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 1.55 by root, Mon Jul 23 22:57:40 2007 UTC vs.
Revision 1.130 by root, Sat Nov 26 06:09:29 2016 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include <assert.h>
6#include "string.h" 6#include <string.h>
7#include "stdlib.h" 7#include <stdlib.h>
8#include "stdio.h" 8#include <stdio.h>
9#include <limits.h>
10#include <float.h>
11#include <inttypes.h>
9 12
10#if defined(__BORLANDC__) || defined(_MSC_VER) 13#if defined(__BORLANDC__) || defined(_MSC_VER)
11# define snprintf _snprintf // C compilers have this in stdio.h 14# define snprintf _snprintf // C compilers have this in stdio.h
12#endif 15#endif
13 16
14// some old perls do not have this, try to make it work, no 17// some old perls do not have this, try to make it work, no
15// guarentees, though. if it breaks, you get to keep the pieces. 18// guarantees, though. if it breaks, you get to keep the pieces.
16#ifndef UTF8_MAXBYTES 19#ifndef UTF8_MAXBYTES
17# define UTF8_MAXBYTES 13 20# define UTF8_MAXBYTES 13
18#endif 21#endif
22
23// compatibility with perl <5.18
24#ifndef HvNAMELEN_get
25# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
26#endif
27#ifndef HvNAMELEN
28# define HvNAMELEN(hv) HvNAMELEN_get (hv)
29#endif
30#ifndef HvNAMEUTF8
31# define HvNAMEUTF8(hv) 0
32#endif
33
34// three extra for rounding, sign, and end of string
35#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
19 36
20#define F_ASCII 0x00000001UL 37#define F_ASCII 0x00000001UL
21#define F_LATIN1 0x00000002UL 38#define F_LATIN1 0x00000002UL
22#define F_UTF8 0x00000004UL 39#define F_UTF8 0x00000004UL
23#define F_INDENT 0x00000008UL 40#define F_INDENT 0x00000008UL
… …
26#define F_SPACE_AFTER 0x00000040UL 43#define F_SPACE_AFTER 0x00000040UL
27#define F_ALLOW_NONREF 0x00000100UL 44#define F_ALLOW_NONREF 0x00000100UL
28#define F_SHRINK 0x00000200UL 45#define F_SHRINK 0x00000200UL
29#define F_ALLOW_BLESSED 0x00000400UL 46#define F_ALLOW_BLESSED 0x00000400UL
30#define F_CONV_BLESSED 0x00000800UL 47#define F_CONV_BLESSED 0x00000800UL
31#define F_MAXDEPTH 0xf8000000UL
32#define S_MAXDEPTH 27
33#define F_MAXSIZE 0x01f00000UL 48#define F_RELAXED 0x00001000UL
34#define S_MAXSIZE 20 49#define F_ALLOW_UNKNOWN 0x00002000UL
50#define F_ALLOW_TAGS 0x00004000UL
35#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing 51#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
36 52
37#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH))
38#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
39
40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 53#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
41#define F_DEFAULT (9UL << S_MAXDEPTH)
42 54
43#define INIT_SIZE 32 // initial scalar size to be allocated 55#define INIT_SIZE 32 // initial scalar size to be allocated
44#define INDENT_STEP 3 // spaces per indentation level 56#define INDENT_STEP 3 // spaces per indentation level
45 57
46#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 58#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
47 59
60#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
61
48#define SB do { 62#define SB do {
49#define SE } while (0) 63#define SE } while (0)
50 64
51#if __GNUC__ >= 3 65#if __GNUC__ >= 3
52# define expect(expr,value) __builtin_expect ((expr),(value)) 66# define expect(expr,value) __builtin_expect ((expr), (value))
53# define inline inline 67# define INLINE static inline
54#else 68#else
55# define expect(expr,value) (expr) 69# define expect(expr,value) (expr)
56# define inline static 70# define INLINE static
57#endif 71#endif
58 72
59#define expect_false(expr) expect ((expr) != 0, 0) 73#define expect_false(expr) expect ((expr) != 0, 0)
60#define expect_true(expr) expect ((expr) != 0, 1) 74#define expect_true(expr) expect ((expr) != 0, 1)
61 75
62static HV *json_stash, *json_boolean_stash; // JSON::XS:: 76#define IN_RANGE_INC(type,val,beg,end) \
63static SV *json_true, *json_false; 77 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
78 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
79
80#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
81
82#ifdef USE_ITHREADS
83# define JSON_SLOW 1
84# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
85# define BOOL_STASH (bool_stash ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1))
86#else
87# define JSON_SLOW 0
88# define JSON_STASH json_stash
89# define BOOL_STASH bool_stash
90#endif
91
92// the amount of HEs to allocate on the stack, when sorting keys
93#define STACK_HES 64
94
95static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean::
96static SV *bool_true, *bool_false, *sv_json;
97
98enum {
99 INCR_M_WS = 0, // initial whitespace skipping, must be 0
100 INCR_M_STR, // inside string
101 INCR_M_BS, // inside backslash
102 INCR_M_C0, // inside comment in initial whitespace sequence
103 INCR_M_C1, // inside comment in other places
104 INCR_M_JSON // outside anything, count nesting
105};
106
107#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
64 108
65typedef struct { 109typedef struct {
