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 1.39 by root, Mon Jun 11 03:42:57 2007 UTC vs.
Revision 1.122 by root, Tue Oct 29 15:55:49 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" 8#include <stdio.h>
9#include <limits.h>
10#include <float.h>
9 11
10#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 14#endif
13 15
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// compatibility with perl <5.18
23#ifndef HvNAMELEN_get
24# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
25#endif
26#ifndef HvNAMELEN
27# define HvNAMELEN(hv) HvNAMELEN_get (hv)
28#endif
29#ifndef HvNAMEUTF8
30# define HvNAMEUTF8(hv) 0
31#endif
32
33// three extra for rounding, sign, and end of string
34#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
35
14#define F_ASCII 0x00000001UL 36#define F_ASCII 0x00000001UL
15#define F_LATIN1 0x00000002UL 37#define F_LATIN1 0x00000002UL
16#define F_UTF8 0x00000004UL 38#define F_UTF8 0x00000004UL
17#define F_INDENT 0x00000008UL 39#define F_INDENT 0x00000008UL
18#define F_CANONICAL 0x00000010UL 40#define F_CANONICAL 0x00000010UL
19#define F_SPACE_BEFORE 0x00000020UL 41#define F_SPACE_BEFORE 0x00000020UL
20#define F_SPACE_AFTER 0x00000040UL 42#define F_SPACE_AFTER 0x00000040UL
21#define F_ALLOW_NONREF 0x00000100UL 43#define F_ALLOW_NONREF 0x00000100UL
22#define F_SHRINK 0x00000200UL 44#define F_SHRINK 0x00000200UL
45#define F_ALLOW_BLESSED 0x00000400UL
46#define F_CONV_BLESSED 0x00000800UL
23#define F_MAXDEPTH 0xf8000000UL 47#define F_RELAXED 0x00001000UL
24#define S_MAXDEPTH 27 48#define F_ALLOW_UNKNOWN 0x00002000UL
25 49#define F_ALLOW_TAGS 0x00004000UL
26#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 50#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
27
28// F_SELFCONVERT? <=> to_json/toJson
29// F_BLESSED? <=> { $__class__$ => }
30 51
31#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 52#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
32#define F_DEFAULT (9UL << S_MAXDEPTH)
33 53
34#define INIT_SIZE 32 // initial scalar size to be allocated 54#define INIT_SIZE 32 // initial scalar size to be allocated
35#define INDENT_STEP 3 // spaces per indentation level 55#define INDENT_STEP 3 // spaces per indentation level
36 56
37#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 57#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
38 58
59#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
60
39#define SB do { 61#define SB do {
40#define SE } while (0) 62#define SE } while (0)
41 63
42#if __GNUC__ >= 3 64#if __GNUC__ >= 3
43# define expect(expr,value) __builtin_expect ((expr),(value)) 65# define expect(expr,value) __builtin_expect ((expr), (value))
44# define inline inline 66# define INLINE static inline
45#else 67#else
46# define expect(expr,value) (expr) 68# define expect(expr,value) (expr)
47# define inline static 69# define INLINE static
48#endif 70#endif
49 71
50#define expect_false(expr) expect ((expr) != 0, 0) 72#define expect_false(expr) expect ((expr) != 0, 0)
51#define expect_true(expr) expect ((expr) != 0, 1) 73#define expect_true(expr) expect ((expr) != 0, 1)
52 74
75#define IN_RANGE_INC(type,val,beg,end) \
76 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
77 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
78
79#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
80
81#ifdef USE_ITHREADS
82# define JSON_SLOW 1
83# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
84#else
85# define JSON_SLOW 0
86# define JSON_STASH json_stash
87#endif
88
89// the amount of HEs to allocate on the stack, when sorting keys
90#define STACK_HES 64
91
53static HV *json_stash; // JSON::XS:: 92static HV *json_stash, *types_boolean_stash; // JSON::XS::
93static SV *types_true, *types_false, *sv_json;
94
95enum {
96 INCR_M_WS = 0, // initial whitespace skipping, must be 0
97 INCR_M_STR, // inside string
98 INCR_M_BS, // inside backslash
99 INCR_M_C0, // inside comment in initial whitespace sequence
100 INCR_M_C1, // inside comment in other places
101 INCR_M_JSON // outside anything, count nesting
102};
103
104#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
105
106typedef struct {
107 U32 flags;
108 U32 max_depth;
109 STRLEN max_size;
110
111 SV *cb_object;
112 HV *cb_sk_object;
113
114 // for the incremental parser
115 SV *incr_text; // the source text so far
116 STRLEN incr_pos; // the current offset into the text
117 int incr_nest; // {[]}-nesting level
118 unsigned char incr_mode;
119} JSON;
120
121INLINE void
122json_init (JSON *json)
123{
124 Zero (json, 1, JSON);
125 json->max_depth = 512;
126}
54 127
55///////////////////////////////////////////////////////////////////////////// 128/////////////////////////////////////////////////////////////////////////////
56// utility functions 129// utility functions
57 130
58static UV * 131INLINE SV *
59SvJSON (SV *sv) 132get_bool (const char *name)
60{ 133{
61 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 134 SV *sv = get_sv (name, 1);
62 croak ("object is not of type JSON::XS");
63 135
64 return &SvUVX (SvRV (sv)); 136 SvREADONLY_on (sv);
65} 137 SvREADONLY_on (SvRV (sv));
66 138
67static void 139 return sv;
140}
141
142INLINE void
68shrink (SV *sv) 143shrink (SV *sv)
69{ 144{
70 sv_utf8_downgrade (sv, 1); 145 sv_utf8_downgrade (sv, 1);
146
71 if (SvLEN (sv) > SvCUR (sv) + 1) 147 if (SvLEN (sv) > SvCUR (sv) + 1)
72 { 148 {
73#ifdef SvPV_shrink_to_cur 149#ifdef SvPV_shrink_to_cur
74 SvPV_shrink_to_cur (sv); 150 SvPV_shrink_to_cur (sv);
75#elif defined (SvPV_renew) 151#elif defined (SvPV_renew)
… …
81// decode an utf-8 character and return it, or (UV)-1 in 157// decode an utf-8 character and return it, or (UV)-1 in
82// case of an error. 158// case of an error.
83// we special-case "safe" characters from U+80 .. U+7FF, 159// we special-case "safe" characters from U+80 .. U+7FF,
84// but use the very good perl function to parse anything else. 160// but use the very good perl function to parse anything else.
85// note that we never call this function for a ascii codepoints 161// note that we never call this function for a ascii codepoints
86inline UV 162INLINE UV
87decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 163decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
88{ 164{
89 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 165 if (expect_true (len >= 2
90 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 166 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
91 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 167 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
92 { 168 {
93 *clen = 2; 169 *clen = 2;
94 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 170 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
95 } 171 }
96 else 172 else
97 { 173 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
98 *clen = (STRLEN)-1; 174}
99 return (UV)-1; 175
176// likewise for encoding, also never called for ascii codepoints
177// this function takes advantage of this fact, although current gccs
178// seem to optimise the check for >= 0x80 away anyways
179INLINE unsigned char *
180encode_utf8 (unsigned char *s, UV ch)
181{
182 if (expect_false (ch < 0x000080))
183 *s++ = ch;
184 else if (expect_true (ch < 0x000800))
185 *s++ = 0xc0 | ( ch >> 6),
186 *s++ = 0x80 | ( ch & 0x3f);
187 else if ( ch < 0x010000)
188 *s++ = 0xe0 | ( ch >> 12),
189 *s++ = 0x80 | ((ch >> 6) & 0x3f),
190 *s++ = 0x80 | ( ch & 0x3f);
191 else if ( ch < 0x110000)
192 *s++ = 0xf0 | ( ch >> 18),
193 *s++ = 0x80 | ((ch >> 12) & 0x3f),
194 *s++ = 0x80 | ((ch >> 6) & 0x3f),
195 *s++ = 0x80 | ( ch & 0x3f);
196
197 return s;
198}
199
200// convert offset pointer to character index, sv must be string
201static STRLEN
202ptr_to_index (SV *sv, char *offset)
203{
204 return SvUTF8 (sv)
205 ? utf8_distance (offset, SvPVX (sv))
206 : offset - SvPVX (sv);
207}
208
209/////////////////////////////////////////////////////////////////////////////
210// fp hell
211
212// scan a group of digits, and a trailing exponent
213static void
214json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
215{
216 UV uaccum = 0;
217 int eaccum = 0;
218
219 // if we recurse too deep, skip all remaining digits
220 // to avoid a stack overflow attack
221 if (expect_false (--maxdepth <= 0))
222 while (((U8)*s - '0') < 10)
223 ++s;
224
225 for (;;)
100 } 226 {
101} 227 U8 dig = (U8)*s - '0';
102 228
229 if (expect_false (dig >= 10))
230 {
231 if (dig == (U8)((U8)'.' - (U8)'0'))
232 {
233 ++s;
234 json_atof_scan1 (s, accum, expo, 1, maxdepth);
235 }
236 else if ((dig | ' ') == 'e' - '0')
237 {
238 int exp2 = 0;
239 int neg = 0;
240
241 ++s;
242
243 if (*s == '-')
244 {
245 ++s;
246 neg = 1;
247 }
248 else if (*s == '+')
249 ++s;
250
251 while ((dig = (U8)*s - '0') < 10)
252 exp2 = exp2 * 10 + *s++ - '0';
253
254 *expo += neg ? -exp2 : exp2;
255 }
256
257 break;
258 }
259
260 ++s;
261
262 uaccum = uaccum * 10 + dig;
263 ++eaccum;
264
265 // if we have too many digits, then recurse for more
266 // we actually do this for rather few digits
267 if (uaccum >= (UV_MAX - 9) / 10)
268 {
269 if (postdp) *expo -= eaccum;
270 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
271 if (postdp) *expo += eaccum;
272
273 break;
274 }
275 }
276
277 // this relies greatly on the quality of the pow ()
278 // implementation of the platform, but a good
279 // implementation is hard to beat.
