf0 | ( ch >> 18),
182 *s++ = 0x80 | ((ch >> 12) & 0x3f),
183 *s++ = 0x80 | ((ch >> 6) & 0x3f),
184 *s++ = 0x80 | ( ch & 0x3f);
185
186 return s;
187}
188
189// convert offset pointer to character index, sv must be string
190static STRLEN
191ptr_to_index (SV *sv, char *offset)
192{
193 return SvUTF8 (sv)
194 ? utf8_distance (offset, SvPVX (sv))
195 : offset - SvPVX (sv);
196}
197
198/////////////////////////////////////////////////////////////////////////////
199// fp hell
200
201// scan a group of digits, and a trailing exponent
202static void
203json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
204{
205 UV uaccum = 0;
206 int eaccum = 0;
207
208 // if we recurse too deep, skip all remaining digits
209 // to avoid a stack overflow attack
210 if (expect_false (--maxdepth <= 0))
211 while (((U8)*s - '0') < 10)
212 ++s;
213
214 for (;;)
215 {
216 U8 dig = (U8)*s - '0';
217
218 if (expect_false (dig >= 10))
219 {
220 if (dig == (U8)((U8)'.' - (U8)'0'))
221 {
222 ++s;
223 json_atof_scan1 (s, accum, expo, 1, maxdepth);
224 }
225 else if ((dig | ' ') == 'e' - '0')
226 {
227 int exp2 = 0;
228 int neg = 0;
229
230 ++s;
231
232 if (*s == '-')
233 {
234 ++s;
235 neg = 1;
236 }
237 else if (*s == '+')
238 ++s;
239
240 while ((dig = (U8)*s - '0') < 10)
241 exp2 = exp2 * 10 + *s++ - '0';
242
243 *expo += neg ? -exp2 : exp2;
244 }
245
246 break;
247 }
248
249 ++s;
250
251 uaccum = uaccum * 10 + dig;
252 ++eaccum;
253
254 // if we have too many digits, then recurse for more
255 // we actually do this for rather few digits
256 if (uaccum >= (UV_MAX - 9) / 10)
257 {
258 if (postdp) *expo -= eaccum;
259 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
260 if (postdp) *expo += eaccum;
261
262 break;
263 }
264 }
265
266 // this relies greatly on the quality of the pow ()
267 // implementation of the platform, but a good
268 // implementation is hard to beat.
269 // (IEEE 754 conformant ones are required to be exact)
270 if (postdp) *expo -= eaccum;
271 *accum += uaccum * Perl_pow (10., *expo);
272 *expo += eaccum;
273}
274
275static NV
276json_atof (const char *s)
277{
278 NV accum = 0.;
279 int expo = 0;
280 int neg = 0;
281
282 if (*s == '-')
283 {
284 ++s;
285 neg = 1;
286 }
287
288 // a recursion depth of ten gives us >>500 bits
289 json_atof_scan1 (s, &accum, &expo, 0, 10);
290
291 return neg ? -accum : accum;
292}
293/////////////////////////////////////////////////////////////////////////////
294// encoder
27 295
28// structure used for encoding JSON 296// structure used for encoding JSON
29typedef struct 297typedef struct
30{ 298{
31 char *cur; 299 char *cur; // SvPVX (sv) + current output position
32 STRLEN len; // SvLEN (sv)
33 char *end; // SvEND (sv) 300 char *end; // SvEND (sv)
34 SV *sv; 301 SV *sv; // result scalar
35 UV flags; 302 JSON json;
36 int max_recurse; 303 U32 indent; // indentation level
37 int indent; 304 UV limit; // escape character values >= this value when encoding
38} enc_t; 305} enc_t;
39 306
40// structure used for decoding JSON 307INLINE void
41typedef struct
42{
43 char *cur;
44 char *end;
45 const char *err;
46 UV flags;
47} dec_t;
48
49static UV *
50SvJSON (SV *sv)
51{
52 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash))
53 croak ("object is not of type JSON::XS");
54
55 return &SvUVX (SvRV (sv));
56}
57
58/////////////////////////////////////////////////////////////////////////////
59
60static void
61need (enc_t *enc, STRLEN len) 308need (enc_t *enc, STRLEN len)
62{ 309{
63 if (enc->cur + len >= enc->end) 310 if (expect_false (enc->cur + len >= enc->end))
64 { 311 {
65 STRLEN cur = enc->cur - SvPVX (enc->sv); 312 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
66 SvGROW (enc->sv, cur + len + 1); 313 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
67 enc->cur = SvPVX (enc->sv) + cur; 314 enc->cur = SvPVX (enc->sv) + cur;
68 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 315 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
69 } 316 }
70} 317}
71 318
72static void 319INLINE void
73encode_ch (enc_t *enc, char ch) 320encode_ch (enc_t *enc, char ch)
74{ 321{
75 need (enc, 1); 322 need (enc, 1);
76 *enc->cur++ = ch; 323 *enc->cur++ = ch;
77} 324}
… …
85 332
86 while (str < end) 333 while (str < end)
87 { 334 {
88 unsigned char ch = *(unsigned char *)str; 335 unsigned char ch = *(unsigned char *)str;
89 336
90 if (ch == '"') 337 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
91 { 338 {
339 if (expect_false (ch == '"')) // but with slow exceptions
340 {
92 need (enc, len += 1); 341 need (enc, len += 1);
93 *enc->cur++ = '\\'; 342 *enc->cur++ = '\\';
94 *enc->cur++ = '"'; 343 *enc->cur++ = '"';
95 ++str;
96 } 344 }
97 else if (ch == '\\') 345 else if (expect_false (ch == '\\'))
98 { 346 {
99 need (enc, len += 1); 347 need (enc, len += 1);
100 *enc->cur++ = '\\'; 348 *enc->cur++ = '\\';
101 *enc->cur++ = '\\'; 349 *enc->cur++ = '\\';
102 ++str;
103 } 350 }
104 else if (ch >= 0x20 && ch < 0x80) // most common case 351 else
105 {
106 *enc->cur++ = ch; 352 *enc->cur++ = ch;
107 ++str; 353
108 }
109 else if (ch == '\015')
110 {
111 need (enc, len += 1);
112 *enc->cur++ = '\\';
113 *enc->cur++ = 'r';
114 ++str;
115 }
116 else if (ch == '\012')
117 {
118 need (enc, len += 1);
119 *enc->cur++ = '\\';
120 *enc->cur++ = 'n';
121 ++str; 354 ++str;
122 } 355 }
123 else 356 else
124 { 357 {
125 STRLEN clen; 358 switch (ch)
126 UV uch;
127
128 if (is_utf8)
129 {
130 uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
131 if (clen == (STRLEN)-1)
132 croak ("malformed UTF-8 character in string, cannot convert to JSON");
133 } 359 {
134 else 360 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
135 { 361 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
136 uch = ch; 362 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
137 clen = 1; 363 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
138 } 364 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
139 365
140 if (uch < 0x80 || enc->flags & F_ASCII) 366 default:
141 {
142 if (uch > 0xFFFFUL)
143 { 367 {
368 STRLEN clen;
369 UV uch;
370
371 if (is_utf8)
372 {
373 uch = decode_utf8 (str, end - str, &clen);
374 if (clen == (STRLEN)-1)
375 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
376 }
377 else
378 {
379 uch = ch;
380 clen = 1;
381 }
382
383 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
384 {
385 if (uch >= 0x10000UL)
386 {
387 if (uch >= 0x110000UL)
388 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
389
144 need (enc, len += 11); 390 need (enc, len += 11);
145 sprintf (enc->cur, "\\u%04x\\u%04x", 391 sprintf (enc->cur, "\\u%04x\\u%04x",
146 (uch - 0x10000) / 0x400 + 0xD800, 392 (int)((uch - 0x10000) / 0x400 + 0xD800),
147 (uch - 0x10000) % 0x400 + 0xDC00); 393 (int)((uch - 0x10000) % 0x400 + 0xDC00));
148 enc->cur += 12; 394 enc->cur += 12;
395 }
396 else
397 {
398 need (enc, len += 5);
399 *enc->cur++ = '\\';
400 *enc->cur++ = 'u';
401 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
402 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
403 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
404 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
405 }
406
407 str += clen;
408 }
409 else if (enc->json.flags & F_LATIN1)
410 {
411 *enc->cur++ = uch;
412 str += clen;
413 }
414 else if (is_utf8)
415 {
416 need (enc, len += clen);
417 do
418 {
419 *enc->cur++ = *str++;
420 }
421 while (--clen);
422 }
423 else
424 {
425 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
426 enc->cur = encode_utf8 (enc->cur, uch);
427 ++str;
428 }
149 } 429 }
150 else
151 {
152 static char hexdigit [16] = "0123456789abcdef";
153 need (enc, len += 5);
154 *enc->cur++ = '\\';
155 *enc->cur++ = 'u';
156 *enc->cur++ = hexdigit [ uch >> 12 ];
157 *enc->cur++ = hexdigit [(uch >> 8) & 15];
158 *enc->cur++ = hexdigit [(uch >> 4) & 15];
159 *enc->cur++ = hexdigit [(uch >> 0) & 15];
160 }
161
162 str += clen;
163 }
164 else if (is_utf8)
165 {
166 need (enc, len += clen);
167 do
168 {
169 *enc->cur++ = *str++;
170 }
171 while (--clen);
172 }
173 else
174 {
175 need (enc, 10); // never more than 11 bytes needed