66 U32 flags; 110 U32 flags;
111 U32 max_depth;
112 STRLEN max_size;
113
67 SV *cb_object; 114 SV *cb_object;
68 HV *cb_sk_object; 115 HV *cb_sk_object;
116
117 // for the incremental parser
118 SV *incr_text; // the source text so far
119 STRLEN incr_pos; // the current offset into the text
120 int incr_nest; // {[]}-nesting level
121 unsigned char incr_mode;
69} JSON; 122} JSON;
123
124INLINE void
125json_init (JSON *json)
126{
127 Zero (json, 1, JSON);
128 json->max_depth = 512;
129}
70 130
71///////////////////////////////////////////////////////////////////////////// 131/////////////////////////////////////////////////////////////////////////////
72// utility functions 132// utility functions
73 133
74inline void 134INLINE SV *
135get_bool (const char *name)
136{
137 SV *sv = get_sv (name, 1);
138
139 SvREADONLY_on (sv);
140 SvREADONLY_on (SvRV (sv));
141
142 return sv;
143}
144
145INLINE void
75shrink (SV *sv) 146shrink (SV *sv)
76{ 147{
77 sv_utf8_downgrade (sv, 1); 148 sv_utf8_downgrade (sv, 1);
149
78 if (SvLEN (sv) > SvCUR (sv) + 1) 150 if (SvLEN (sv) > SvCUR (sv) + 1)
79 { 151 {
80#ifdef SvPV_shrink_to_cur 152#ifdef SvPV_shrink_to_cur
81 SvPV_shrink_to_cur (sv); 153 SvPV_shrink_to_cur (sv);
82#elif defined (SvPV_renew) 154#elif defined (SvPV_renew)
… …
88// decode an utf-8 character and return it, or (UV)-1 in 160// decode an utf-8 character and return it, or (UV)-1 in
89// case of an error. 161// case of an error.
90// we special-case "safe" characters from U+80 .. U+7FF, 162// we special-case "safe" characters from U+80 .. U+7FF,
91// but use the very good perl function to parse anything else. 163// but use the very good perl function to parse anything else.
92// note that we never call this function for a ascii codepoints 164// note that we never call this function for a ascii codepoints
93inline UV 165INLINE UV
94decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 166decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
95{ 167{
96 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 168 if (expect_true (len >= 2
97 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 169 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
98 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 170 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
99 { 171 {
100 *clen = 2; 172 *clen = 2;
101 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 173 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
102 } 174 }
103 else 175 else
176 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
177}
178
179// likewise for encoding, also never called for ascii codepoints
180// this function takes advantage of this fact, although current gccs
181// seem to optimise the check for >= 0x80 away anyways
182INLINE unsigned char *
183encode_utf8 (unsigned char *s, UV ch)
184{
185 if (expect_false (ch < 0x000080))
186 *s++ = ch;
187 else if (expect_true (ch < 0x000800))
188 *s++ = 0xc0 | ( ch >> 6),
189 *s++ = 0x80 | ( ch & 0x3f);
190 else if ( ch < 0x010000)
191 *s++ = 0xe0 | ( ch >> 12),
192 *s++ = 0x80 | ((ch >> 6) & 0x3f),
193 *s++ = 0x80 | ( ch & 0x3f);
194 else if ( ch < 0x110000)
195 *s++ = 0xf0 | ( ch >> 18),
196 *s++ = 0x80 | ((ch >> 12) & 0x3f),
197 *s++ = 0x80 | ((ch >> 6) & 0x3f),
198 *s++ = 0x80 | ( ch & 0x3f);
199
200 return s;
201}
202
203// convert offset pointer to character index, sv must be string
204static STRLEN
205ptr_to_index (SV *sv, char *offset)
206{
207 return SvUTF8 (sv)
208 ? utf8_distance (offset, SvPVX (sv))
209 : offset - SvPVX (sv);
210}
211
212/////////////////////////////////////////////////////////////////////////////
213// fp hell
214
215// scan a group of digits, and a trailing exponent
216static void
217json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
218{
219 UV uaccum = 0;
220 int eaccum = 0;
221
222 // if we recurse too deep, skip all remaining digits
223 // to avoid a stack overflow attack
224 if (expect_false (--maxdepth <= 0))
225 while (((U8)*s - '0') < 10)
226 ++s;
227
228 for (;;)
229 {
230 U8 dig = (U8)*s - '0';
231
232 if (expect_false (dig >= 10))
233 {
234 if (dig == (U8)((U8)'.' - (U8)'0'))
235 {
236 ++s;
237 json_atof_scan1 (s, accum, expo, 1, maxdepth);
238 }
239 else if ((dig | ' ') == 'e' - '0')
240 {
241 int exp2 = 0;
242 int neg = 0;
243
244 ++s;
245
246 if (*s == '-')
247 {
248 ++s;
249 neg = 1;
250 }
251 else if (*s == '+')
252 ++s;
253
254 while ((dig = (U8)*s - '0') < 10)
255 exp2 = exp2 * 10 + *s++ - '0';
256
257 *expo += neg ? -exp2 : exp2;
258 }
259
260 break;
261 }
262
263 ++s;
264
265 uaccum = uaccum * 10 + dig;
266 ++eaccum;
267
268 // if we have too many digits, then recurse for more
269 // we actually do this for rather few digits
270 if (uaccum >= (UV_MAX - 9) / 10)
271 {
272 if (postdp) *expo -= eaccum;
273 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
274 if (postdp) *expo += eaccum;
275
276 break;
277 }
104 { 278 }
105 *clen = (STRLEN)-1; 279
280 // this relies greatly on the quality of the pow ()
281 // implementation of the platform, but a good
282 // implementation is hard to beat.