280 // (IEEE 754 conformant ones are required to be exact)
281 if (postdp) *expo -= eaccum;
282 *accum += uaccum * Perl_pow (10., *expo);
283 *expo += eaccum;
284}
285
286static NV
287json_atof (const char *s)
288{
289 NV accum = 0.;
290 int expo = 0;
291 int neg = 0;
292
293 if (*s == '-')
294 {
295 ++s;
296 neg = 1;
297 }
298
299 // a recursion depth of ten gives us >>500 bits
300 json_atof_scan1 (s, &accum, &expo, 0, 10);
301
302 return neg ? -accum : accum;
303}
103///////////////////////////////////////////////////////////////////////////// 304/////////////////////////////////////////////////////////////////////////////
104// encoder 305// encoder
105 306
106// structure used for encoding JSON 307// structure used for encoding JSON
107typedef struct 308typedef struct
108{ 309{
109 char *cur; // SvPVX (sv) + current output position 310 char *cur; // SvPVX (sv) + current output position
110 char *end; // SvEND (sv) 311 char *end; // SvEND (sv)
111 SV *sv; // result scalar 312 SV *sv; // result scalar
112 U32 flags; // F_* 313 JSON json;
113 U32 indent; // indentation level 314 U32 indent; // indentation level
114 U32 maxdepth; // max. indentation/recursion level 315 UV limit; // escape character values >= this value when encoding
115} enc_t; 316} enc_t;
116 317
117inline void 318INLINE void
118need (enc_t *enc, STRLEN len) 319need (enc_t *enc, STRLEN len)
119{ 320{
120 if (expect_false (enc->cur + len >= enc->end)) 321 if (expect_false (enc->cur + len >= enc->end))
121 { 322 {
122 STRLEN cur = enc->cur - SvPVX (enc->sv); 323 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
123 SvGROW (enc->sv, cur + len + 1); 324 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
124 enc->cur = SvPVX (enc->sv) + cur; 325 enc->cur = SvPVX (enc->sv) + cur;
125 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 326 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
126 } 327 }
127} 328}
128 329
129inline void 330INLINE void
130encode_ch (enc_t *enc, char ch) 331encode_ch (enc_t *enc, char ch)
131{ 332{
132 need (enc, 1); 333 need (enc, 1);
133 *enc->cur++ = ch; 334 *enc->cur++ = ch;
134} 335}
… …
178 STRLEN clen; 379 STRLEN clen;
179 UV uch; 380 UV uch;
180 381
181 if (is_utf8) 382 if (is_utf8)
182 { 383 {
183 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
184 uch = decode_utf8 (str, end - str, &clen); 384 uch = decode_utf8 (str, end - str, &clen);
185 if (clen == (STRLEN)-1) 385 if (clen == (STRLEN)-1)
186 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 386 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
187 } 387 }
188 else 388 else
189 { 389 {
190 uch = ch; 390 uch = ch;
191 clen = 1; 391 clen = 1;
192 } 392 }
193 393
194 if (uch > 0x10FFFFUL) 394 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
195 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
196
197 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
198 { 395 {
199 if (uch > 0xFFFFUL) 396 if (uch >= 0x10000UL)
200 { 397 {
398 if (uch >= 0x110000UL)
399 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
400
201 need (enc, len += 11); 401 need (enc, len += 11);
202 sprintf (enc->cur, "\\u%04x\\u%04x", 402 sprintf (enc->cur, "\\u%04x\\u%04x",
203 (int)((uch - 0x10000) / 0x400 + 0xD800), 403 (int)((uch - 0x10000) / 0x400 + 0xD800),
204 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 404 (int)((uch - 0x10000) % 0x400 + 0xDC00));
205 enc->cur += 12; 405 enc->cur += 12;
206 } 406 }
207 else 407 else
208 { 408 {
209 static char hexdigit [16] = "0123456789abcdef";
210 need (enc, len += 5); 409 need (enc, len += 5);
211 *enc->cur++ = '\\'; 410 *enc->cur++ = '\\';
212 *enc->cur++ = 'u'; 411 *enc->cur++ = 'u';
213 *enc->cur++ = hexdigit [ uch >> 12 ]; 412 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
214 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 413 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
215 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 414 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
216 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 415 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
217 } 416 }
218 417
219 str += clen; 418 str += clen;
220 } 419 }
221 else if (enc->flags & F_LATIN1) 420 else if (enc->json.flags & F_LATIN1)
222 { 421 {
223 *enc->cur++ = uch; 422 *enc->cur++ = uch;
224 str += clen; 423 str += clen;
225 } 424 }
226 else if (is_utf8) 425 else if (is_utf8)
… …
233 while (--clen); 432 while (--clen);
234 } 433 }
235 else 434 else
236 { 435 {
237 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 436 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
238 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 437 enc->cur = encode_utf8 (enc->cur, uch);
239 ++str; 438 ++str;
240 } 439 }
241 } 440 }
242 } 441 }
243 } 442 }
244 443
245 --len; 444 --len;
246 } 445 }
247} 446}
248 447
249inline void 448INLINE void
250encode_indent (enc_t *enc) 449encode_indent (enc_t *enc)
251{ 450{
252 if (enc->flags & F_INDENT) 451 if (enc->json.flags & F_INDENT)
253 { 452 {
254 int spaces = enc->indent * INDENT_STEP; 453 int spaces = enc->indent * INDENT_STEP;
255 454
256 need (enc, spaces); 455 need (enc, spaces);
257 memset (enc->cur, ' ', spaces); 456 memset (enc->cur, ' ', spaces);
258 enc->cur += spaces; 457 enc->cur += spaces;
259 } 458 }
260} 459}
261 460
262inline void 461INLINE void
263encode_space (enc_t *enc) 462encode_space (enc_t *enc)
264{ 463{
265 need (enc, 1); 464 need (enc, 1);
266 encode_ch (enc, ' '); 465 encode_ch (enc, ' ');
267} 466}
268 467
269inline void 468INLINE void
270encode_nl (enc_t *enc) 469encode_nl (enc_t *enc)
271{ 470{
272 if (enc->flags & F_INDENT) 471 if (enc->json.flags & F_INDENT)
273 { 472 {
274 need (enc, 1); 473 need (enc, 1);
275 encode_ch (enc, '\n'); 474 encode_ch (enc, '\n');
276 } 475 }
277} 476}
278 477
279inline void 478INLINE void
280encode_comma (enc_t *enc) 479encode_comma (enc_t *enc)
281{ 480{
282 encode_ch (enc, ','); 481 encode_ch (enc, ',');
283 482
284 if (enc->flags & F_INDENT) 483 if (enc->json.flags & F_INDENT)
285 encode_nl (enc); 484 encode_nl (enc);
286 else if (enc->flags & F_SPACE_AFTER) 485 else if (enc->json.flags & F_SPACE_AFTER)
287 encode_space (enc); 486 encode_space (enc);
288} 487}
289 488
290static void encode_sv (enc_t *enc, SV *sv); 489static void encode_sv (enc_t *enc, SV *sv);
291 490
292static void 491static void
293encode_av (enc_t *enc, AV *av) 492encode_av (enc_t *enc, AV *av)
294{ 493{
295 int i, len = av_len (av); 494 int i, len = av_len (av);
296 495
297 if (enc->indent >= enc->maxdepth) 496 if (enc->indent >= enc->json.max_depth)