176 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0);
177 ++str;
178 } 430 }
179 } 431 }
180 432
181 --len; 433 --len;
182 } 434 }
183} 435}
184 436
185#define INDENT SB \ 437INLINE void
438encode_indent (enc_t *enc)
439{
186 if (enc->flags & F_INDENT) \ 440 if (enc->json.flags & F_INDENT)
187 { \ 441 {
188 int i_; \ 442 int spaces = enc->indent * INDENT_STEP;
189 need (enc, enc->indent); \ 443
190 for (i_ = enc->indent * 3; i_--; )\ 444 need (enc, spaces);
445 memset (enc->cur, ' ', spaces);
446 enc->cur += spaces;
447 }
448}
449
450INLINE void
451encode_space (enc_t *enc)
452{
453 need (enc, 1);
454 encode_ch (enc, ' ');
455}
456
457INLINE void
458encode_nl (enc_t *enc)
459{
460 if (enc->json.flags & F_INDENT)
461 {
462 need (enc, 1);
191 encode_ch (enc, ' '); \ 463 encode_ch (enc, '\n');
192 } \ 464 }
193 SE 465}
194 466
195#define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE 467INLINE void
196#define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE 468encode_comma (enc_t *enc)
197#define COMMA SB \ 469{
198 encode_ch (enc, ','); \ 470 encode_ch (enc, ',');
471
199 if (enc->flags & F_INDENT) \ 472 if (enc->json.flags & F_INDENT)
200 NL; \ 473 encode_nl (enc);
201 else if (enc->flags & F_SPACE_AFTER) \ 474 else if (enc->json.flags & F_SPACE_AFTER)
202 SPACE; \ 475 encode_space (enc);
203 SE 476}
204 477
205static void encode_sv (enc_t *enc, SV *sv); 478static void encode_sv (enc_t *enc, SV *sv);
206 479
207static void 480static void
208encode_av (enc_t *enc, AV *av) 481encode_av (enc_t *enc, AV *av)
209{ 482{
210 int i, len = av_len (av); 483 int i, len = av_len (av);
211 484
485 if (enc->indent >= enc->json.max_depth)
486 croak (ERR_NESTING_EXCEEDED);
487
212 encode_ch (enc, '['); NL; 488 encode_ch (enc, '[');
213 ++enc->indent;
214 489
490 if (len >= 0)
491 {
492 encode_nl (enc); ++enc->indent;
493
215 for (i = 0; i <= len; ++i) 494 for (i = 0; i <= len; ++i)
216 { 495 {
217 INDENT; 496 SV **svp = av_fetch (av, i, 0);
218 encode_sv (enc, *av_fetch (av, i, 0));
219 497
498 encode_indent (enc);
499
500 if (svp)
501 encode_sv (enc, *svp);
502 else
503 encode_str (enc, "null", 4, 0);
504
220 if (i < len) 505 if (i < len)
221 COMMA; 506 encode_comma (enc);
222 } 507 }
223 508
224 NL; 509 encode_nl (enc); --enc->indent; encode_indent (enc);
510 }
225 511
226 --enc->indent;
227 INDENT; encode_ch (enc, ']'); 512 encode_ch (enc, ']');
228} 513}
229 514
230static void 515static void
231encode_he (enc_t *enc, HE *he) 516encode_hk (enc_t *enc, HE *he)
232{ 517{
233 encode_ch (enc, '"'); 518 encode_ch (enc, '"');
234 519
235 if (HeKLEN (he) == HEf_SVKEY) 520 if (HeKLEN (he) == HEf_SVKEY)
236 { 521 {
237 SV *sv = HeSVKEY (he); 522 SV *sv = HeSVKEY (he);
238 STRLEN len; 523 STRLEN len;
239 char *str; 524 char *str;
240 525
241 SvGETMAGIC (sv); 526 SvGETMAGIC (sv);
242 str = SvPV (sv, len); 527 str = SvPV (sv, len);
243 528
244 encode_str (enc, str, len, SvUTF8 (sv)); 529 encode_str (enc, str, len, SvUTF8 (sv));
245 } 530 }
246 else 531 else
247 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 532 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
248 533
249 encode_ch (enc, '"'); 534 encode_ch (enc, '"');
250 535
251 if (enc->flags & F_SPACE_BEFORE) SPACE; 536 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
252 encode_ch (enc, ':'); 537 encode_ch (enc, ':');
253 if (enc->flags & F_SPACE_AFTER ) SPACE; 538 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
254 encode_sv (enc, HeVAL (he));
255} 539}
256 540
257// compare hash entries, used when all keys are bytestrings 541// compare hash entries, used when all keys are bytestrings
258static int 542static int
259he_cmp_fast (const void *a_, const void *b_) 543he_cmp_fast (const void *a_, const void *b_)
… …
264 HE *b = *(HE **)b_; 548 HE *b = *(HE **)b_;
265 549
266 STRLEN la = HeKLEN (a); 550 STRLEN la = HeKLEN (a);
267 STRLEN lb = HeKLEN (b); 551 STRLEN lb = HeKLEN (b);
268 552
269 if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 553 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
270 cmp = la < lb ? -1 : la == lb ? 0 : 1; 554 cmp = lb - la;
271 555
272 return cmp; 556 return cmp;
273} 557}
274 558
275// compare hash entries, used when some keys are sv's or utf-x 559// compare hash entries, used when some keys are sv's or utf-x
276static int 560static int
277he_cmp_slow (const void *a, const void *b) 561he_cmp_slow (const void *a, const void *b)
278{ 562{
279 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 563 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
280} 564}
281 565
282static void 566static void
283encode_hv (enc_t *enc, HV *hv) 567encode_hv (enc_t *enc, HV *hv)
284{ 568{
285 int count, i; 569 HE *he;
286 570
287 encode_ch (enc, '{'); NL; ++enc->indent; 571 if (enc->indent >= enc->json.max_depth)
572 croak (ERR_NESTING_EXCEEDED);
288 573
289 if ((count = hv_iterinit (hv))) 574 encode_ch (enc, '{');
290 { 575
291 // for canonical output we have to sort by keys first 576 // for canonical output we have to sort by keys first
292 // actually, this is mostly due to the stupid so-called 577 // actually, this is mostly due to the stupid so-called
293 // security workaround added somewhere in 5.8.x. 578 // security workaround added somewhere in 5.8.x
294 // that randomises hash orderings 579 // that randomises hash orderings
295 if (enc->flags & F_CANONICAL) 580 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
581 {
582 int count = hv_iterinit (hv);
583
584 if (SvMAGICAL (hv))
296 { 585 {
297 HE *he, *hes [count]; 586 // need to count by iterating. could improve by dynamically building the vector below
587 // but I don't care for the speed of this special case.
588 // note also that we will run into undefined behaviour when the two iterations
589 // do not result in the same count, something I might care for in some later release.
590
591 count = 0;
592 while (hv_iternext (hv))
593 ++count;
594
595 hv_iterinit (hv);
596 }
597
598 if (count)
599 {
298 int fast = 1; 600 int i, fast = 1;
601 HE *hes_stack [STACK_HES];
602 HE **hes = hes_stack;
603
604 // allocate larger arrays on the heap
605 if (count > STACK_HES)
606 {
607 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
608 hes = (HE **)SvPVX (sv);
609 }
299 610
300 i = 0; 611 i = 0;
301 while ((he = hv_iternext (hv))) 612 while ((he = hv_iternext (hv)))
302 { 613 {
303 hes [i++] = he; 614 hes [i++] = he;
… …
309 620
310 if (fast) 621 if (fast)
311 qsort (hes, count, sizeof (HE *), he_cmp_fast); 622 qsort (hes, count, sizeof (HE *), he_cmp_fast);
312 else 623 else
313 { 624 {
314 // hack to disable "use bytes" 625 // hack to forcefully disable "use bytes"
315 COP *oldcop = PL_curcop, cop; 626 COP cop = *PL_curcop;
316 cop.op_private = 0; 627 cop.op_private = 0;
628
629 ENTER;
630 SAVETMPS;
631
632 SAVEVPTR (PL_curcop);
317 PL_curcop = &cop; 633 PL_curcop = &cop;
318 634
319 SAVETMPS;
320 qsort (hes, count, sizeof (HE *), he_cmp_slow); 635 qsort (hes, count, sizeof (HE *), he_cmp_slow);
636
321 FREETMPS; 637 FREETMPS;
322 638 LEAVE;
323 PL_curcop = oldcop;
324 }
325
326 for (i = 0; i < count; ++i)
327 { 639 }
328 INDENT;
329 encode_he (enc, hes [i]);
330 640
331 if (i < count - 1) 641 encode_nl (enc); ++enc->indent;
332 COMMA; 642
643 while (count--)
333 } 644 {
645 encode_indent (enc);
646 he = hes [count];
647 encode_hk (enc, he);
648 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
334 649
650 if (count)
651 encode_comma (enc);
335 NL; 652 }
653
654 encode_nl (enc); --enc->indent; encode_indent (enc);
336 } 655 }
656 }
657 else
658 {
659 if (hv_iterinit (hv) || SvMAGICAL (hv))
660 if ((he = hv_iternext (hv)))
661 {
662 encode_nl (enc); ++enc->indent;
663
664 for (;;)
665 {
666 encode_indent (enc);
667 encode_hk (enc, he);
668 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
669
670 if (!(he = hv_iternext (hv)))
671 break;
672
673 encode_comma (enc);
674 }
675
676 encode_nl (enc); --enc->indent; encode_indent (enc);
677 }
678 }
679
680 encode_ch (enc, '}');
681}
682
683// encode objects, arrays and special \0=false and \1=true values.