283 // (IEEE 754 conformant ones are required to be exact)
284 if (postdp) *expo -= eaccum;
285 *accum += uaccum * Perl_pow (10., *expo);
286 *expo += eaccum;
287}
288
289static NV
290json_atof (const char *s)
291{
292 NV accum = 0.;
293 int expo = 0;
294 int neg = 0;
295
296 if (*s == '-')
297 {
298 ++s;
299 neg = 1;
300 }
301
302 // a recursion depth of ten gives us >>500 bits
303 json_atof_scan1 (s, &accum, &expo, 0, 10);
304
305 return neg ? -accum : accum;
306}
307
308// target of scalar reference is bool? -1 == nope, 0 == false, 1 == true
309static int
310ref_bool_type (SV *sv)
311{
312 svtype svt = SvTYPE (sv);
313
314 if (svt < SVt_PVAV)
315 {
316 STRLEN len = 0;
317 char *pv = svt ? SvPV (sv, len) : 0;
318
319 if (len == 1)
320 if (*pv == '1')
321 return 1;
322 else if (*pv == '0')
323 return 0;
324 }
325
326 return -1;
327}
328
329// returns whether scalar is not a reference in the sense of allow_nonref
330static int
331json_nonref (SV *scalar)
332{
333 if (!SvROK (scalar))
334 return 1;
335
336 scalar = SvRV (scalar);
337
338 if (SvTYPE (scalar) >= SVt_PVMG)
339 {
340 if (SvSTASH (scalar) == bool_stash)
106 return (UV)-1; 341 return 1;
342
343 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
344 return 1;
107 } 345 }
346
347 return 0;
108} 348}
109 349
110///////////////////////////////////////////////////////////////////////////// 350/////////////////////////////////////////////////////////////////////////////
111// encoder 351// encoder
112 352
… …
116 char *cur; // SvPVX (sv) + current output position 356 char *cur; // SvPVX (sv) + current output position
117 char *end; // SvEND (sv) 357 char *end; // SvEND (sv)
118 SV *sv; // result scalar 358 SV *sv; // result scalar
119 JSON json; 359 JSON json;
120 U32 indent; // indentation level 360 U32 indent; // indentation level
121 U32 maxdepth; // max. indentation/recursion level 361 UV limit; // escape character values >= this value when encoding
122} enc_t; 362} enc_t;
123 363
124inline void 364INLINE void
125need (enc_t *enc, STRLEN len) 365need (enc_t *enc, STRLEN len)
126{ 366{
127 if (expect_false (enc->cur + len >= enc->end)) 367 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
128 { 368 {
129 STRLEN cur = enc->cur - SvPVX (enc->sv); 369 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
130 SvGROW (enc->sv, cur + len + 1); 370 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
131 enc->cur = SvPVX (enc->sv) + cur; 371 enc->cur = SvPVX (enc->sv) + cur;
132 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 372 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
133 } 373 }
134} 374}
135 375
136inline void 376INLINE void
137encode_ch (enc_t *enc, char ch) 377encode_ch (enc_t *enc, char ch)
138{ 378{
139 need (enc, 1); 379 need (enc, 1);
140 *enc->cur++ = ch; 380 *enc->cur++ = ch;
141} 381}
… …
195 { 435 {
196 uch = ch; 436 uch = ch;
197 clen = 1; 437 clen = 1;
198 } 438 }
199 439
200 if (uch > 0x10FFFFUL) 440 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
201 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
202
203 if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF))
204 { 441 {
205 if (uch > 0xFFFFUL) 442 if (uch >= 0x10000UL)
206 { 443 {
444 if (uch >= 0x110000UL)
445 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
446
207 need (enc, len += 11); 447 need (enc, len += 11);
208 sprintf (enc->cur, "\\u%04x\\u%04x", 448 sprintf (enc->cur, "\\u%04x\\u%04x",
209 (int)((uch - 0x10000) / 0x400 + 0xD800), 449 (int)((uch - 0x10000) / 0x400 + 0xD800),
210 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 450 (int)((uch - 0x10000) % 0x400 + 0xDC00));