298 croak ("data structure too deep (hit recursion limit)"); 497 croak (ERR_NESTING_EXCEEDED);
299 498
300 encode_ch (enc, '['); encode_nl (enc); 499 encode_ch (enc, '[');
301 ++enc->indent;
302 500
501 if (len >= 0)
502 {
503 encode_nl (enc); ++enc->indent;
504
303 for (i = 0; i <= len; ++i) 505 for (i = 0; i <= len; ++i)
304 { 506 {
507 SV **svp = av_fetch (av, i, 0);
508
305 encode_indent (enc); 509 encode_indent (enc);
306 encode_sv (enc, *av_fetch (av, i, 0));
307 510
511 if (svp)
512 encode_sv (enc, *svp);
513 else
514 encode_str (enc, "null", 4, 0);
515
308 if (i < len) 516 if (i < len)
309 encode_comma (enc); 517 encode_comma (enc);
310 } 518 }
311 519
520 encode_nl (enc); --enc->indent; encode_indent (enc);
521 }
522
312 encode_nl (enc); 523 encode_ch (enc, ']');
313
314 --enc->indent;
315 encode_indent (enc); encode_ch (enc, ']');
316} 524}
317 525
318static void 526static void
319encode_he (enc_t *enc, HE *he) 527encode_hk (enc_t *enc, HE *he)
320{ 528{
321 encode_ch (enc, '"'); 529 encode_ch (enc, '"');
322 530
323 if (HeKLEN (he) == HEf_SVKEY) 531 if (HeKLEN (he) == HEf_SVKEY)
324 { 532 {
325 SV *sv = HeSVKEY (he); 533 SV *sv = HeSVKEY (he);
326 STRLEN len; 534 STRLEN len;
327 char *str; 535 char *str;
328 536
329 SvGETMAGIC (sv); 537 SvGETMAGIC (sv);
330 str = SvPV (sv, len); 538 str = SvPV (sv, len);
331 539
332 encode_str (enc, str, len, SvUTF8 (sv)); 540 encode_str (enc, str, len, SvUTF8 (sv));
333 } 541 }
334 else 542 else
335 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 543 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
336 544
337 encode_ch (enc, '"'); 545 encode_ch (enc, '"');
338 546
339 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 547 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
340 encode_ch (enc, ':'); 548 encode_ch (enc, ':');
341 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 549 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
342 encode_sv (enc, HeVAL (he));
343} 550}
344 551
345// compare hash entries, used when all keys are bytestrings 552// compare hash entries, used when all keys are bytestrings
346static int 553static int
347he_cmp_fast (const void *a_, const void *b_) 554he_cmp_fast (const void *a_, const void *b_)
… …
352 HE *b = *(HE **)b_; 559 HE *b = *(HE **)b_;
353 560
354 STRLEN la = HeKLEN (a); 561 STRLEN la = HeKLEN (a);
355 STRLEN lb = HeKLEN (b); 562 STRLEN lb = HeKLEN (b);
356 563
357 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 564 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
358 cmp = la - lb; 565 cmp = lb - la;
359 566
360 return cmp; 567 return cmp;
361} 568}
362 569
363// compare hash entries, used when some keys are sv's or utf-x 570// compare hash entries, used when some keys are sv's or utf-x
364static int 571static int
365he_cmp_slow (const void *a, const void *b) 572he_cmp_slow (const void *a, const void *b)
366{ 573{
367 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 574 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
368} 575}
369 576
370static void 577static void
371encode_hv (enc_t *enc, HV *hv) 578encode_hv (enc_t *enc, HV *hv)
372{ 579{
373 int count, i; 580 HE *he;
374 581
375 if (enc->indent >= enc->maxdepth) 582 if (enc->indent >= enc->json.max_depth)
376 croak ("data structure too deep (hit recursion limit)"); 583 croak (ERR_NESTING_EXCEEDED);
377 584
378 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 585 encode_ch (enc, '{');
379 586
380 if ((count = hv_iterinit (hv)))
381 {
382 // for canonical output we have to sort by keys first 587 // for canonical output we have to sort by keys first
383 // actually, this is mostly due to the stupid so-called 588 // actually, this is mostly due to the stupid so-called
384 // security workaround added somewhere in 5.8.x. 589 // security workaround added somewhere in 5.8.x
385 // that randomises hash orderings 590 // that randomises hash orderings
386 if (enc->flags & F_CANONICAL) 591 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
592 {
593 int count = hv_iterinit (hv);
594
595 if (SvMAGICAL (hv))
387 { 596 {
597 // need to count by iterating. could improve by dynamically building the vector below
598 // but I don't care for the speed of this special case.
599 // note also that we will run into undefined behaviour when the two iterations
600 // do not result in the same count, something I might care for in some later release.
601
602 count = 0;
603 while (hv_iternext (hv))
604 ++count;
605
606 hv_iterinit (hv);
607 }
608
609 if (count)
610 {
388 int fast = 1; 611 int i, fast = 1;
389 HE *he; 612 HE *hes_stack [STACK_HES];
390#if defined(__BORLANDC__) || defined(_MSC_VER) 613 HE **hes = hes_stack;
391 HE **hes = _alloca (count * sizeof (HE)); 614
392#else 615 // allocate larger arrays on the heap
393 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 616 if (count > STACK_HES)
394#endif 617 {
618 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
619 hes = (HE **)SvPVX (sv);
620 }
395 621
396 i = 0; 622 i = 0;
397 while ((he = hv_iternext (hv))) 623 while ((he = hv_iternext (hv)))
398 { 624 {
399 hes [i++] = he; 625 hes [i++] = he;
… …
421 647
422 FREETMPS; 648 FREETMPS;
423 LEAVE; 649 LEAVE;
424 } 650 }
425 651
426 for (i = 0; i < count; ++i) 652 encode_nl (enc); ++enc->indent;
653
654 while (count--)
427 { 655 {
428 encode_indent (enc); 656 encode_indent (enc);
657 he = hes [count];
429 encode_he (enc, hes [i]); 658 encode_hk (enc, he);
659 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
430 660
431 if (i < count - 1) 661 if (count)
432 encode_comma (enc); 662 encode_comma (enc);
433 } 663 }
434 664
435 encode_nl (enc); 665 encode_nl (enc); --enc->indent; encode_indent (enc);
436 } 666 }
667 }
437 else 668 else
669 {
670 if (hv_iterinit (hv) || SvMAGICAL (hv))
671 if ((he = hv_iternext (hv)))
438 { 672 {
439 HE *he = hv_iternext (hv); 673 encode_nl (enc); ++enc->indent;
440 674
441 for (;;) 675 for (;;)
442 { 676 {
443 encode_indent (enc); 677 encode_indent (enc);
444 encode_he (enc, he); 678 encode_hk (enc, he);
679 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
445 680
446 if (!(he = hv_iternext (hv))) 681 if (!(he = hv_iternext (hv)))
447 break; 682 break;
448 683
449 encode_comma (enc); 684 encode_comma (enc);
450 } 685 }
451 686
452 encode_nl (enc); 687 encode_nl (enc); --enc->indent; encode_indent (enc);
453 } 688 }
454 } 689 }
455 690
456 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 691 encode_ch (enc, '}');
457} 692}
458 693
459// encode objects, arrays and special \0=false and \1=true values. 694// encode objects, arrays and special \0=false and \1=true values.