684static void
685encode_rv (enc_t *enc, SV *sv)
686{
687 svtype svt;
688
689 SvGETMAGIC (sv);
690 svt = SvTYPE (sv);
691
692 if (expect_false (SvOBJECT (sv)))
693 {
694 HV *boolean_stash = !JSON_SLOW || types_boolean_stash
695 ? types_boolean_stash
696 : gv_stashpv ("Types::Serialiser::Boolean", 1);
697 HV *stash = SvSTASH (sv);
698
699 if (stash == boolean_stash)
700 {
701 if (SvIV (sv))
702 encode_str (enc, "true", 4, 0);
703 else
704 encode_str (enc, "false", 5, 0);
705 }
706 else if (enc->json.flags & F_CONV_BLESSED)
707 {
708 GV *to_json = gv_fetchmethod_autoload (stash, "TO_JSON", 0);
709
710 if (to_json)
711 {
712 dSP;
713
714 ENTER; SAVETMPS; PUSHMARK (SP);
715 // we re-bless the reference to get overload and other niceties right
716 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
717
718 // calling with G_SCALAR ensures that we always get a 1 return value
719 PUTBACK;
720 call_sv ((SV *)GvCV (to_json), G_SCALAR);
721 SPAGAIN;
722
723 // catch this surprisingly common error
724 if (SvROK (TOPs) && SvRV (TOPs) == sv)
725 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
726
727 sv = POPs;
728 PUTBACK;
729
730 encode_sv (enc, sv);
731
732 FREETMPS; LEAVE;
733 }
734 else if (enc->json.flags & F_ALLOW_BLESSED)
735 encode_str (enc, "null", 4, 0);
736 else
737 croak ("encountered object '%s' when conv_object is enabled, but TO_JSON is missing and allow_blessed disabled",
738 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
739 }
740 else if (enc->json.flags & F_ALLOW_TAGS)
741 {
742 GV *method = gv_fetchmethod_autoload (stash, "FREEZE", 0);
743
744 if (method)
745 {
746 int count;
747 dSP;
748
749 ENTER; SAVETMPS; PUSHMARK (SP);
750 EXTEND (SP, 2);
751 // we re-bless the reference to get overload and other niceties right
752 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
753 PUSHs (sv_json);
754
755 PUTBACK;
756 count = call_sv ((SV *)GvCV (method), G_ARRAY);
757 SPAGAIN;
758
759 // catch this surprisingly common error
760 if (SvROK (TOPs) && SvRV (TOPs) == sv)
761 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
762
763 encode_ch (enc, '(');
764 encode_ch (enc, '"');
765 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
766 encode_ch (enc, '"');
767 encode_ch (enc, ')');
768 encode_ch (enc, '[');
769
770 while (count)
771 {
772 encode_sv (enc, SP[1 - count--]);
773
774 if (count)
775 encode_ch (enc, ',');
776 }
777
778 encode_ch (enc, ']');
779
780 PUTBACK;
781
782 FREETMPS; LEAVE;
783 }
784 else if (enc->json.flags & F_ALLOW_BLESSED)
785 encode_str (enc, "null", 4, 0);
786 else
787 croak ("encountered object '%s' when allow_tags is enabled, but FREEZE is missing and allow_blessed disabled",
788 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
789 }
790 else if (enc->json.flags & F_ALLOW_BLESSED)
791 encode_str (enc, "null", 4, 0);
337 else 792 else
338 { 793 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled",
339 SV *sv; 794 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
340 HE *he = hv_iternext (hv); 795 }
341 796 else if (svt == SVt_PVHV)
342 for (;;) 797 encode_hv (enc, (HV *)sv);
343 { 798 else if (svt == SVt_PVAV)
344 INDENT; 799 encode_av (enc, (AV *)sv);
345 encode_he (enc, he); 800 else if (svt < SVt_PVAV)
346
347 if (!(he = hv_iternext (hv)))
348 break;
349
350 COMMA;
351 }
352
353 NL;
354 }
355 } 801 {
802 STRLEN len = 0;
803 char *pv = svt ? SvPV (sv, len) : 0;
356 804
357 --enc->indent; INDENT; encode_ch (enc, '}'); 805 if (len == 1 && *pv == '1')
806 encode_str (enc, "true", 4, 0);
807 else if (len == 1 && *pv == '0')
808 encode_str (enc, "false", 5, 0);
809 else if (enc->json.flags & F_ALLOW_UNKNOWN)
810 encode_str (enc, "null", 4, 0);
811 else
812 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
813 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
814 }
815 else if (enc->json.flags & F_ALLOW_UNKNOWN)
816 encode_str (enc, "null", 4, 0);
817 else
818 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
819 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
358} 820}
359 821
360static void 822static void
361encode_sv (enc_t *enc, SV *sv) 823encode_sv (enc_t *enc, SV *sv)
362{ 824{
… …
370 encode_str (enc, str, len, SvUTF8 (sv)); 832 encode_str (enc, str, len, SvUTF8 (sv));
371 encode_ch (enc, '"'); 833 encode_ch (enc, '"');
372 } 834 }
373 else if (SvNOKp (sv)) 835 else if (SvNOKp (sv))
374 { 836 {
837 // trust that perl will do the right thing w.r.t. JSON syntax.
375 need (enc, NV_DIG + 32); 838 need (enc, NV_DIG + 32);
376 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 839 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
377 enc->cur += strlen (enc->cur); 840 enc->cur += strlen (enc->cur);
378 } 841 }
379 else if (SvIOKp (sv)) 842 else if (SvIOKp (sv))
380 { 843 {
844 // we assume we can always read an IV as a UV and vice versa
845 // we assume two's complement
846 // we assume no aliasing issues in the union
847 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
848 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
849 {
850 // optimise the "small number case"
851 // code will likely be branchless and use only a single multiplication
852 // works for numbers up to 59074
853 I32 i = SvIVX (sv);
854 U32 u;
855 char digit, nz = 0;
856
381 need (enc, 64); 857 need (enc, 6);
858
859 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
860 u = i < 0 ? -i : i;
861
862 // convert to 4.28 fixed-point representation
863 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
864
865 // now output digit by digit, each time masking out the integer part
866 // and multiplying by 5 while moving the decimal point one to the right,
867 // resulting in a net multiplication by 10.
868 // we always write the digit to memory but conditionally increment
869 // the pointer, to enable the use of conditional move instructions.
870 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
871 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
872 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
873 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
874 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
875 }
876 else
877 {
878 // large integer, use the (rather slow) snprintf way.