211 enc->cur += 12; 451 enc->cur += 12;
212 } 452 }
213 else 453 else
214 { 454 {
215 static char hexdigit [16] = "0123456789abcdef";
216 need (enc, len += 5); 455 need (enc, len += 5);
217 *enc->cur++ = '\\'; 456 *enc->cur++ = '\\';
218 *enc->cur++ = 'u'; 457 *enc->cur++ = 'u';
219 *enc->cur++ = hexdigit [ uch >> 12 ]; 458 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
220 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 459 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
221 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 460 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
222 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 461 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
223 } 462 }
224 463
225 str += clen; 464 str += clen;
226 } 465 }
227 else if (enc->json.flags & F_LATIN1) 466 else if (enc->json.flags & F_LATIN1)
… …
239 while (--clen); 478 while (--clen);
240 } 479 }
241 else 480 else
242 { 481 {
243 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 482 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
244 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 483 enc->cur = encode_utf8 (enc->cur, uch);
245 ++str; 484 ++str;
246 } 485 }
247 } 486 }
248 } 487 }
249 } 488 }
250 489
251 --len; 490 --len;
252 } 491 }
253} 492}
254 493
255inline void 494INLINE void
256encode_indent (enc_t *enc) 495encode_indent (enc_t *enc)
257{ 496{
258 if (enc->json.flags & F_INDENT) 497 if (enc->json.flags & F_INDENT)
259 { 498 {
260 int spaces = enc->indent * INDENT_STEP; 499 int spaces = enc->indent * INDENT_STEP;
… …
263 memset (enc->cur, ' ', spaces); 502 memset (enc->cur, ' ', spaces);
264 enc->cur += spaces; 503 enc->cur += spaces;
265 } 504 }
266} 505}
267 506
268inline void 507INLINE void
269encode_space (enc_t *enc) 508encode_space (enc_t *enc)
270{ 509{
271 need (enc, 1); 510 need (enc, 1);
272 encode_ch (enc, ' '); 511 encode_ch (enc, ' ');
273} 512}
274 513
275inline void 514INLINE void
276encode_nl (enc_t *enc) 515encode_nl (enc_t *enc)
277{ 516{
278 if (enc->json.flags & F_INDENT) 517 if (enc->json.flags & F_INDENT)
279 { 518 {
280 need (enc, 1); 519 need (enc, 1);
281 encode_ch (enc, '\n'); 520 encode_ch (enc, '\n');
282 } 521 }
283} 522}
284 523
285inline void 524INLINE void
286encode_comma (enc_t *enc) 525encode_comma (enc_t *enc)
287{ 526{
288 encode_ch (enc, ','); 527 encode_ch (enc, ',');
289 528
290 if (enc->json.flags & F_INDENT) 529 if (enc->json.flags & F_INDENT)
… …
298static void 537static void
299encode_av (enc_t *enc, AV *av) 538encode_av (enc_t *enc, AV *av)
300{ 539{
301 int i, len = av_len (av); 540 int i, len = av_len (av);
302 541
303 if (enc->indent >= enc->maxdepth) 542 if (enc->indent >= enc->json.max_depth)
304 croak ("data structure too deep (hit recursion limit)"); 543 croak (ERR_NESTING_EXCEEDED);
305 544
306 encode_ch (enc, '['); encode_nl (enc); 545 encode_ch (enc, '[');
307 ++enc->indent;
308 546
547 if (len >= 0)
548 {
549 encode_nl (enc); ++enc->indent;
550
309 for (i = 0; i <= len; ++i) 551 for (i = 0; i <= len; ++i)
310 { 552 {
311 SV **svp = av_fetch (av, i, 0); 553 SV **svp = av_fetch (av, i, 0);
312 554
313 encode_indent (enc); 555 encode_indent (enc);
314 556
315 if (svp) 557 if (svp)
316 encode_sv (enc, *svp); 558 encode_sv (enc, *svp);
317 else 559 else
318 encode_str (enc, "null", 4, 0); 560 encode_str (enc, "null", 4, 0);
319 561
320 if (i < len) 562 if (i < len)
321 encode_comma (enc); 563 encode_comma (enc);
322 } 564 }
323 565
566 encode_nl (enc); --enc->indent; encode_indent (enc);