460static void 695static void
461encode_rv (enc_t *enc, SV *sv) 696encode_rv (enc_t *enc, SV *sv)
462{ 697{
463 svtype svt; 698 svtype svt;
699 GV *method;
464 700
465 SvGETMAGIC (sv); 701 SvGETMAGIC (sv);
466 svt = SvTYPE (sv); 702 svt = SvTYPE (sv);
467 703
704 if (expect_false (SvOBJECT (sv)))
705 {
706 HV *boolean_stash = !JSON_SLOW || types_boolean_stash
707 ? types_boolean_stash
708 : gv_stashpv ("Types::Serialiser::Boolean", 1);
709 HV *stash = SvSTASH (sv);
710
711 if (stash == boolean_stash)
712 {
713 if (SvIV (sv))
714 encode_str (enc, "true", 4, 0);
715 else
716 encode_str (enc, "false", 5, 0);
717 }
718 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
719 {
720 int count;
721 dSP;
722
723 ENTER; SAVETMPS; PUSHMARK (SP);
724 EXTEND (SP, 2);
725 // we re-bless the reference to get overload and other niceties right
726 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
727 PUSHs (sv_json);
728
729 PUTBACK;
730 count = call_sv ((SV *)GvCV (method), G_ARRAY);
731 SPAGAIN;
732
733 // catch this surprisingly common error
734 if (SvROK (TOPs) && SvRV (TOPs) == sv)
735 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
736
737 encode_ch (enc, '(');
738 encode_ch (enc, '"');
739 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
740 encode_ch (enc, '"');
741 encode_ch (enc, ')');
742 encode_ch (enc, '[');
743
744 while (count)
745 {
746 encode_sv (enc, SP[1 - count--]);
747
748 if (count)
749 encode_ch (enc, ',');
750 }
751
752 encode_ch (enc, ']');
753
754 PUTBACK;
755
756 FREETMPS; LEAVE;
757 }
758 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
759 {
760 dSP;
761
762 ENTER; SAVETMPS; PUSHMARK (SP);
763 // we re-bless the reference to get overload and other niceties right
764 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
765
766 // calling with G_SCALAR ensures that we always get a 1 return value
767 PUTBACK;
768 call_sv ((SV *)GvCV (method), G_SCALAR);
769 SPAGAIN;
770
771 // catch this surprisingly common error
772 if (SvROK (TOPs) && SvRV (TOPs) == sv)
773 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
774
775 sv = POPs;
776 PUTBACK;
777
778 encode_sv (enc, sv);
779
780 FREETMPS; LEAVE;
781 }
782 else if (enc->json.flags & F_ALLOW_BLESSED)
783 encode_str (enc, "null", 4, 0);
784 else
785 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
786 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
787 }
468 if (svt == SVt_PVHV) 788 else if (svt == SVt_PVHV)
469 encode_hv (enc, (HV *)sv); 789 encode_hv (enc, (HV *)sv);
470 else if (svt == SVt_PVAV) 790 else if (svt == SVt_PVAV)
471 encode_av (enc, (AV *)sv); 791 encode_av (enc, (AV *)sv);
472 else if (svt < SVt_PVAV) 792 else if (svt < SVt_PVAV)
473 { 793 {
474 if (SvNIOK (sv) && SvIV (sv) == 0) 794 STRLEN len = 0;
795 char *pv = svt ? SvPV (sv, len) : 0;
796
797 if (len == 1 && *pv == '1')
798 encode_str (enc, "true", 4, 0);
799 else if (len == 1 && *pv == '0')
475 encode_str (enc, "false", 5, 0); 800 encode_str (enc, "false", 5, 0);
476 else if (SvNIOK (sv) && SvIV (sv) == 1) 801 else if (enc->json.flags & F_ALLOW_UNKNOWN)
477 encode_str (enc, "true", 4, 0); 802 encode_str (enc, "null", 4, 0);
478 else 803 else
479 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 804 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
480 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 805 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
481 } 806 }
807 else if (enc->json.flags & F_ALLOW_UNKNOWN)
808 encode_str (enc, "null", 4, 0);
482 else 809 else
483 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 810 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
484 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 811 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
485} 812}
486 813
… …
504 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 831 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
505 enc->cur += strlen (enc->cur); 832 enc->cur += strlen (enc->cur);
506 } 833 }
507 else if (SvIOKp (sv)) 834 else if (SvIOKp (sv))
508 { 835 {
509 // we assume we can always read an IV as a UV 836 // we assume we can always read an IV as a UV and vice versa
510 if (SvUV (sv) & ~(UV)0x7fff) 837 // we assume two's complement
511 { 838 // we assume no aliasing issues in the union
512 // large integer, use the (rather slow) snprintf way. 839 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
513 need (enc, sizeof (UV) * 3); 840 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
514 enc->cur +=
515 SvIsUV(sv)
516 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
517 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
518 }
519 else
520 { 841 {
521 // optimise the "small number case" 842 // optimise the "small number case"
522 // code will likely be branchless and use only a single multiplication 843 // code will likely be branchless and use only a single multiplication
844 // works for numbers up to 59074
523 I32 i = SvIV (sv); 845 I32 i = SvIVX (sv);
524 U32 u; 846 U32 u;
525 char digit, nz = 0; 847 char digit, nz = 0;
526 848
527 need (enc, 6); 849 need (enc, 6);
528 850
… …
534 856
535 // now output digit by digit, each time masking out the integer part 857 // now output digit by digit, each time masking out the integer part
536 // and multiplying by 5 while moving the decimal point one to the right, 858 // and multiplying by 5 while moving the decimal point one to the right,
537 // resulting in a net multiplication by 10. 859 // resulting in a net multiplication by 10.
538 // we always write the digit to memory but conditionally increment 860 // we always write the digit to memory but conditionally increment
539 // the pointer, to ease the usage of conditional move instructions. 861 // the pointer, to enable the use of conditional move instructions.
540 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 862 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
541 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 863 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
542 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 864 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
543 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 865 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
544 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 866 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
545 } 867 }
868 else
869 {
870 // large integer, use the (rather slow) snprintf way.
871 need (enc, IVUV_MAXCHARS);
872 enc->cur +=
873 SvIsUV(sv)
874 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
875 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
876 }
546 } 877 }
547 else if (SvROK (sv)) 878 else if (SvROK (sv))
548 encode_rv (enc, SvRV (sv)); 879 encode_rv (enc, SvRV (sv));
549 else if (!SvOK (sv)) 880 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
550 encode_str (enc, "null", 4, 0); 881 encode_str (enc, "null", 4, 0);
551 else 882 else
552 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 883 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
553 SvPV_nolen (sv), SvFLAGS (sv)); 884 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
554} 885}
555 886
556static SV * 887static SV *
557encode_json (SV *scalar, U32 flags) 888encode_json (SV *scalar, JSON *json)
558{ 889{
559 enc_t enc; 890 enc_t enc;
560 891
561 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 892 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
562 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 893 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
563 894
564 enc.flags = flags; 895 enc.json = *json;
565 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 896 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
566 enc.cur = SvPVX (enc.sv); 897 enc.cur = SvPVX (enc.sv);
567 enc.end = SvEND (enc.sv); 898 enc.end = SvEND (enc.sv);
568 enc.indent = 0; 899 enc.indent = 0;
569 enc.maxdepth = DEC_DEPTH (flags); 900 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
901 : enc.json.flags & F_LATIN1 ? 0x000100UL
902 : 0x110000UL;
570 903
571 SvPOK_only (enc.sv); 904 SvPOK_only (enc.sv);
572 encode_sv (&enc, scalar); 905 encode_sv (&enc, scalar);
906 encode_nl (&enc);
573 907
574 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 908 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
575 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 909 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
576 910
577 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 911 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
578 SvUTF8_on (enc.sv); 912 SvUTF8_on (enc.sv);
579 913
580 if (enc.flags & F_SHRINK) 914 if (enc.json.flags & F_SHRINK)
581 shrink (enc.sv); 915 shrink (enc.sv);
582 916
583 return enc.sv; 917 return enc.sv;
584} 918}
585 919
… …
590typedef struct 924typedef struct
591{ 925{
592 char *cur; // current parser pointer 926 char *cur; // current parser pointer
593 char *end; // end of input string 927 char *end; // end of input string
594 const char *err; // parse error, if != 0 928 const char *err; // parse error, if != 0
595 U32 flags; // F_* 929 JSON json;
596 U32 depth; // recursion depth 930 U32 depth; // recursion depth
597 U32 maxdepth; // recursion depth limit 931 U32 maxdepth; // recursion depth limit
598} dec_t; 932} dec_t;
599 933
600inline void 934INLINE void
935decode_comment (dec_t *dec)
936{
937 // only '#'-style comments allowed a.t.m.