879 need (enc, IVUV_MAXCHARS);
382 enc->cur += 880 enc->cur +=
383 SvIsUV(sv) 881 SvIsUV(sv)
384 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 882 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
385 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 883 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
884 }
386 } 885 }
387 else if (SvROK (sv)) 886 else if (SvROK (sv))
388 { 887 encode_rv (enc, SvRV (sv));
389 if (!--enc->max_recurse) 888 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
390 croak ("data structure too deep (hit recursion limit)");
391
392 sv = SvRV (sv);
393
394 switch (SvTYPE (sv))
395 {
396 case SVt_PVAV: encode_av (enc, (AV *)sv); break;
397 case SVt_PVHV: encode_hv (enc, (HV *)sv); break;
398
399 default:
400 croak ("JSON can only represent references to arrays or hashes");
401 }
402 }
403 else if (!SvOK (sv))
404 encode_str (enc, "null", 4, 0); 889 encode_str (enc, "null", 4, 0);
405 else 890 else
406 croak ("encountered perl type that JSON cannot handle"); 891 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
892 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
407} 893}
408 894
409static SV * 895static SV *
410encode_json (SV *scalar, UV flags) 896encode_json (SV *scalar, JSON *json)
411{ 897{
412 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar))
413 croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)");
414
415 enc_t enc; 898 enc_t enc;
416 enc.flags = flags; 899
900 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
901 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
902
903 enc.json = *json;
417 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 904 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
418 enc.cur = SvPVX (enc.sv); 905 enc.cur = SvPVX (enc.sv);
419 enc.end = SvEND (enc.sv); 906 enc.end = SvEND (enc.sv);
420 enc.max_recurse = 0;
421 enc.indent = 0; 907 enc.indent = 0;
908 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
909 : enc.json.flags & F_LATIN1 ? 0x000100UL
910 : 0x110000UL;
422 911
423 SvPOK_only (enc.sv); 912 SvPOK_only (enc.sv);
424 encode_sv (&enc, scalar); 913 encode_sv (&enc, scalar);
914 encode_nl (&enc);
425 915
916 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
917 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
918
426 if (!(flags & (F_ASCII | F_UTF8))) 919 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
427 SvUTF8_on (enc.sv); 920 SvUTF8_on (enc.sv);
428 921
429 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 922 if (enc.json.flags & F_SHRINK)
923 shrink (enc.sv);
924
430 return enc.sv; 925 return enc.sv;
431} 926}
432 927
433///////////////////////////////////////////////////////////////////////////// 928/////////////////////////////////////////////////////////////////////////////
929// decoder
434 930
435#define WS \ 931// structure used for decoding JSON
932typedef struct
933{
934 char *cur; // current parser pointer
935 char *end; // end of input string
936 const char *err; // parse error, if != 0
937 JSON json;
938 U32 depth; // recursion depth
939 U32 maxdepth; // recursion depth limit
940} dec_t;
941
942INLINE void
943decode_comment (dec_t *dec)
944{
945 // only '#'-style comments allowed a.t.m.
946
947 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
948 ++dec->cur;
949}
950
951INLINE void
952decode_ws (dec_t *dec)
953{
436 for (;;) \ 954 for (;;)
437 { \ 955 {
438 char ch = *dec->cur; \ 956 char ch = *dec->cur;
957
439 if (ch > 0x20 \ 958 if (ch > 0x20)
959 {
960 if (expect_false (ch == '#'))
961 {
962 if (dec->json.flags & F_RELAXED)
963 decode_comment (dec);
964 else
965 break;
966 }
967 else
968 break;
969 }
440 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ 970 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
441 break; \ 971 break; // parse error, but let higher level handle it, gives better error messages
972
442 ++dec->cur; \ 973 ++dec->cur;
443 } 974 }
975}
444 976
445#define ERR(reason) SB dec->err = reason; goto fail; SE 977#define ERR(reason) SB dec->err = reason; goto fail; SE
978
446#define EXPECT_CH(ch) SB \ 979#define EXPECT_CH(ch) SB \
447 if (*dec->cur != ch) \ 980 if (*dec->cur != ch) \
448 ERR (# ch " expected"); \ 981 ERR (# ch " expected"); \
449 ++dec->cur; \ 982 ++dec->cur; \
450 SE 983 SE
451 984
985#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
986#define DEC_DEC_DEPTH --dec->depth
987
452static SV *decode_sv (dec_t *dec); 988static SV *decode_sv (dec_t *dec);
453 989
454static signed char decode_hexdigit[256]; 990static signed char decode_hexdigit[256];
455 991
456static UV 992static UV
457decode_4hex (dec_t *dec) 993decode_4hex (dec_t *dec)
458{ 994{
459 signed char d1, d2, d3, d4; 995 signed char d1, d2, d3, d4;
996 unsigned char *cur = (unsigned char *)dec->cur;
460 997
461 d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; 998 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
462 if (d1 < 0) ERR ("four hexadecimal digits expected"); 999 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
463 d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; 1000 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
464 if (d2 < 0) ERR ("four hexadecimal digits expected"); 1001 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
465 d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]];
466 if (d3 < 0) ERR ("four hexadecimal digits expected");
467 d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]];
468 if (d4 < 0) ERR ("four hexadecimal digits expected");
469 1002
470 dec->cur += 4; 1003 dec->cur += 4;
471 1004
472 return ((UV)d1) << 12 1005 return ((UV)d1) << 12
473 | ((UV)d2) << 8 1006 | ((UV)d2) << 8
… …
476 1009
477fail: 1010fail:
478 return (UV)-1; 1011 return (UV)-1;
479} 1012}
480 1013
481#define APPEND_GROW(n) SB \
482 if (cur + (n) >= end) \
483 { \
484 STRLEN ofs = cur - SvPVX (sv); \
485 SvGROW (sv, ofs + (n) + 1); \
486 cur = SvPVX (sv) + ofs; \
487 end = SvEND (sv); \
488 } \
489 SE
490
491#define APPEND_CH(ch) SB \
492 APPEND_GROW (1); \
493 *cur++ = (ch); \
494 SE
495
496static SV * 1014static SV *
497decode_str (dec_t *dec) 1015decode_str (dec_t *dec)
498{ 1016{
499 SV *sv = NEWSV (0,2); 1017 SV *sv = 0;
500 int utf8 = 0; 1018 int utf8 = 0;
501 char *cur = SvPVX (sv); 1019 char *dec_cur = dec->cur;
502 char *end = SvEND (sv);
503 1020
504 for (;;) 1021 do
505 { 1022 {
506 unsigned char ch = *(unsigned char *)dec->cur; 1023 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
1024 char *cur = buf;
507 1025
508 if (ch == '"') 1026 do
509 break;
510 else if (ch == '\\')
511 { 1027 {
512 switch (*++dec->cur) 1028 unsigned char ch = *(unsigned char *)dec_cur++;
1029
1030 if (expect_false (ch == '"'))
1031 {
1032 --dec_cur;
1033 break;
513 { 1034 }
514 case '\\': 1035 else if (expect_false (ch == '\\'))
515 case '/': 1036 {
516 case '"': APPEND_CH (*dec->cur++); break; 1037 switch (*dec_cur)
517
518 case 'b': APPEND_CH ('\010'); ++dec->cur; break;
519 case 't': APPEND_CH ('\011'); ++dec->cur; break;
520 case 'n': APPEND_CH ('\012'); ++dec->cur; break;
521 case 'f': APPEND_CH ('\014'); ++dec->cur; break;
522 case 'r': APPEND_CH ('\015'); ++dec->cur; break;
523
524 case 'u':
525 { 1038 {
526 UV lo, hi; 1039 case '\\':
527 ++dec->cur; 1040 case '/':
1041 case '"': *cur++ = *dec_cur++; break;
528 1042
529 hi = decode_4hex (dec); 1043 case 'b': ++dec_cur; *cur++ = '\010'; break;
530 if (hi == (UV)-1) 1044 case 't': ++dec_cur; *cur++ = '\011'; break;
531 goto fail; 1045 case 'n': ++dec_cur; *cur++ = '\012'; break;
1046 case 'f': ++dec_cur; *cur++ = '\014'; break;
1047 case 'r': ++dec_cur; *cur++ = '\015'; break;
532 1048
533 // possibly a surrogate pair 1049 case 'u':
534 if (hi >= 0xd800 && hi < 0xdc00)
535 { 1050 {
536 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 1051 UV lo, hi;