567 }
568
324 encode_nl (enc); 569 encode_ch (enc, ']');
325
326 --enc->indent;
327 encode_indent (enc); encode_ch (enc, ']');
328} 570}
329 571
330static void 572static void
331encode_he (enc_t *enc, HE *he) 573encode_hk (enc_t *enc, HE *he)
332{ 574{
333 encode_ch (enc, '"'); 575 encode_ch (enc, '"');
334 576
335 if (HeKLEN (he) == HEf_SVKEY) 577 if (HeKLEN (he) == HEf_SVKEY)
336 { 578 {
337 SV *sv = HeSVKEY (he); 579 SV *sv = HeSVKEY (he);
338 STRLEN len; 580 STRLEN len;
339 char *str; 581 char *str;
340 582
341 SvGETMAGIC (sv); 583 SvGETMAGIC (sv);
342 str = SvPV (sv, len); 584 str = SvPV (sv, len);
343 585
344 encode_str (enc, str, len, SvUTF8 (sv)); 586 encode_str (enc, str, len, SvUTF8 (sv));
345 } 587 }
… …
349 encode_ch (enc, '"'); 591 encode_ch (enc, '"');
350 592
351 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); 593 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
352 encode_ch (enc, ':'); 594 encode_ch (enc, ':');
353 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); 595 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
354 encode_sv (enc, HeVAL (he));
355} 596}
356 597
357// compare hash entries, used when all keys are bytestrings 598// compare hash entries, used when all keys are bytestrings
358static int 599static int
359he_cmp_fast (const void *a_, const void *b_) 600he_cmp_fast (const void *a_, const void *b_)
… …
364 HE *b = *(HE **)b_; 605 HE *b = *(HE **)b_;
365 606
366 STRLEN la = HeKLEN (a); 607 STRLEN la = HeKLEN (a);
367 STRLEN lb = HeKLEN (b); 608 STRLEN lb = HeKLEN (b);
368 609
369 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 610 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
370 cmp = la - lb; 611 cmp = lb - la;
371 612
372 return cmp; 613 return cmp;
373} 614}
374 615
375// compare hash entries, used when some keys are sv's or utf-x 616// compare hash entries, used when some keys are sv's or utf-x
376static int 617static int
377he_cmp_slow (const void *a, const void *b) 618he_cmp_slow (const void *a, const void *b)
378{ 619{
379 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 620 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
380} 621}
381 622
382static void 623static void
383encode_hv (enc_t *enc, HV *hv) 624encode_hv (enc_t *enc, HV *hv)
384{ 625{
385 int count, i; 626 HE *he;
386 627
387 if (enc->indent >= enc->maxdepth) 628 if (enc->indent >= enc->json.max_depth)
388 croak ("data structure too deep (hit recursion limit)"); 629 croak (ERR_NESTING_EXCEEDED);
389 630
390 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 631 encode_ch (enc, '{');
391 632
392 if ((count = hv_iterinit (hv)))
393 {
394 // for canonical output we have to sort by keys first 633 // for canonical output we have to sort by keys first
395 // actually, this is mostly due to the stupid so-called 634 // actually, this is mostly due to the stupid so-called
396 // security workaround added somewhere in 5.8.x. 635 // security workaround added somewhere in 5.8.x
397 // that randomises hash orderings 636 // that randomises hash orderings
398 if (enc->json.flags & F_CANONICAL) 637 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
638 {
639 int count = hv_iterinit (hv);
640
641 if (SvMAGICAL (hv))
399 { 642 {
643 // need to count by iterating. could improve by dynamically building the vector below
644 // but I don't care for the speed of this special case.
645 // note also that we will run into undefined behaviour when the two iterations
646 // do not result in the same count, something I might care for in some later release.