938
939 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
940 ++dec->cur;
941}
942
943INLINE void
601decode_ws (dec_t *dec) 944decode_ws (dec_t *dec)
602{ 945{
603 for (;;) 946 for (;;)
604 { 947 {
605 char ch = *dec->cur; 948 char ch = *dec->cur;
606 949
607 if (ch > 0x20 950 if (ch > 0x20)
951 {
952 if (expect_false (ch == '#'))
953 {
954 if (dec->json.flags & F_RELAXED)
955 decode_comment (dec);
956 else
957 break;
958 }
959 else
960 break;
961 }
608 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 962 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
609 break; 963 break; // parse error, but let higher level handle it, gives better error messages
610 964
611 ++dec->cur; 965 ++dec->cur;
612 } 966 }
613} 967}
614 968
… …
618 if (*dec->cur != ch) \ 972 if (*dec->cur != ch) \
619 ERR (# ch " expected"); \ 973 ERR (# ch " expected"); \
620 ++dec->cur; \ 974 ++dec->cur; \
621 SE 975 SE
622 976
623#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 977#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
624#define DEC_DEC_DEPTH --dec->depth 978#define DEC_DEC_DEPTH --dec->depth
625 979
626static SV *decode_sv (dec_t *dec); 980static SV *decode_sv (dec_t *dec);
627 981
628static signed char decode_hexdigit[256]; 982static signed char decode_hexdigit[256];
… …
631decode_4hex (dec_t *dec) 985decode_4hex (dec_t *dec)
632{ 986{
633 signed char d1, d2, d3, d4; 987 signed char d1, d2, d3, d4;
634 unsigned char *cur = (unsigned char *)dec->cur; 988 unsigned char *cur = (unsigned char *)dec->cur;
635 989
636 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected"); 990 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
637 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected"); 991 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
638 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected"); 992 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
639 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected"); 993 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
640 994
641 dec->cur += 4; 995 dec->cur += 4;
642 996
643 return ((UV)d1) << 12 997 return ((UV)d1) << 12
644 | ((UV)d2) << 8 998 | ((UV)d2) << 8
… …
720 1074
721 if (hi >= 0x80) 1075 if (hi >= 0x80)
722 { 1076 {
723 utf8 = 1; 1077 utf8 = 1;
724 1078
725 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1079 cur = encode_utf8 (cur, hi);
726 } 1080 }
727 else 1081 else
728 *cur++ = hi; 1082 *cur++ = hi;
729 } 1083 }
730 break; 1084 break;
… …
732 default: 1086 default:
733 --dec_cur; 1087 --dec_cur;
734 ERR ("illegal backslash escape sequence in string"); 1088 ERR ("illegal backslash escape sequence in string");
735 } 1089 }
736 } 1090 }
737 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1091 else if (expect_true (ch >= 0x20 && ch < 0x80))
738 *cur++ = ch; 1092 *cur++ = ch;
739 else if (ch >= 0x80) 1093 else if (ch >= 0x80)
740 { 1094 {
741 STRLEN clen; 1095 STRLEN clen;
742 UV uch;
743 1096
744 --dec_cur; 1097 --dec_cur;
745 1098
746 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1099 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
747 if (clen == (STRLEN)-1) 1100 if (clen == (STRLEN)-1)
748 ERR ("malformed UTF-8 character in JSON string"); 1101 ERR ("malformed UTF-8 character in JSON string");
749 1102
750 do 1103 do
751 *cur++ = *dec_cur++; 1104 *cur++ = *dec_cur++;
… …
768 { 1121 {
769 STRLEN len = cur - buf; 1122 STRLEN len = cur - buf;
770 1123
771 if (sv) 1124 if (sv)
772 { 1125 {
773 SvGROW (sv, SvCUR (sv) + len + 1); 1126 STRLEN cur = SvCUR (sv);
1127
1128 if (SvLEN (sv) <= cur + len)
1129 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1130
774 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1131 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
775 SvCUR_set (sv, SvCUR (sv) + len); 1132 SvCUR_set (sv, SvCUR (sv) + len);
776 } 1133 }
777 else 1134 else
778 sv = newSVpvn (buf, len); 1135 sv = newSVpvn (buf, len);
… …
863 is_nv = 1; 1220 is_nv = 1;
864 } 1221 }
865 1222
866 if (!is_nv) 1223 if (!is_nv)
867 { 1224 {
1225 int len = dec->cur - start;
1226
868 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1227 // special case the rather common 1..5-digit-int case
869 if (*start == '-') 1228 if (*start == '-')
870 switch (dec->cur - start) 1229 switch (len)
871 { 1230 {
872 case 2: return newSViv (-( start [1] - '0' )); 1231 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
873 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1232 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
874 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1233 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
875 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1234 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1235 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
876 } 1236 }
877 else 1237 else
878 switch (dec->cur - start) 1238 switch (len)
879 { 1239 {
880 case 1: return newSViv ( start [0] - '0' ); 1240 case 1: return newSViv ( start [0] - '0' * 1);
881 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1241 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
882 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1242 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
883 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1243 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1244 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
884 } 1245 }
885 1246
886 { 1247 {
887 UV uv; 1248 UV uv;
888 int numtype = grok_number (start, dec->cur - start, &uv); 1249 int numtype = grok_number (start, len, &uv);
889 if (numtype & IS_NUMBER_IN_UV) 1250 if (numtype & IS_NUMBER_IN_UV)
890 if (numtype & IS_NUMBER_NEG) 1251 if (numtype & IS_NUMBER_NEG)
891 { 1252 {
892 if (uv < (UV)IV_MIN) 1253 if (uv < (UV)IV_MIN)
893 return newSViv (-(IV)uv); 1254 return newSViv (-(IV)uv);
894 } 1255 }
895 else 1256 else
896 return newSVuv (uv); 1257 return newSVuv (uv);
897 } 1258 }
898 }
899 1259
1260 len -= *start == '-' ? 1 : 0;
1261
1262 // does not fit into IV or UV, try NV
1263 if (len <= NV_DIG)
1264 // fits into NV without loss of precision
1265 return newSVnv (json_atof (start));
1266
1267 // everything else fails, convert it to a string
1268 return newSVpvn (start, dec->cur - start);
1269 }
1270
1271 // loss of precision here
900 return newSVnv (Atof (start)); 1272 return newSVnv (json_atof (start));
901 1273
902fail: 1274fail:
903 return 0; 1275 return 0;
904} 1276}
905 1277
… …
929 if (*dec->cur == ']') 1301 if (*dec->cur == ']')
930 { 1302 {
931 ++dec->cur; 1303 ++dec->cur;
932 break; 1304 break;
933 } 1305 }
934 1306
935 if (*dec->cur != ',') 1307 if (*dec->cur != ',')
936 ERR (", or ] expected while parsing array"); 1308 ERR (", or ] expected while parsing array");
937 1309
938 ++dec->cur; 1310 ++dec->cur;
1311
1312 decode_ws (dec);
1313
1314 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1315 {
1316 ++dec->cur;
1317 break;
1318 }
939 } 1319 }
940 1320
941 DEC_DEC_DEPTH; 1321 DEC_DEC_DEPTH;
942 return newRV_noinc ((SV *)av); 1322 return newRV_noinc ((SV *)av);
943 1323
… …
948} 1328}
949 1329
950static SV * 1330static SV *
951decode_hv (dec_t *dec) 1331decode_hv (dec_t *dec)
952{ 1332{
1333 SV *sv;
953 HV *hv = newHV (); 1334 HV *hv = newHV ();
954 1335
955 DEC_INC_DEPTH; 1336 DEC_INC_DEPTH;
956 decode_ws (dec); 1337 decode_ws (dec);
957 1338
958 if (*dec->cur == '}') 1339 if (*dec->cur == '}')
959 ++dec->cur; 1340 ++dec->cur;
960 else 1341 else
961 for (;;) 1342 for (;;)
962 { 1343 {
1344 EXPECT_CH ('"');
1345
1346 // heuristic: assume that
1347 // a) decode_str + hv_store_ent are abysmally slow.
1348 // b) most hash keys are short, simple ascii text.
1349 // => try to "fast-match" such strings to avoid
1350 // the overhead of decode_str + hv_store_ent.
1351 {
963 SV *key, *value; 1352 SV *value;
1353 char *p = dec->cur;
1354 char *e = p + 24; // only try up to 24 bytes
964 1355
965 decode_ws (dec); EXPECT_CH ('"'); 1356 for (;;)
966
967 key = decode_str (dec);
968 if (!key)
969 goto fail;
970
971 decode_ws (dec); EXPECT_CH (':');
972
973 value = decode_sv (dec);
974 if (!value)
975 { 1357 {
1358 // the >= 0x80 is false on most architectures
1359 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1360 {
1361 // slow path, back up and use decode_str
1362 SV *key = decode_str (dec);
1363 if (!key)
1364 goto fail;
1365
1366 decode_ws (dec); EXPECT_CH (':');
1367
1368 decode_ws (dec);
1369 value = decode_sv (dec);
1370 if (!value)
1371 {
1372 SvREFCNT_dec (key);
1373 goto fail;
1374 }
1375
1376 hv_store_ent (hv, key, value, 0);
976 SvREFCNT_dec (key); 1377 SvREFCNT_dec (key);
1378
1379 break;
1380 }
1381 else if (*p == '"')
1382 {
1383 // fast path, got a simple key
1384 char *key = dec->cur;
1385 int len = p - key;
1386 dec->cur = p + 1;
1387
1388 decode_ws (dec); EXPECT_CH (':');
1389
1390 decode_ws (dec);
1391 value = decode_sv (dec);
1392 if (!value)
977 goto fail; 1393 goto fail;
1394
1395 hv_store (hv, key, len, value, 0);
1396
1397 break;
1398 }
1399
1400 ++p;
978 } 1401 }
979 1402 }
980 hv_store_ent (hv, key, value, 0);
981 SvREFCNT_dec (key);
982 1403
983 decode_ws (dec); 1404 decode_ws (dec);
984 1405
985 if (*dec->cur == '}') 1406 if (*dec->cur == '}')
986 { 1407 {
… …
990 1411
991 if (*dec->cur != ',') 1412 if (*dec->cur != ',')
992 ERR (", or } expected while parsing object/hash"); 1413 ERR (", or } expected while parsing object/hash");
993 1414
994 ++dec->cur; 1415 ++dec->cur;
1416
1417 decode_ws (dec);
1418
1419 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1420 {
1421 ++dec->cur;
1422 break;
1423 }
995 } 1424 }
996 1425
997 DEC_DEC_DEPTH; 1426 DEC_DEC_DEPTH;
998 return newRV_noinc ((SV *)hv); 1427 sv = newRV_noinc ((SV *)hv);
1428
1429 // check filter callbacks
1430 if (dec->json.flags & F_HOOK)
1431 {
1432 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1433 {
1434 HE *cb, *he;
1435
1436 hv_iterinit (hv);
1437 he = hv_iternext (hv);
1438 hv_iterinit (hv);
1439
1440 // the next line creates a mortal sv each time its called.