537 ERR ("missing low surrogate character in surrogate pair"); 1052 ++dec_cur;
538 1053
539 dec->cur += 2; 1054 dec->cur = dec_cur;
540
541 lo = decode_4hex (dec); 1055 hi = decode_4hex (dec);
1056 dec_cur = dec->cur;
542 if (lo == (UV)-1) 1057 if (hi == (UV)-1)
543 goto fail; 1058 goto fail;
544 1059
1060 // possibly a surrogate pair
1061 if (hi >= 0xd800)
1062 if (hi < 0xdc00)
1063 {
1064 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
1065 ERR ("missing low surrogate character in surrogate pair");
1066
1067 dec_cur += 2;
1068
1069 dec->cur = dec_cur;
1070 lo = decode_4hex (dec);
1071 dec_cur = dec->cur;
1072 if (lo == (UV)-1)
1073 goto fail;
1074
545 if (lo < 0xdc00 || lo >= 0xe000) 1075 if (lo < 0xdc00 || lo >= 0xe000)
546 ERR ("surrogate pair expected"); 1076 ERR ("surrogate pair expected");
547 1077
548 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; 1078 hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000;
1079 }
1080 else if (hi < 0xe000)
1081 ERR ("missing high surrogate character in surrogate pair");
1082
1083 if (hi >= 0x80)
1084 {
1085 utf8 = 1;
1086
1087 cur = encode_utf8 (cur, hi);
1088 }
1089 else
1090 *cur++ = hi;
549 } 1091 }
550 else if (hi >= 0xdc00 && hi < 0xe000)
551 ERR ("missing high surrogate character in surrogate pair");
552
553 if (hi >= 0x80)
554 { 1092 break;
555 utf8 = 1;
556 1093
557 APPEND_GROW (4); // at most 4 bytes for 21 bits
558 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0);
559 }
560 else 1094 default:
561 APPEND_CH (hi); 1095 --dec_cur;
1096 ERR ("illegal backslash escape sequence in string");
562 } 1097 }
563 break; 1098 }
1099 else if (expect_true (ch >= 0x20 && ch < 0x80))
1100 *cur++ = ch;
1101 else if (ch >= 0x80)
1102 {
1103 STRLEN clen;
564 1104
565 default:
566 --dec->cur; 1105 --dec_cur;
567 ERR ("illegal backslash escape sequence in string"); 1106
1107 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
1108 if (clen == (STRLEN)-1)
1109 ERR ("malformed UTF-8 character in JSON string");
1110
1111 do
1112 *cur++ = *dec_cur++;
1113 while (--clen);
1114
1115 utf8 = 1;
1116 }
1117 else
1118 {
1119 --dec_cur;
1120
1121 if (!ch)
1122 ERR ("unexpected end of string while parsing JSON string");
1123 else
1124 ERR ("invalid character encountered while parsing JSON string");
568 } 1125 }
569 } 1126 }
570 else if (ch >= 0x20 && ch <= 0x7f) 1127 while (cur < buf + SHORT_STRING_LEN);
571 APPEND_CH (*dec->cur++); 1128
572 else if (ch >= 0x80)
573 { 1129 {
574 STRLEN clen; 1130 STRLEN len = cur - buf;
575 UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
576 if (clen == (STRLEN)-1)
577 ERR ("malformed UTF-8 character in string, cannot convert to JSON");
578 1131
579 APPEND_GROW (clen); 1132 if (sv)
580 do
581 { 1133 {
582 *cur++ = *dec->cur++; 1134 STRLEN cur = SvCUR (sv);
1135
1136 if (SvLEN (sv) <= cur + len)
1137 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1138
1139 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
1140 SvCUR_set (sv, SvCUR (sv) + len);
583 } 1141 }
584 while (--clen);
585
586 utf8 = 1;
587 }
588 else if (dec->cur == dec->end)
589 ERR ("unexpected end of string while parsing json string");
590 else 1142 else
591 ERR ("invalid character encountered"); 1143 sv = newSVpvn (buf, len);
592 } 1144 }
1145 }
1146 while (*dec_cur != '"');
593 1147
594 ++dec->cur; 1148 ++dec_cur;
595 1149
596 SvCUR_set (sv, cur - SvPVX (sv)); 1150 if (sv)
597 1151 {
598 SvPOK_only (sv); 1152 SvPOK_only (sv);
599 *SvEND (sv) = 0; 1153 *SvEND (sv) = 0;
600 1154
601 if (utf8) 1155 if (utf8)
602 SvUTF8_on (sv); 1156 SvUTF8_on (sv);
1157 }
1158 else
1159 sv = newSVpvn ("", 0);
603 1160
1161 dec->cur = dec_cur;
604 return sv; 1162 return sv;
605 1163
606fail: 1164fail:
607 SvREFCNT_dec (sv); 1165 dec->cur = dec_cur;
608 return 0; 1166 return 0;
609} 1167}
610 1168
611static SV * 1169static SV *
612decode_num (dec_t *dec) 1170decode_num (dec_t *dec)
… …
670 is_nv = 1; 1228 is_nv = 1;
671 } 1229 }
672 1230
673 if (!is_nv) 1231 if (!is_nv)
674 { 1232 {
675 UV uv; 1233 int len = dec->cur - start;
676 int numtype = grok_number (start, dec->cur - start, &uv); 1234
677 if (numtype & IS_NUMBER_IN_UV) 1235 // special case the rather common 1..5-digit-int case
678 if (numtype & IS_NUMBER_NEG) 1236 if (*start == '-')
1237 switch (len)
679 { 1238 {
680 if (uv < (UV)IV_MIN) 1239 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
681 return newSViv (-(IV)uv); 1240 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1241 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1242 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1243 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
682 } 1244 }
1245 else
1246 switch (len)
1247 {
1248 case 1: return newSViv ( start [0] - '0' * 1);
1249 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1250 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1251 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1252 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1253 }
1254
1255 {
1256 UV uv;
1257 int numtype = grok_number (start, len, &uv);
1258 if (numtype & IS_NUMBER_IN_UV)
1259 if (numtype & IS_NUMBER_NEG)
1260 {
1261 if (uv < (UV)IV_MIN)
1262 return newSViv (-(IV)uv);
1263 }
683 else 1264 else
684 return newSVuv (uv); 1265 return newSVuv (uv);
685 } 1266 }
686 1267
1268 len -= *start == '-' ? 1 : 0;
1269
1270 // does not fit into IV or UV, try NV
1271 if (len <= NV_DIG)
1272 // fits into NV without loss of precision
1273 return newSVnv (json_atof (start));
1274
1275 // everything else fails, convert it to a string
1276 return newSVpvn (start, dec->cur - start);
1277 }
1278
1279 // loss of precision here
687 return newSVnv (Atof (start)); 1280 return newSVnv (json_atof (start));
688 1281
689fail: 1282fail:
690 return 0; 1283 return 0;
691} 1284}
692 1285
693static SV * 1286static SV *
694decode_av (dec_t *dec) 1287decode_av (dec_t *dec)
695{ 1288{
696 AV *av = newAV (); 1289 AV *av = newAV ();
697 1290
698 WS; 1291 DEC_INC_DEPTH;
1292 decode_ws (dec);
1293
699 if (*dec->cur == ']') 1294 if (*dec->cur == ']')
700 ++dec->cur; 1295 ++dec->cur;
701 else 1296 else
702 for (;;) 1297 for (;;)
703 { 1298 {
… …
707 if (!value) 1302 if (!value)
708 goto fail; 1303 goto fail;
709 1304
710 av_push (av, value); 1305 av_push (av, value);
711 1306
712 WS; 1307 decode_ws (dec);
713 1308
714 if (*dec->cur == ']') 1309 if (*dec->cur == ']')
715 { 1310 {
716 ++dec->cur; 1311 ++dec->cur;
717 break; 1312 break;
718 } 1313 }
719 1314
720 if (*dec->cur != ',') 1315 if (*dec->cur != ',')
721 ERR (", or ] expected while parsing array"); 1316 ERR (", or ] expected while parsing array");
722 1317
723 ++dec->cur; 1318 ++dec->cur;
1319
1320 decode_ws (dec);
1321
1322 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1323 {
1324 ++dec->cur;
1325 break;
1326 }
724 } 1327 }
725 1328
1329 DEC_DEC_DEPTH;
726 return newRV_noinc ((SV *)av); 1330 return newRV_noinc ((SV *)av);
727 1331
728fail: 1332fail:
729 SvREFCNT_dec (av); 1333 SvREFCNT_dec (av);
1334 DEC_DEC_DEPTH;
730 return 0; 1335 return 0;
731} 1336}
732 1337
733static SV * 1338static SV *
734decode_hv (dec_t *dec) 1339decode_hv (dec_t *dec)
735{ 1340{
1341 SV *sv;
736 HV *hv = newHV (); 1342 HV *hv = newHV ();
737 1343
738 WS; 1344 DEC_INC_DEPTH;
1345 decode_ws (dec);
1346
739 if (*dec->cur == '}') 1347 if (*dec->cur == '}')
740 ++dec->cur; 1348 ++dec->cur;
741 else 1349 else
742 for (;;) 1350 for (;;)
743 { 1351 {
744 SV *key, *value;
745
746 WS; EXPECT_CH ('"'); 1352 EXPECT_CH ('"');
747 1353
748 key = decode_str (dec); 1354 // heuristic: assume that
749 if (!key) 1355 // a) decode_str + hv_store_ent are abysmally slow.
750 goto fail; 1356 // b) most hash keys are short, simple ascii text.
1357 // => try to "fast-match" such strings to avoid
1358 // the overhead of decode_str + hv_store_ent.