647
648 count = 0;
649 while (hv_iternext (hv))
650 ++count;
651
652 hv_iterinit (hv);
653 }
654
655 if (count)
656 {
400 int fast = 1; 657 int i, fast = 1;
401 HE *he; 658 HE *hes_stack [STACK_HES];
402#if defined(__BORLANDC__) || defined(_MSC_VER) 659 HE **hes = hes_stack;
403 HE **hes = _alloca (count * sizeof (HE)); 660
404#else 661 // allocate larger arrays on the heap
405 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 662 if (count > STACK_HES)
406#endif 663 {
664 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
665 hes = (HE **)SvPVX (sv);
666 }
407 667
408 i = 0; 668 i = 0;
409 while ((he = hv_iternext (hv))) 669 while ((he = hv_iternext (hv)))
410 { 670 {
411 hes [i++] = he; 671 hes [i++] = he;
… …
433 693
434 FREETMPS; 694 FREETMPS;
435 LEAVE; 695 LEAVE;
436 } 696 }
437 697
438 for (i = 0; i < count; ++i) 698 encode_nl (enc); ++enc->indent;
699
700 while (count--)
439 { 701 {
440 encode_indent (enc); 702 encode_indent (enc);
703 he = hes [count];
441 encode_he (enc, hes [i]); 704 encode_hk (enc, he);
705 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
442 706
443 if (i < count - 1) 707 if (count)
444 encode_comma (enc); 708 encode_comma (enc);
445 } 709 }
446 710
447 encode_nl (enc); 711 encode_nl (enc); --enc->indent; encode_indent (enc);
448 } 712 }
713 }
449 else 714 else
715 {
716 if (hv_iterinit (hv) || SvMAGICAL (hv))
717 if ((he = hv_iternext (hv)))
450 { 718 {
451 HE *he = hv_iternext (hv); 719 encode_nl (enc); ++enc->indent;
452 720
453 for (;;) 721 for (;;)
454 { 722 {
455 encode_indent (enc); 723 encode_indent (enc);
456 encode_he (enc, he); 724 encode_hk (enc, he);
725 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
457 726
458 if (!(he = hv_iternext (hv))) 727 if (!(he = hv_iternext (hv)))
459 break; 728 break;
460 729
461 encode_comma (enc); 730 encode_comma (enc);
462 } 731 }
463 732
464 encode_nl (enc); 733 encode_nl (enc); --enc->indent; encode_indent (enc);
465 } 734 }
466 } 735 }
467 736
468 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 737 encode_ch (enc, '}');
469} 738}
470 739
471// encode objects, arrays and special \0=false and \1=true values. 740// encode objects, arrays and special \0=false and \1=true values.
472static void 741static void
473encode_rv (enc_t *enc, SV *sv) 742encode_rv (enc_t *enc, SV *sv)
474{ 743{
475 svtype svt; 744 svtype svt;
745 GV *method;
476 746
477 SvGETMAGIC (sv); 747 SvGETMAGIC (sv);
478 svt = SvTYPE (sv); 748 svt = SvTYPE (sv);
479 749
480 if (expect_false (SvOBJECT (sv))) 750 if (expect_false (SvOBJECT (sv)))
481 { 751 {
482 if (SvSTASH (sv) == json_boolean_stash) 752 HV *stash = SvSTASH (sv);
753
754 if (stash == bool_stash)
483 { 755 {
484 if (SvIV (sv)) 756 if (SvIV (sv))
485 encode_str (enc, "true", 4, 0); 757 encode_str (enc, "true", 4, 0);
486 else 758 else
487 encode_str (enc, "false", 5, 0); 759 encode_str (enc, "false", 5, 0);
488 } 760 }
761 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
762 {
763 int count;
764 dSP;
765
766 ENTER; SAVETMPS;
767 SAVESTACK_POS ();
768 PUSHMARK (SP);
769 EXTEND (SP, 2);
770 // we re-bless the reference to get overload and other niceties right
771 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
772 PUSHs (sv_json);
773
774 PUTBACK;
775 count = call_sv ((SV *)GvCV (method), G_ARRAY);
776 SPAGAIN;
777
778 // catch this surprisingly common error
779 if (SvROK (TOPs) && SvRV (TOPs) == sv)
780 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
781
782 encode_ch (enc, '(');
783 encode_ch (enc, '"');
784 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
785 encode_ch (enc, '"');
786 encode_ch (enc, ')');
787 encode_ch (enc, '[');
788
789 while (count)
790 {
791 encode_sv (enc, SP[1 - count--]);
792
793 if (count)
794 encode_ch (enc, ',');
795 }
796
797 encode_ch (enc, ']');
798
799 FREETMPS; LEAVE;
800 }
801 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
802 {
803 dSP;
804
805 ENTER; SAVETMPS;
806 PUSHMARK (SP);
807 // we re-bless the reference to get overload and other niceties right
808 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
809
810 // calling with G_SCALAR ensures that we always get a 1 return value
811 PUTBACK;
812 call_sv ((SV *)GvCV (method), G_SCALAR);
813 SPAGAIN;
814
815 // catch this surprisingly common error
816 if (SvROK (TOPs) && SvRV (TOPs) == sv)
817 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
818
819 sv = POPs;
820 PUTBACK;
821
822 encode_sv (enc, sv);
823
824 FREETMPS; LEAVE;
825 }
826 else if (enc->json.flags & F_ALLOW_BLESSED)
827 encode_str (enc, "null", 4, 0);
489 else 828 else
490 {
491#if 0
492 if (0 && sv_derived_from (rv, "JSON::Literal"))
493 {
494 // not yet
495 }
496#endif
497 if (enc->json.flags & F_CONV_BLESSED)
498 {
499 // we re-bless the reference to get overload and other niceties right
500 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
501
502 if (to_json)
503 {
504 int count;
505 dSP;
506
507 ENTER; SAVETMPS; PUSHMARK (SP);
508 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
509