1441 // might want to optimise this for common cases.
1442 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1443
1444 if (cb)
1445 {
1446 dSP;
1447 int count;
1448
1449 ENTER; SAVETMPS; PUSHMARK (SP);
1450 XPUSHs (HeVAL (he));
1451 sv_2mortal (sv);
1452
1453 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1454
1455 if (count == 1)
1456 {
1457 sv = newSVsv (POPs);
1458 FREETMPS; LEAVE;
1459 return sv;
1460 }
1461
1462 SvREFCNT_inc (sv);
1463 FREETMPS; LEAVE;
1464 }
1465 }
1466
1467 if (dec->json.cb_object)
1468 {
1469 dSP;
1470 int count;
1471
1472 ENTER; SAVETMPS; PUSHMARK (SP);
1473 XPUSHs (sv_2mortal (sv));
1474
1475 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1476
1477 if (count == 1)
1478 {
1479 sv = newSVsv (POPs);
1480 FREETMPS; LEAVE;
1481 return sv;
1482 }
1483
1484 SvREFCNT_inc (sv);
1485 FREETMPS; LEAVE;
1486 }
1487 }
1488
1489 return sv;
999 1490
1000fail: 1491fail:
1001 SvREFCNT_dec (hv); 1492 SvREFCNT_dec (hv);
1002 DEC_DEC_DEPTH; 1493 DEC_DEC_DEPTH;
1003 return 0; 1494 return 0;
1004} 1495}
1005 1496
1006static SV * 1497static SV *
1498decode_tag (dec_t *dec)
1499{
1500 SV *tag = 0;
1501 SV *val = 0;
1502
1503 if (!(dec->json.flags & F_ALLOW_TAGS))
1504 ERR ("malformed JSON string, neither array, object, number, string or atom");
1505
1506 ++dec->cur;
1507
1508 decode_ws (dec);
1509
1510 tag = decode_sv (dec);
1511 if (!tag)
1512 goto fail;
1513
1514 if (!SvPOK (tag))
1515 ERR ("malformed JSON string, (tag) must be a string");
1516
1517 decode_ws (dec);
1518
1519 if (*dec->cur != ')')
1520 ERR (") expected after tag");
1521
1522 ++dec->cur;
1523
1524 decode_ws (dec);
1525
1526 val = decode_sv (dec);
1527 if (!val)
1528 goto fail;
1529
1530 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1531 ERR ("malformed JSON string, tag value must be an array");
1532
1533 {
1534 AV *av = (AV *)SvRV (val);
1535 int i, len = av_len (av) + 1;
1536 HV *stash = gv_stashsv (tag, 0);
1537 SV *sv;
1538
1539 if (!stash)
1540 ERR ("cannot decode perl-object (package does not exist)");
1541
1542 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1543
1544 if (!method)
1545 ERR ("cannot decode perl-object (package does not have a THAW method)");
1546
1547 dSP;
1548
1549 ENTER; SAVETMPS; PUSHMARK (SP);
1550 EXTEND (SP, len + 2);
1551 // we re-bless the reference to get overload and other niceties right
1552 PUSHs (tag);
1553 PUSHs (sv_json);
1554
1555 for (i = 0; i < len; ++i)
1556 PUSHs (*av_fetch (av, i, 1));
1557
1558 PUTBACK;
1559 call_sv ((SV *)GvCV (method), G_SCALAR);
1560 SPAGAIN;
1561
1562 SvREFCNT_dec (tag);
1563 SvREFCNT_dec (val);
1564 sv = SvREFCNT_inc (POPs);
1565
1566 PUTBACK;
1567
1568 FREETMPS; LEAVE;
1569
1570 return sv;
1571 }
1572
1573fail:
1574 SvREFCNT_dec (tag);
1575 SvREFCNT_dec (val);
1576 return 0;
1577}
1578
1579static SV *
1007decode_sv (dec_t *dec) 1580decode_sv (dec_t *dec)
1008{ 1581{
1009 decode_ws (dec); 1582 // the beauty of JSON: you need exactly one character lookahead
1583 // to parse everything.
1010 switch (*dec->cur) 1584 switch (*dec->cur)
1011 { 1585 {
1012 case '"': ++dec->cur; return decode_str (dec); 1586 case '"': ++dec->cur; return decode_str (dec);
1013 case '[': ++dec->cur; return decode_av (dec); 1587 case '[': ++dec->cur; return decode_av (dec);
1014 case '{': ++dec->cur; return decode_hv (dec); 1588 case '{': ++dec->cur; return decode_hv (dec);
1589 case '(': return decode_tag (dec);
1015 1590
1016 case '-': 1591 case '-':
1017 case '0': case '1': case '2': case '3': case '4': 1592 case '0': case '1': case '2': case '3': case '4':
1018 case '5': case '6': case '7': case '8': case '9': 1593 case '5': case '6': case '7': case '8': case '9':
1019 return decode_num (dec); 1594 return decode_num (dec);
1020 1595
1021 case 't': 1596 case 't':
1022 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1597 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1023 { 1598 {
1024 dec->cur += 4; 1599 dec->cur += 4;
1600#if JSON_SLOW
1601 types_true = get_bool ("Types::Serialiser::true");
1602#endif
1025 return newSViv (1); 1603 return newSVsv (types_true);
1026 } 1604 }
1027 else 1605 else
1028 ERR ("'true' expected"); 1606 ERR ("'true' expected");
1029 1607
1030 break; 1608 break;
1031 1609
1032 case 'f': 1610 case 'f':
1033 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1611 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1034 { 1612 {
1035 dec->cur += 5; 1613 dec->cur += 5;
1614#if JSON_SLOW
1615 types_false = get_bool ("Types::Serialiser::false");
1616#endif
1036 return newSViv (0); 1617 return newSVsv (types_false);
1037 } 1618 }
1038 else 1619 else
1039 ERR ("'false' expected"); 1620 ERR ("'false' expected");
1040 1621
1041 break; 1622 break;
… …
1050 ERR ("'null' expected"); 1631 ERR ("'null' expected");
1051 1632
1052 break; 1633 break;
1053 1634
1054 default: 1635 default:
1055 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1636 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1056 break; 1637 break;
1057 } 1638 }
1058 1639
1059fail: 1640fail:
1060 return 0; 1641 return 0;
1061} 1642}
1062 1643
1063static SV * 1644static SV *
1064decode_json (SV *string, U32 flags, UV *offset_return) 1645decode_json (SV *string, JSON *json, char **offset_return)
1065{ 1646{
1066 dec_t dec; 1647 dec_t dec;
1067 UV offset;
1068 SV *sv; 1648 SV *sv;
1069 1649
1650 /* work around bugs in 5.10 where manipulating magic values
1651 * makes perl ignore the magic in subsequent accesses.
1652 * also make a copy of non-PV values, to get them into a clean
1653 * state (SvPV should do that, but it's buggy, see below).
1654 */
1070 SvGETMAGIC (string); 1655 /*SvGETMAGIC (string);*/
1656 if (SvMAGICAL (string) || !SvPOK (string))
1657 string = sv_2mortal (newSVsv (string));
1658
1071 SvUPGRADE (string, SVt_PV); 1659 SvUPGRADE (string, SVt_PV);
1072 1660
1073 if (flags & F_UTF8) 1661 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1662 * assertion with -DDEBUGGING, although SvCUR is documented to
1663 * return the xpv_cur field which certainly exists after upgrading.
1664 * according to nicholas clark, calling SvPOK fixes this.
1665 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1666 * and hope for the best.
1667 * Damnit, SvPV_nolen still trips over yet another assertion. This
1668 * assertion business is seriously broken, try yet another workaround
1669 * for the broken -DDEBUGGING.