1359 {
1360 SV *value;
1361 char *p = dec->cur;
1362 char *e = p + 24; // only try up to 24 bytes
751 1363
752 WS; EXPECT_CH (':'); 1364 for (;;)
753
754 value = decode_sv (dec);
755 if (!value)
756 { 1365 {
1366 // the >= 0x80 is false on most architectures
1367 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1368 {
1369 // slow path, back up and use decode_str
1370 SV *key = decode_str (dec);
1371 if (!key)
1372 goto fail;
1373
1374 decode_ws (dec); EXPECT_CH (':');
1375
1376 decode_ws (dec);
1377 value = decode_sv (dec);
1378 if (!value)
1379 {
1380 SvREFCNT_dec (key);
1381 goto fail;
1382 }
1383
1384 hv_store_ent (hv, key, value, 0);
757 SvREFCNT_dec (key); 1385 SvREFCNT_dec (key);
1386
1387 break;
1388 }
1389 else if (*p == '"')
1390 {
1391 // fast path, got a simple key
1392 char *key = dec->cur;
1393 int len = p - key;
1394 dec->cur = p + 1;
1395
1396 decode_ws (dec); EXPECT_CH (':');
1397
1398 decode_ws (dec);
1399 value = decode_sv (dec);
1400 if (!value)
758 goto fail; 1401 goto fail;
1402
1403 hv_store (hv, key, len, value, 0);
1404
1405 break;
1406 }
1407
1408 ++p;
759 } 1409 }
760
761 //TODO: optimise
762 hv_store_ent (hv, key, value, 0);
763
764 WS; 1410 }
1411
1412 decode_ws (dec);
765 1413
766 if (*dec->cur == '}') 1414 if (*dec->cur == '}')
767 { 1415 {
768 ++dec->cur; 1416 ++dec->cur;
769 break; 1417 break;
… …
771 1419
772 if (*dec->cur != ',') 1420 if (*dec->cur != ',')
773 ERR (", or } expected while parsing object/hash"); 1421 ERR (", or } expected while parsing object/hash");
774 1422
775 ++dec->cur; 1423 ++dec->cur;
1424
1425 decode_ws (dec);
1426
1427 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1428 {
1429 ++dec->cur;
1430 break;
1431 }
776 } 1432 }
777 1433
1434 DEC_DEC_DEPTH;
778 return newRV_noinc ((SV *)hv); 1435 sv = newRV_noinc ((SV *)hv);
1436
1437 // check filter callbacks
1438 if (dec->json.flags & F_HOOK)
1439 {
1440 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1441 {
1442 HE *cb, *he;
1443
1444 hv_iterinit (hv);
1445 he = hv_iternext (hv);
1446 hv_iterinit (hv);
1447
1448 // the next line creates a mortal sv each time its called.
1449 // might want to optimise this for common cases.
1450 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1451
1452 if (cb)
1453 {
1454 dSP;
1455 int count;
1456
1457 ENTER; SAVETMPS; PUSHMARK (SP);
1458 XPUSHs (HeVAL (he));
1459 sv_2mortal (sv);
1460
1461 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1462
1463 if (count == 1)
1464 {
1465 sv = newSVsv (POPs);
1466 FREETMPS; LEAVE;
1467 return sv;
1468 }
1469
1470 SvREFCNT_inc (sv);
1471 FREETMPS; LEAVE;
1472 }
1473 }
1474
1475 if (dec->json.cb_object)
1476 {
1477 dSP;
1478 int count;
1479
1480 ENTER; SAVETMPS; PUSHMARK (SP);
1481 XPUSHs (sv_2mortal (sv));
1482
1483 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1484
1485 if (count == 1)
1486 {
1487 sv = newSVsv (POPs);
1488 FREETMPS; LEAVE;
1489 return sv;
1490 }
1491
1492 SvREFCNT_inc (sv);
1493 FREETMPS; LEAVE;
1494 }
1495 }
1496
1497 return sv;
779 1498
780fail: 1499fail:
781 SvREFCNT_dec (hv); 1500 SvREFCNT_dec (hv);
1501 DEC_DEC_DEPTH;
1502 return 0;
1503}
1504
1505static SV *
1506decode_tag (dec_t *dec)
1507{
1508 SV *tag = 0;
1509 SV *val = 0;
1510
1511 if (!(dec->json.flags & F_ALLOW_TAGS))
1512 ERR ("malformed JSON string, neither array, object, number, string or atom");
1513
1514 ++dec->cur;
1515
1516 tag = decode_sv (dec);
1517 if (!tag)
1518 goto fail;
1519
1520 if (!SvPOK (tag))
1521 ERR ("malformed JSON string, (tag) must be a string");
1522
1523 if (*dec->cur != ')')
1524 ERR (") expected after tag");
1525
1526 ++dec->cur;
1527
1528 val = decode_sv (dec);
1529 if (!val)
1530 goto fail;
1531
1532 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1533 ERR ("malformed JSON string, tag value must be an array");
1534
1535 {
1536 AV *av = (AV *)SvRV (val);
1537 int i, len = av_len (av) + 1;
1538 HV *stash = gv_stashsv (tag, 0);
1539 SV *sv;
1540
1541 if (!stash)
1542 ERR ("cannot decode perl-object (package does not exist)");
1543
1544 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1545
1546 if (!method)
1547 ERR ("cannot decode perl-object (package does not have a THAW method)");
1548
1549 dSP;
1550
1551 ENTER; SAVETMPS; PUSHMARK (SP);
1552 EXTEND (SP, len + 2);
1553 // we re-bless the reference to get overload and other niceties right
1554 PUSHs (tag);
1555 PUSHs (sv_json);
1556
1557 for (i = 0; i < len; ++i)
1558 PUSHs (*av_fetch (av, i, 1));
1559
1560 PUTBACK;
1561 call_sv ((SV *)GvCV (method), G_SCALAR);
1562 SPAGAIN;
1563
1564 SvREFCNT_dec (tag);
1565 SvREFCNT_dec (val);
1566 sv = SvREFCNT_inc (POPs);
1567
1568 PUTBACK;
1569
1570 FREETMPS; LEAVE;
1571
1572 return sv;
1573 }
1574
1575fail:
1576 SvREFCNT_dec (tag);
1577 SvREFCNT_dec (val);
782 return 0; 1578 return 0;
783} 1579}
784 1580
785static SV * 1581static SV *
786decode_sv (dec_t *dec) 1582decode_sv (dec_t *dec)
787{ 1583{
788 WS; 1584 // the beauty of JSON: you need exactly one character lookahead
1585 // to parse everything.
789 switch (*dec->cur) 1586 switch (*dec->cur)
790 { 1587 {
791 case '"': ++dec->cur; return decode_str (dec); 1588 case '"': ++dec->cur; return decode_str (dec);
792 case '[': ++dec->cur; return decode_av (dec); 1589 case '[': ++dec->cur; return decode_av (dec);
793 case '{': ++dec->cur; return decode_hv (dec); 1590 case '{': ++dec->cur; return decode_hv (dec);
1591 case '(': return decode_tag (dec);
794 1592
795 case '-': 1593 case '-':
796 case '0': case '1': case '2': case '3': case '4': 1594 case '0': case '1': case '2': case '3': case '4':
797 case '5': case '6': case '7': case '8': case '9': 1595 case '5': case '6': case '7': case '8': case '9':
798 return decode_num (dec); 1596 return decode_num (dec);
799 1597
800 case 't': 1598 case 't':
801 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1599 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
802 { 1600 {
803 dec->cur += 4; 1601 dec->cur += 4;
1602#if JSON_SLOW
1603 types_true = get_bool ("Types::Serialiser::true");
1604#endif
804 return newSViv (1); 1605 return newSVsv (types_true);
805 } 1606 }
806 else 1607 else
807 ERR ("'true' expected"); 1608 ERR ("'true' expected");
808 1609
809 break; 1610 break;
810 1611
811 case 'f': 1612 case 'f':
812 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1613 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
813 { 1614 {
814 dec->cur += 5; 1615 dec->cur += 5;
1616#if JSON_SLOW
1617 types_false = get_bool ("Types::Serialiser::false");
1618#endif
815 return newSViv (0); 1619 return newSVsv (types_false);
816 } 1620 }
817 else 1621 else
818 ERR ("'false' expected"); 1622 ERR ("'false' expected");
819 1623
820 break; 1624 break;
… …
829 ERR ("'null' expected"); 1633 ERR ("'null' expected");
830 1634
831 break; 1635 break;
832 1636
833 default: 1637 default:
834 ERR ("malformed json string"); 1638 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
835 break; 1639 break;
836 } 1640 }
837 1641
838fail: 1642fail:
839 return 0; 1643 return 0;
840} 1644}
841 1645
842static SV * 1646static SV *
843decode_json (SV *string, UV flags) 1647decode_json (SV *string, JSON *json, char **offset_return)
844{ 1648{
1649 dec_t dec;
845 SV *sv; 1650 SV *sv;
846 1651
847 if (flags & F_UTF8) 1652 /* work around bugs in 5.10 where manipulating magic values
1653 * makes perl ignore the magic in subsequent accesses.
1654 * also make a copy of non-PV values, to get them into a clean
1655 * state (SvPV should do that, but it's buggy, see below).
1656 */
1657 /*SvGETMAGIC (string);*/
1658 if (SvMAGICAL (string) || !SvPOK (string))
1659 string = sv_2mortal (newSVsv (string));
1660
1661 SvUPGRADE (string, SVt_PV);
1662
1663 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1664 * assertion with -DDEBUGGING, although SvCUR is documented to
1665 * return the xpv_cur field which certainly exists after upgrading.