510 // calling with G_SCALAR ensures that we always get a 1 return value
511 PUTBACK;
512 call_sv ((SV *)GvCV (to_json), G_SCALAR);
513 SPAGAIN;
514
515 // catch this surprisingly common error
516 if (SvROK (TOPs) && SvRV (TOPs) == sv)
517 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
518
519 sv = POPs;
520 PUTBACK;
521
522 encode_sv (enc, sv);
523
524 FREETMPS; LEAVE;
525 }
526 else if (enc->json.flags & F_ALLOW_BLESSED)
527 encode_str (enc, "null", 4, 0);
528 else
529 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
530 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
531 }
532 else if (enc->json.flags & F_ALLOW_BLESSED)
533 encode_str (enc, "null", 4, 0);
534 else
535 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 829 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
536 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 830 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
537 }
538 } 831 }
539 else if (svt == SVt_PVHV) 832 else if (svt == SVt_PVHV)
540 encode_hv (enc, (HV *)sv); 833 encode_hv (enc, (HV *)sv);
541 else if (svt == SVt_PVAV) 834 else if (svt == SVt_PVAV)
542 encode_av (enc, (AV *)sv); 835 encode_av (enc, (AV *)sv);
543 else if (svt < SVt_PVAV) 836 else if (svt < SVt_PVAV)
544 { 837 {
545 STRLEN len = 0; 838 int bool_type = ref_bool_type (sv);
546 char *pv = svt ? SvPV (sv, len) : 0;
547 839
548 if (len == 1 && *pv == '1') 840 if (bool_type == 1)
549 encode_str (enc, "true", 4, 0); 841 encode_str (enc, "true", 4, 0);
550 else if (len == 1 && *pv == '0') 842 else if (bool_type == 0)
551 encode_str (enc, "false", 5, 0); 843 encode_str (enc, "false", 5, 0);
844 else if (enc->json.flags & F_ALLOW_UNKNOWN)
845 encode_str (enc, "null", 4, 0);
552 else 846 else
553 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 847 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
554 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 848 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
555 } 849 }
850 else if (enc->json.flags & F_ALLOW_UNKNOWN)
851 encode_str (enc, "null", 4, 0);
556 else 852 else
557 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 853 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
558 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 854 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
559} 855}
560 856
… …
578 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 874 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
579 enc->cur += strlen (enc->cur); 875 enc->cur += strlen (enc->cur);
580 } 876 }
581 else if (SvIOKp (sv)) 877 else if (SvIOKp (sv))
582 { 878 {
583 // we assume we can always read an IV as a UV 879 // we assume we can always read an IV as a UV and vice versa
584 if (SvUV (sv) & ~(UV)0x7fff) 880 // we assume two's complement
585 { 881 // we assume no aliasing issues in the union
586 // large integer, use the (rather slow) snprintf way. 882 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
587 need (enc, sizeof (UV) * 3); 883 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
588 enc->cur +=
589 SvIsUV(sv)
590 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
591 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
592 }
593 else
594 { 884 {
595 // optimise the "small number case" 885 // optimise the "small number case"
596 // code will likely be branchless and use only a single multiplication 886 // code will likely be branchless and use only a single multiplication
887 // works for numbers up to 59074
597 I32 i = SvIV (sv); 888 I32 i = SvIVX (sv);
598 U32 u; 889 U32 u;
599 char digit, nz = 0; 890 char digit, nz = 0;
600 891
601 need (enc, 6); 892 need (enc, 6);
602 893
… …
608 899
609 // now output digit by digit, each time masking out the integer part 900 // now output digit by digit, each time masking out the integer part
610 // and multiplying by 5 while moving the decimal point one to the right, 901 // and multiplying by 5 while moving the decimal point one to the right,
611 // resulting in a net multiplication by 10. 902 // resulting in a net multiplication by 10.
612 // we always write the digit to memory but conditionally increment 903 // we always write the digit to memory but conditionally increment
613 // the pointer, to ease the usage of conditional move instructions. 904 // the pointer, to enable the use of conditional move instructions.
614 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 905 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
615 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 906 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
616 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 907 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
617 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 908 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
618 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 909 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
619 } 910 }
911 else
912 {
913 // large integer, use the (rather slow) snprintf way.