1670 */
1671 {
1672#ifdef DEBUGGING
1673 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1674#else
1675 STRLEN offset = SvCUR (string);
1676#endif
1677
1678 if (offset > json->max_size && json->max_size)
1679 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1680 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1681 }
1682
1683 if (DECODE_WANTS_OCTETS (json))
1074 sv_utf8_downgrade (string, 0); 1684 sv_utf8_downgrade (string, 0);
1075 else 1685 else
1076 sv_utf8_upgrade (string); 1686 sv_utf8_upgrade (string);
1077 1687
1078 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1688 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1079 1689
1080 dec.flags = flags; 1690 dec.json = *json;
1081 dec.cur = SvPVX (string); 1691 dec.cur = SvPVX (string);
1082 dec.end = SvEND (string); 1692 dec.end = SvEND (string);
1083 dec.err = 0; 1693 dec.err = 0;
1084 dec.depth = 0; 1694 dec.depth = 0;
1085 dec.maxdepth = DEC_DEPTH (dec.flags); 1695
1696 if (dec.json.cb_object || dec.json.cb_sk_object)
1697 dec.json.flags |= F_HOOK;
1086 1698
1087 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1699 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1700
1701 decode_ws (&dec);
1088 sv = decode_sv (&dec); 1702 sv = decode_sv (&dec);
1703
1704 if (offset_return)
1705 *offset_return = dec.cur;
1089 1706
1090 if (!(offset_return || !sv)) 1707 if (!(offset_return || !sv))
1091 { 1708 {
1092 // check for trailing garbage 1709 // check for trailing garbage
1093 decode_ws (&dec); 1710 decode_ws (&dec);
… …
1096 { 1713 {
1097 dec.err = "garbage after JSON object"; 1714 dec.err = "garbage after JSON object";
1098 SvREFCNT_dec (sv); 1715 SvREFCNT_dec (sv);
1099 sv = 0; 1716 sv = 0;
1100 } 1717 }
1101 }
1102
1103 if (offset_return || !sv)
1104 {
1105 offset = dec.flags & F_UTF8
1106 ? dec.cur - SvPVX (string)
1107 : utf8_distance (dec.cur, SvPVX (string));
1108
1109 if (offset_return)
1110 *offset_return = offset;
1111 } 1718 }
1112 1719
1113 if (!sv) 1720 if (!sv)
1114 { 1721 {
1115 SV *uni = sv_newmortal (); 1722 SV *uni = sv_newmortal ();
… …
1121 SAVEVPTR (PL_curcop); 1728 SAVEVPTR (PL_curcop);
1122 PL_curcop = &cop; 1729 PL_curcop = &cop;
1123 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1730 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1124 LEAVE; 1731 LEAVE;
1125 1732
1126 croak ("%s, at character offset %d [\"%s\"]", 1733 croak ("%s, at character offset %d (before \"%s\")",
1127 dec.err, 1734 dec.err,
1128 (int)offset, 1735 (int)ptr_to_index (string, dec.cur),
1129 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1736 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1130 } 1737 }
1131 1738
1132 sv = sv_2mortal (sv); 1739 sv = sv_2mortal (sv);
1133 1740
1134 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1741 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
1135 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1742 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1136 1743
1137 return sv; 1744 return sv;
1745}
1746
1747/////////////////////////////////////////////////////////////////////////////
1748// incremental parser
1749
1750static void
1751incr_parse (JSON *self)
1752{
1753 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1754
1755 // the state machine here is a bit convoluted and could be simplified a lot
1756 // but this would make it slower, so...
1757
1758 for (;;)
1759 {
1760 //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
1761 switch (self->incr_mode)
1762 {
1763 // only used for initial whitespace skipping
1764 case INCR_M_WS:
1765 for (;;)
1766 {
1767 if (*p > 0x20)
1768 {
1769 if (*p == '#')
1770 {
1771 self->incr_mode = INCR_M_C0;
1772 goto incr_m_c;
1773 }
1774 else
1775 {
1776 self->incr_mode = INCR_M_JSON;
1777 goto incr_m_json;
1778 }
1779 }
1780 else if (!*p)
1781 goto interrupt;
1782
1783 ++p;
1784 }
1785
1786 // skip a single char inside a string (for \\-processing)
1787 case INCR_M_BS:
1788 if (!*p)
1789 goto interrupt;
1790
1791 ++p;
1792 self->incr_mode = INCR_M_STR;
1793 goto incr_m_str;
1794
1795 // inside #-style comments
1796 case INCR_M_C0:
1797 case INCR_M_C1:
1798 incr_m_c:
1799 for (;;)
1800 {
1801 if (*p == '\n')
1802 {
1803 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1804 break;
1805 }
1806 else if (!*p)
1807 goto interrupt;
1808
1809 ++p;
1810 }
1811
1812 break;
1813
1814 // inside a string
1815 case INCR_M_STR:
1816 incr_m_str:
1817 for (;;)
1818 {
1819 if (*p == '"')
1820 {
1821 ++p;
1822 self->incr_mode = INCR_M_JSON;
1823
1824 if (!self->incr_nest)
1825 goto interrupt;
1826
1827 goto incr_m_json;
1828 }
1829 else if (*p == '\\')
1830 {
1831 ++p; // "virtually" consumes character after \
1832
1833 if (!*p) // if at end of string we have to switch modes
1834 {
1835 self->incr_mode = INCR_M_BS;
1836 goto interrupt;
1837 }
1838 }
1839 else if (!*p)
1840 goto interrupt;
1841
1842 ++p;
1843 }
1844
1845 // after initial ws, outside string
1846 case INCR_M_JSON:
1847 incr_m_json:
1848 for (;;)
1849 {
1850 switch (*p++)
1851 {
1852 case 0:
1853 --p;
1854 goto interrupt;
1855
1856 case 0x09:
1857 case 0x0a:
1858 case 0x0d:
1859 case 0x20:
1860 if (!self->incr_nest)
1861 {
1862 --p; // do not eat the whitespace, let the next round do it
1863 goto interrupt;
1864 }
1865 break;
1866
1867 case '"':
1868 self->incr_mode = INCR_M_STR;
1869 goto incr_m_str;
1870
1871 case '[':
1872 case '{':
1873 case '(':
1874 if (++self->incr_nest > self->max_depth)
1875 croak (ERR_NESTING_EXCEEDED);
1876 break;
1877
1878 case ']':
1879 case '}':
1880 if (--self->incr_nest <= 0)
1881 goto interrupt;
1882 break;
1883
1884 case ')':
1885 --self->incr_nest;
1886 break;
1887
1888 case '#':
1889 self->incr_mode = INCR_M_C1;
1890 goto incr_m_c;
1891 }
1892 }
1893 }
1894
1895 modechange:
1896 ;
1897 }
1898
1899interrupt:
1900 self->incr_pos = p - SvPVX (self->incr_text);
1901 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1902 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1138} 1903}
1139 1904
1140///////////////////////////////////////////////////////////////////////////// 1905/////////////////////////////////////////////////////////////////////////////
1141// XS interface functions 1906// XS interface functions
1142 1907
1143MODULE = JSON::XS PACKAGE = JSON::XS 1908MODULE = JSON::XS PACKAGE = JSON::XS
1144 1909
1145BOOT: 1910BOOT:
1146{ 1911{
1147 int i; 1912 int i;
1148
1149 memset (decode_hexdigit, 0xff, 256);
1150 1913
1151 for (i = 0; i < 256; ++i) 1914 for (i = 0; i < 256; ++i)
1152 decode_hexdigit [i] = 1915 decode_hexdigit [i] =
1153 i >= '0' && i <= '9' ? i - '0' 1916 i >= '0' && i <= '9' ? i - '0'
1154 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1917 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1155 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1918 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1156 : -1; 1919 : -1;
1157 1920
1158 json_stash = gv_stashpv ("JSON::XS", 1); 1921 json_stash = gv_stashpv ("JSON::XS" , 1);
1922 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1923
1924 types_true = get_bool ("Types::Serialiser::true");
1925 types_false = get_bool ("Types::Serialiser::false");
1926
1927 sv_json = newSVpv ("JSON", 0);
1928 SvREADONLY_on (sv_json);
1929
1930 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1159} 1931}
1160 1932
1161PROTOTYPES: DISABLE 1933PROTOTYPES: DISABLE
1162 1934
1163SV *new (char *dummy) 1935void CLONE (...)