1666 * according to nicholas clark, calling SvPOK fixes this.
1667 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1668 * and hope for the best.
1669 * Damnit, SvPV_nolen still trips over yet another assertion. This
1670 * assertion business is seriously broken, try yet another workaround
1671 * for the broken -DDEBUGGING.
1672 */
1673 {
1674#ifdef DEBUGGING
1675 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1676#else
1677 STRLEN offset = SvCUR (string);
1678#endif
1679
1680 if (offset > json->max_size && json->max_size)
1681 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1682 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1683 }
1684
1685 if (DECODE_WANTS_OCTETS (json))
848 sv_utf8_downgrade (string, 0); 1686 sv_utf8_downgrade (string, 0);
849 else 1687 else
850 sv_utf8_upgrade (string); 1688 sv_utf8_upgrade (string);
851 1689
852 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1690 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
853 1691
854 dec_t dec; 1692 dec.json = *json;
855 dec.flags = flags;
856 dec.cur = SvPVX (string); 1693 dec.cur = SvPVX (string);
857 dec.end = SvEND (string); 1694 dec.end = SvEND (string);
858 dec.err = 0; 1695 dec.err = 0;
1696 dec.depth = 0;
859 1697
1698 if (dec.json.cb_object || dec.json.cb_sk_object)
1699 dec.json.flags |= F_HOOK;
1700
1701 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1702
1703 decode_ws (&dec);
860 sv = decode_sv (&dec); 1704 sv = decode_sv (&dec);
861 1705
1706 if (offset_return)
1707 *offset_return = dec.cur;
1708
1709 if (!(offset_return || !sv))
1710 {
1711 // check for trailing garbage
1712 decode_ws (&dec);
1713
1714 if (*dec.cur)
1715 {
1716 dec.err = "garbage after JSON object";
1717 SvREFCNT_dec (sv);
1718 sv = 0;
1719 }
1720 }
1721
862 if (!sv) 1722 if (!sv)
863 { 1723 {
864 IV offset = utf8_distance (dec.cur, SvPVX (string));
865 SV *uni = sv_newmortal (); 1724 SV *uni = sv_newmortal ();
1725
866 // horrible hack to silence warning inside pv_uni_display 1726 // horrible hack to silence warning inside pv_uni_display
867 COP cop; 1727 COP cop = *PL_curcop;
868 memset (&cop, 0, sizeof (cop));
869 cop.cop_warnings = pWARN_NONE; 1728 cop.cop_warnings = pWARN_NONE;
1729 ENTER;
870 SAVEVPTR (PL_curcop); 1730 SAVEVPTR (PL_curcop);
871 PL_curcop = &cop; 1731 PL_curcop = &cop;
872
873 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1732 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1733 LEAVE;
1734
874 croak ("%s, at character offset %d (%s)", 1735 croak ("%s, at character offset %d (before \"%s\")",
875 dec.err, 1736 dec.err,
876 (int)offset, 1737 (int)ptr_to_index (string, dec.cur),
877 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1738 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
878 } 1739 }
879 1740
880 sv = sv_2mortal (sv); 1741 sv = sv_2mortal (sv);
881 1742
882 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1743 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
883 croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)"); 1744 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
884 1745
885 return sv; 1746 return sv;
886} 1747}
887 1748
1749/////////////////////////////////////////////////////////////////////////////
1750// incremental parser
1751
1752static void
1753incr_parse (JSON *self)
1754{
1755 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1756
1757 // the state machine here is a bit convoluted and could be simplified a lot
1758 // but this would make it slower, so...
1759
1760 for (;;)
1761 {
1762 //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D
1763 switch (self->incr_mode)
1764 {
1765 // only used for initial whitespace skipping
1766 case INCR_M_WS:
1767 for (;;)
1768 {
1769 if (*p > 0x20)
1770 {
1771 if (*p == '#')
1772 {
1773 self->incr_mode = INCR_M_C0;
1774 goto incr_m_c;
1775 }
1776 else
1777 {
1778 self->incr_mode = INCR_M_JSON;
1779 goto incr_m_json;
1780 }
1781 }
1782 else if (!*p)
1783 goto interrupt;
1784
1785 ++p;
1786 }
1787
1788 // skip a single char inside a string (for \\-processing)
1789 case INCR_M_BS:
1790 if (!*p)
1791 goto interrupt;
1792
1793 ++p;
1794 self->incr_mode = INCR_M_STR;
1795 goto incr_m_str;
1796
1797 // inside #-style comments
1798 case INCR_M_C0:
1799 case INCR_M_C1:
1800 incr_m_c:
1801 for (;;)
1802 {
1803 if (*p == '\n')
1804 {
1805 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1806 break;
1807 }
1808 else if (!*p)
1809 goto interrupt;
1810
1811 ++p;
1812 }
1813
1814 break;
1815
1816 // inside a string
1817 case INCR_M_STR:
1818 incr_m_str:
1819 for (;;)
1820 {
1821 if (*p == '"')
1822 {
1823 ++p;
1824 self->incr_mode = INCR_M_JSON;
1825
1826 if (!self->incr_nest)
1827 goto interrupt;
1828
1829 goto incr_m_json;
1830 }
1831 else if (*p == '\\')
1832 {
1833 ++p; // "virtually" consumes character after \
1834
1835 if (!*p) // if at end of string we have to switch modes
1836 {
1837 self->incr_mode = INCR_M_BS;
1838 goto interrupt;
1839 }
1840 }
1841 else if (!*p)
1842 goto interrupt;
1843
1844 ++p;
1845 }
1846
1847 // after initial ws, outside string
1848 case INCR_M_JSON:
1849 incr_m_json:
1850 for (;;)
1851 {
1852 switch (*p++)
1853 {
1854 case 0:
1855 --p;
1856 goto interrupt;
1857
1858 case 0x09:
1859 case 0x0a:
1860 case 0x0d:
1861 case 0x20:
1862 if (!self->incr_nest)
1863 {
1864 --p; // do not eat the whitespace, let the next round do it
1865 goto interrupt;
1866 }
1867 break;
1868
1869 case '"':
1870 self->incr_mode = INCR_M_STR;
1871 goto incr_m_str;
1872
1873 case '[':
1874 case '{':
1875 case '(':
1876 if (++self->incr_nest > self->max_depth)
1877 croak (ERR_NESTING_EXCEEDED);
1878 break;
1879
1880 case ']':
1881 case '}':
1882 if (--self->incr_nest <= 0)
1883 goto interrupt;
1884 break;
1885
1886 case ')':
1887 --self->incr_nest;
1888 break;
1889
1890 case '#':
1891 self->incr_mode = INCR_M_C1;
1892 goto incr_m_c;
1893 }
1894 }
1895 }
1896
1897 modechange:
1898 ;
1899 }
1900
1901interrupt:
1902 self->incr_pos = p - SvPVX (self->incr_text);
1903 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1904 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1905}
1906
1907/////////////////////////////////////////////////////////////////////////////
1908// XS interface functions
1909
888MODULE = JSON::XS PACKAGE = JSON::XS 1910MODULE = JSON::XS PACKAGE = JSON::XS
889 1911
890BOOT: 1912BOOT:
891{ 1913{
892 int i; 1914 int i;
893 1915
894 memset (decode_hexdigit, 0xff, 256);
895 for (i = 10; i--; ) 1916 for (i = 0; i < 256; ++i)
896 decode_hexdigit ['0' + i] = i; 1917 decode_hexdigit [i] =
1918 i >= '0' && i <= '9' ? i - '0'
1919 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1920 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1921 : -1;
897 1922
898 for (i = 7; i--; )
899 {
900 decode_hexdigit ['a' + i] = 10 + i;
901 decode_hexdigit ['A' + i] = 10 + i;
902 }
903
904 json_stash = gv_stashpv ("JSON::XS", 1); 1923 json_stash = gv_stashpv ("JSON::XS" , 1);
1924 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1925
1926 types_true = get_bool ("Types::Serialiser::true");
1927 types_false = get_bool ("Types::Serialiser::false");
1928
1929 sv_json = newSVpv ("JSON", 0);
1930 SvREADONLY_on (sv_json);
1931
1932 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
905} 1933}
906 1934
907PROTOTYPES: DISABLE 1935PROTOTYPES: DISABLE
908 1936
909SV *new (char *dummy) 1937void CLONE (...)