1164 CODE: 1936 CODE:
1165 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1937 json_stash = 0;
1938 types_boolean_stash = 0;
1939
1940void new (char *klass)
1941 PPCODE:
1942{
1943 SV *pv = NEWSV (0, sizeof (JSON));
1944 SvPOK_only (pv);
1945 json_init ((JSON *)SvPVX (pv));
1946 XPUSHs (sv_2mortal (sv_bless (
1947 newRV_noinc (pv),
1948 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1949 )));
1950}
1951
1952void ascii (JSON *self, int enable = 1)
1953 ALIAS:
1954 ascii = F_ASCII
1955 latin1 = F_LATIN1
1956 utf8 = F_UTF8
1957 indent = F_INDENT
1958 canonical = F_CANONICAL
1959 space_before = F_SPACE_BEFORE
1960 space_after = F_SPACE_AFTER
1961 pretty = F_PRETTY
1962 allow_nonref = F_ALLOW_NONREF
1963 shrink = F_SHRINK
1964 allow_blessed = F_ALLOW_BLESSED
1965 convert_blessed = F_CONV_BLESSED
1966 relaxed = F_RELAXED
1967 allow_unknown = F_ALLOW_UNKNOWN
1968 allow_tags = F_ALLOW_TAGS
1969 PPCODE:
1970{
1971 if (enable)
1972 self->flags |= ix;
1973 else
1974 self->flags &= ~ix;
1975
1976 XPUSHs (ST (0));
1977}
1978
1979void get_ascii (JSON *self)
1980 ALIAS:
1981 get_ascii = F_ASCII
1982 get_latin1 = F_LATIN1
1983 get_utf8 = F_UTF8
1984 get_indent = F_INDENT
1985 get_canonical = F_CANONICAL
1986 get_space_before = F_SPACE_BEFORE
1987 get_space_after = F_SPACE_AFTER
1988 get_allow_nonref = F_ALLOW_NONREF
1989 get_shrink = F_SHRINK
1990 get_allow_blessed = F_ALLOW_BLESSED
1991 get_convert_blessed = F_CONV_BLESSED
1992 get_relaxed = F_RELAXED
1993 get_allow_unknown = F_ALLOW_UNKNOWN
1994 get_allow_tags = F_ALLOW_TAGS
1995 PPCODE:
1996 XPUSHs (boolSV (self->flags & ix));
1997
1998void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1999 PPCODE:
2000 self->max_depth = max_depth;
2001 XPUSHs (ST (0));
2002
2003U32 get_max_depth (JSON *self)
2004 CODE:
2005 RETVAL = self->max_depth;
1166 OUTPUT: 2006 OUTPUT:
1167 RETVAL 2007 RETVAL
1168 2008
1169SV *ascii (SV *self, int enable = 1) 2009void max_size (JSON *self, U32 max_size = 0)
1170 ALIAS: 2010 PPCODE:
1171 ascii = F_ASCII 2011 self->max_size = max_size;
1172 latin1 = F_LATIN1 2012 XPUSHs (ST (0));
1173 utf8 = F_UTF8 2013
1174 indent = F_INDENT 2014int get_max_size (JSON *self)
1175 canonical = F_CANONICAL
1176 space_before = F_SPACE_BEFORE
1177 space_after = F_SPACE_AFTER
1178 pretty = F_PRETTY
1179 allow_nonref = F_ALLOW_NONREF
1180 shrink = F_SHRINK
1181 CODE: 2015 CODE:
1182{ 2016 RETVAL = self->max_size;
1183 UV *uv = SvJSON (self);
1184 if (enable)
1185 *uv |= ix;
1186 else
1187 *uv &= ~ix;
1188
1189 RETVAL = newSVsv (self);
1190}
1191 OUTPUT: 2017 OUTPUT:
1192 RETVAL 2018 RETVAL
1193 2019
1194SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 2020void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
2021 PPCODE:
2022{
2023 SvREFCNT_dec (self->cb_object);
2024 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
2025
2026 XPUSHs (ST (0));
2027}
2028
2029void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
2030 PPCODE:
2031{
2032 if (!self->cb_sk_object)
2033 self->cb_sk_object = newHV ();
2034
2035 if (SvOK (cb))
2036 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
2037 else
2038 {
2039 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
2040
2041 if (!HvKEYS (self->cb_sk_object))
2042 {
2043 SvREFCNT_dec (self->cb_sk_object);
2044 self->cb_sk_object = 0;
2045 }
2046 }
2047
2048 XPUSHs (ST (0));
2049}
2050
2051void encode (JSON *self, SV *scalar)
2052 PPCODE:
2053 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2054 XPUSHs (scalar);
2055
2056void decode (JSON *self, SV *jsonstr)
2057 PPCODE:
2058 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2059 XPUSHs (jsonstr);
2060
2061void decode_prefix (JSON *self, SV *jsonstr)
2062 PPCODE:
2063{
2064 SV *sv;
2065 char *offset;
2066 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
2067 EXTEND (SP, 2);
2068 PUSHs (sv);
2069 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
2070}
2071
2072void incr_parse (JSON *self, SV *jsonstr = 0)
2073 PPCODE:
2074{
2075 if (!self->incr_text)
2076 self->incr_text = newSVpvn ("", 0);
2077
2078 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2079 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2080 if (DECODE_WANTS_OCTETS (self))
2081 {
2082 if (self->incr_pos)
2083 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2084 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2085
2086 sv_utf8_downgrade (self->incr_text, 0);
2087 }
2088 else
2089 {
2090 sv_utf8_upgrade (self->incr_text);
2091
2092 if (self->incr_pos)
2093 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2094 - (U8 *)SvPVX (self->incr_text);
2095 }
2096
2097 // append data, if any
2098 if (jsonstr)
2099 {
2100 /* make sure both strings have same encoding */
2101 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2102 if (SvUTF8 (jsonstr))
2103 sv_utf8_downgrade (jsonstr, 0);
2104 else
2105 sv_utf8_upgrade (jsonstr);
2106
2107 /* and then just blindly append */
2108 {
2109 STRLEN len;
2110 const char *str = SvPV (jsonstr, len);
2111 STRLEN cur = SvCUR (self->incr_text);
2112
2113 if (SvLEN (self->incr_text) <= cur + len)
2114 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
2115
2116 Move (str, SvEND (self->incr_text), len, char);
2117 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2118 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2119 }
2120 }
2121
2122 if (GIMME_V != G_VOID)
2123 do
2124 {
2125 SV *sv;
2126 char *offset;
2127
2128 if (!INCR_DONE (self))
2129 {
2130 incr_parse (self);
2131
2132 if (self->incr_pos > self->max_size && self->max_size)
2133 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2134 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2135
2136 if (!INCR_DONE (self))
2137 {
2138 // as an optimisation, do not accumulate white space in the incr buffer
2139 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2140 {
2141 self->incr_pos = 0;
2142 SvCUR_set (self->incr_text, 0);
2143 }
2144
2145 break;
2146 }
2147 }
2148
2149 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2150 XPUSHs (sv);
2151
2152 self->incr_pos -= offset - SvPVX (self->incr_text);
2153 self->incr_nest = 0;
2154 self->incr_mode = 0;
2155
2156 sv_chop (self->incr_text, offset);
2157 }
2158 while (GIMME_V == G_ARRAY);
2159}
2160
2161SV *incr_text (JSON *self)
2162 ATTRS: lvalue
1195 CODE: 2163 CODE:
1196{ 2164{
1197 UV *uv = SvJSON (self); 2165 if (self->incr_pos)
1198 UV log2 = 0; 2166 croak ("incr_text can not be called when the incremental parser already started parsing");
1199 2167
1200 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 2168 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1201
1202 while ((1UL << log2) < max_depth)
1203 ++log2;
1204
1205 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1206
1207 RETVAL = newSVsv (self);
1208} 2169}
1209 OUTPUT: 2170 OUTPUT:
1210 RETVAL 2171 RETVAL
1211 2172
2173void incr_skip (JSON *self)
2174 CODE:
2175{
2176 if (self->incr_pos)
2177 {
2178 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2179 self->incr_pos = 0;
2180 self->incr_nest = 0;
2181 self->incr_mode = 0;
2182 }
2183}
2184
2185void incr_reset (JSON *self)
2186 CODE:
2187{
2188 SvREFCNT_dec (self->incr_text);
2189 self->incr_text = 0;
2190 self->incr_pos = 0;
2191 self->incr_nest = 0;
2192 self->incr_mode = 0;
2193}
2194
2195void DESTROY (JSON *self)
2196 CODE:
2197 SvREFCNT_dec (self->cb_sk_object);
2198 SvREFCNT_dec (self->cb_object);
2199 SvREFCNT_dec (self->incr_text);
2200
2201PROTOTYPES: ENABLE
2202
1212void encode (SV *self, SV *scalar) 2203void encode_json (SV *scalar)
1213 PPCODE: 2204 PPCODE:
1214 XPUSHs (encode_json (scalar, *SvJSON (self))); 2205{
2206 JSON json;
2207 json_init (&json);
2208 json.flags |= F_UTF8;
2209 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2210 XPUSHs (scalar);
2211}
1215 2212
1216void decode (SV *self, SV *jsonstr) 2213void decode_json (SV *jsonstr)
1217 PPCODE: 2214 PPCODE:
1218 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1219
1220void decode_prefix (SV *self, SV *jsonstr)
1221 PPCODE:
1222{ 2215{
1223 UV offset; 2216 JSON json;
1224 EXTEND (SP, 2); 2217 json_init (&json);
1225 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 2218 json.flags |= F_UTF8;
1226 PUSHs (sv_2mortal (newSVuv (offset))); 2219 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2220 XPUSHs (jsonstr);
1227} 2221}
1228 2222
1229PROTOTYPES: ENABLE
1230
1231void to_json (SV *scalar)
1232 ALIAS:
1233 objToJson = 0
1234 PPCODE:
1235 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1236
1237void from_json (SV *jsonstr)
1238 ALIAS:
1239 jsonToObj = 0
1240 PPCODE:
1241 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1242

Diff Legend

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