910 CODE: 1938 CODE:
911 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1939 json_stash = 0;
1940 types_boolean_stash = 0;
1941
1942void new (char *klass)
1943 PPCODE:
1944{
1945 SV *pv = NEWSV (0, sizeof (JSON));
1946 SvPOK_only (pv);
1947 json_init ((JSON *)SvPVX (pv));
1948 XPUSHs (sv_2mortal (sv_bless (
1949 newRV_noinc (pv),
1950 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1951 )));
1952}
1953
1954void ascii (JSON *self, int enable = 1)
1955 ALIAS:
1956 ascii = F_ASCII
1957 latin1 = F_LATIN1
1958 utf8 = F_UTF8
1959 indent = F_INDENT
1960 canonical = F_CANONICAL
1961 space_before = F_SPACE_BEFORE
1962 space_after = F_SPACE_AFTER
1963 pretty = F_PRETTY
1964 allow_nonref = F_ALLOW_NONREF
1965 shrink = F_SHRINK
1966 allow_blessed = F_ALLOW_BLESSED
1967 convert_blessed = F_CONV_BLESSED
1968 relaxed = F_RELAXED
1969 allow_unknown = F_ALLOW_UNKNOWN
1970 allow_tags = F_ALLOW_TAGS
1971 PPCODE:
1972{
1973 if (enable)
1974 self->flags |= ix;
1975 else
1976 self->flags &= ~ix;
1977
1978 XPUSHs (ST (0));
1979}
1980
1981void get_ascii (JSON *self)
1982 ALIAS:
1983 get_ascii = F_ASCII
1984 get_latin1 = F_LATIN1
1985 get_utf8 = F_UTF8
1986 get_indent = F_INDENT
1987 get_canonical = F_CANONICAL
1988 get_space_before = F_SPACE_BEFORE
1989 get_space_after = F_SPACE_AFTER
1990 get_allow_nonref = F_ALLOW_NONREF
1991 get_shrink = F_SHRINK
1992 get_allow_blessed = F_ALLOW_BLESSED
1993 get_convert_blessed = F_CONV_BLESSED
1994 get_relaxed = F_RELAXED
1995 get_allow_unknown = F_ALLOW_UNKNOWN
1996 get_allow_tags = F_ALLOW_TAGS
1997 PPCODE:
1998 XPUSHs (boolSV (self->flags & ix));
1999
2000void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
2001 PPCODE:
2002 self->max_depth = max_depth;
2003 XPUSHs (ST (0));
2004
2005U32 get_max_depth (JSON *self)
2006 CODE:
2007 RETVAL = self->max_depth;
912 OUTPUT: 2008 OUTPUT:
913 RETVAL 2009 RETVAL
914 2010
915SV *ascii (SV *self, int enable) 2011void max_size (JSON *self, U32 max_size = 0)
916 ALIAS: 2012 PPCODE:
917 ascii = F_ASCII 2013 self->max_size = max_size;
918 utf8 = F_UTF8 2014 XPUSHs (ST (0));
919 indent = F_INDENT 2015
920 canonical = F_CANONICAL 2016int get_max_size (JSON *self)
921 space_before = F_SPACE_BEFORE
922 space_after = F_SPACE_AFTER
923 json_rpc = F_JSON_RPC
924 pretty = F_PRETTY
925 allow_nonref = F_ALLOW_NONREF
926 CODE: 2017 CODE:
927{ 2018 RETVAL = self->max_size;
928 UV *uv = SvJSON (self);
929 if (enable)
930 *uv |= ix;
931 else
932 *uv &= ~ix;
933
934 RETVAL = newSVsv (self);
935}
936 OUTPUT: 2019 OUTPUT:
937 RETVAL 2020 RETVAL
938 2021
939void encode (SV *self, SV *scalar) 2022void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
940 PPCODE: 2023 PPCODE:
941 XPUSHs (encode_json (scalar, *SvJSON (self))); 2024{
2025 SvREFCNT_dec (self->cb_object);
2026 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
942 2027
943void decode (SV *self, SV *jsonstr) 2028 XPUSHs (ST (0));
2029}
2030
2031void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
944 PPCODE: 2032 PPCODE:
945 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 2033{
2034 if (!self->cb_sk_object)
2035 self->cb_sk_object = newHV ();
2036
2037 if (SvOK (cb))
2038 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
2039 else
2040 {
2041 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
2042
2043 if (!HvKEYS (self->cb_sk_object))
2044 {
2045 SvREFCNT_dec (self->cb_sk_object);
2046 self->cb_sk_object = 0;
2047 }
2048 }
2049
2050 XPUSHs (ST (0));
2051}
2052
2053void encode (JSON *self, SV *scalar)
2054 PPCODE:
2055 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2056 XPUSHs (scalar);
2057
2058void decode (JSON *self, SV *jsonstr)
2059 PPCODE:
2060 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2061 XPUSHs (jsonstr);
2062
2063void decode_prefix (JSON *self, SV *jsonstr)
2064 PPCODE:
2065{
2066 SV *sv;
2067 char *offset;
2068 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
2069 EXTEND (SP, 2);
2070 PUSHs (sv);
2071 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
2072}
2073
2074void incr_parse (JSON *self, SV *jsonstr = 0)
2075 PPCODE:
2076{
2077 if (!self->incr_text)
2078 self->incr_text = newSVpvn ("", 0);
2079
2080 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2081 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2082 if (DECODE_WANTS_OCTETS (self))
2083 {
2084 if (self->incr_pos)
2085 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2086 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2087
2088 sv_utf8_downgrade (self->incr_text, 0);
2089 }
2090 else
2091 {
2092 sv_utf8_upgrade (self->incr_text);
2093
2094 if (self->incr_pos)
2095 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2096 - (U8 *)SvPVX (self->incr_text);
2097 }
2098
2099 // append data, if any
2100 if (jsonstr)
2101 {
2102 /* make sure both strings have same encoding */
2103 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2104 if (SvUTF8 (jsonstr))
2105 sv_utf8_downgrade (jsonstr, 0);
2106 else
2107 sv_utf8_upgrade (jsonstr);
2108
2109 /* and then just blindly append */
2110 {
2111 STRLEN len;
2112 const char *str = SvPV (jsonstr, len);
2113 STRLEN cur = SvCUR (self->incr_text);
2114
2115 if (SvLEN (self->incr_text) <= cur + len)
2116 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
2117
2118 Move (str, SvEND (self->incr_text), len, char);
2119 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2120 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2121 }
2122 }
2123
2124 if (GIMME_V != G_VOID)
2125 do
2126 {
2127 SV *sv;
2128 char *offset;
2129
2130 if (!INCR_DONE (self))
2131 {
2132 incr_parse (self);
2133
2134 if (self->incr_pos > self->max_size && self->max_size)
2135 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2136 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2137
2138 if (!INCR_DONE (self))
2139 {
2140 // as an optimisation, do not accumulate white space in the incr buffer
2141 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2142 {
2143 self->incr_pos = 0;
2144 SvCUR_set (self->incr_text, 0);
2145 }
2146
2147 break;
2148 }
2149 }
2150
2151 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2152 XPUSHs (sv);
2153
2154 self->incr_pos -= offset - SvPVX (self->incr_text);
2155 self->incr_nest = 0;
2156 self->incr_mode = 0;
2157
2158 sv_chop (self->incr_text, offset);
2159 }
2160 while (GIMME_V == G_ARRAY);
2161}
2162
2163SV *incr_text (JSON *self)
2164 ATTRS: lvalue
2165 CODE:
2166{
2167 if (self->incr_pos)
2168 croak ("incr_text can not be called when the incremental parser already started parsing");
2169
2170 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
2171}
2172 OUTPUT:
2173 RETVAL
2174
2175void incr_skip (JSON *self)
2176 CODE:
2177{
2178 if (self->incr_pos)
2179 {
2180 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2181 self->incr_pos = 0;
2182 self->incr_nest = 0;
2183 self->incr_mode = 0;
2184 }
2185}
2186
2187void incr_reset (JSON *self)
2188 CODE:
2189{
2190 SvREFCNT_dec (self->incr_text);
2191 self->incr_text = 0;
2192 self->incr_pos = 0;
2193 self->incr_nest = 0;
2194 self->incr_mode = 0;
2195}
2196
2197void DESTROY (JSON *self)
2198 CODE:
2199 SvREFCNT_dec (self->cb_sk_object);
2200 SvREFCNT_dec (self->cb_object);
2201 SvREFCNT_dec (self->incr_text);
946 2202
947PROTOTYPES: ENABLE 2203PROTOTYPES: ENABLE
948 2204
949void to_json (SV *scalar) 2205void encode_json (SV *scalar)
950 PPCODE: 2206 PPCODE:
951 XPUSHs (encode_json (scalar, F_UTF8)); 2207{
2208 JSON json;
2209 json_init (&json);
2210 json.flags |= F_UTF8;
2211 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2212 XPUSHs (scalar);
2213}
952 2214
953void from_json (SV *jsonstr) 2215void decode_json (SV *jsonstr)
954 PPCODE: 2216 PPCODE:
955 XPUSHs (decode_json (jsonstr, F_UTF8)); 2217{
2218 JSON json;
2219 json_init (&json);
2220 json.flags |= F_UTF8;
2221 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2222 XPUSHs (jsonstr);
2223}
956 2224

Diff Legend

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