914 need (enc, IVUV_MAXCHARS);
915 enc->cur +=
916 SvIsUV(sv)
917 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
918 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
919 }
620 } 920 }
621 else if (SvROK (sv)) 921 else if (SvROK (sv))
622 encode_rv (enc, SvRV (sv)); 922 encode_rv (enc, SvRV (sv));
623 else if (!SvOK (sv)) 923 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
624 encode_str (enc, "null", 4, 0); 924 encode_str (enc, "null", 4, 0);
625 else 925 else
626 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 926 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
627 SvPV_nolen (sv), SvFLAGS (sv)); 927 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
628} 928}
629 929
630static SV * 930static SV *
631encode_json (SV *scalar, JSON *json) 931encode_json (SV *scalar, JSON *json)
632{ 932{
633 enc_t enc; 933 enc_t enc;
634 934
635 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) 935 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
636 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 936 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
637 937
638 enc.json = *json; 938 enc.json = *json;
639 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 939 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
640 enc.cur = SvPVX (enc.sv); 940 enc.cur = SvPVX (enc.sv);
641 enc.end = SvEND (enc.sv); 941 enc.end = SvEND (enc.sv);
642 enc.indent = 0; 942 enc.indent = 0;
643 enc.maxdepth = DEC_DEPTH (enc.json.flags); 943 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
944 : enc.json.flags & F_LATIN1 ? 0x000100UL
945 : 0x110000UL;
644 946
645 SvPOK_only (enc.sv); 947 SvPOK_only (enc.sv);
646 encode_sv (&enc, scalar); 948 encode_sv (&enc, scalar);
949 encode_nl (&enc);
647 950
648 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 951 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
649 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 952 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
650 953
651 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 954 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
669 JSON json; 972 JSON json;
670 U32 depth; // recursion depth 973 U32 depth; // recursion depth
671 U32 maxdepth; // recursion depth limit 974 U32 maxdepth; // recursion depth limit
672} dec_t; 975} dec_t;
673 976
674inline void 977INLINE void
978decode_comment (dec_t *dec)
979{
980 // only '#'-style comments allowed a.t.m.
981
982 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
983 ++dec->cur;
984}
985
986INLINE void
675decode_ws (dec_t *dec) 987decode_ws (dec_t *dec)
676{ 988{
677 for (;;) 989 for (;;)
678 { 990 {
679 char ch = *dec->cur; 991 char ch = *dec->cur;
680 992
681 if (ch > 0x20 993 if (ch > 0x20)
994 {
995 if (expect_false (ch == '#'))
996 {
997 if (dec->json.flags & F_RELAXED)
998 decode_comment (dec);
999 else
1000 break;
1001 }
1002 else
1003 break;
1004 }
682 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 1005 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
683 break; 1006 break; // parse error, but let higher level handle it, gives better error messages
684 1007
685 ++dec->cur; 1008 ++dec->cur;
686 } 1009 }
687} 1010}
688 1011
… …
692 if (*dec->cur != ch) \ 1015 if (*dec->cur != ch) \
693 ERR (# ch " expected"); \ 1016 ERR (# ch " expected"); \
694 ++dec->cur; \ 1017 ++dec->cur; \
695 SE 1018 SE
696 1019
697#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 1020#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
698#define DEC_DEC_DEPTH --dec->depth 1021#define DEC_DEC_DEPTH --dec->depth
699 1022
700static SV *decode_sv (dec_t *dec); 1023static SV *decode_sv (dec_t *dec);
701 1024
702static signed char decode_hexdigit[256]; 1025static signed char decode_hexdigit[256];
… …
794 1117
795 if (hi >= 0x80) 1118 if (hi >= 0x80)
796 { 1119 {
797 utf8 = 1; 1120 utf8 = 1;
798 1121
799 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1122 cur = encode_utf8 (cur, hi);
800 } 1123 }
801 else 1124 else
802 *cur++ = hi; 1125 *cur++ = hi;
803 } 1126 }
804 break; 1127 break;
… …
806 default: 1129 default:
807 --dec_cur; 1130 --dec_cur;
808 ERR ("illegal backslash escape sequence in string"); 1131 ERR ("illegal backslash escape sequence in string");
809 } 1132 }
810 } 1133 }
811 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1134 else if (expect_true (ch >= 0x20 && ch < 0x80))
812 *cur++ = ch; 1135 *cur++ = ch;
813 else if (ch >= 0x80) 1136 else if (ch >= 0x80)
814 { 1137 {
815 STRLEN clen; 1138 STRLEN clen;
816 UV uch;
817 1139
818 --dec_cur; 1140 --dec_cur;
819 1141
820 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1142 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
821 if (clen == (STRLEN)-1) 1143 if (clen == (STRLEN)-1)
822 ERR ("malformed UTF-8 character in JSON string"); 1144 ERR ("malformed UTF-8 character in JSON string");
823 1145
824 do 1146 do
825 *cur++ = *dec_cur++; 1147 *cur++ = *dec_cur++;
826 while (--clen); 1148 while (--clen);
827 1149
828 utf8 = 1; 1150 utf8 = 1;
829 } 1151 }
1152 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1153 *cur++ = ch;
830 else 1154 else
831 { 1155 {
832 --dec_cur; 1156 --dec_cur;
833 1157
834 if (!ch) 1158 if (!ch)
… …
842 { 1166 {
843 STRLEN len = cur - buf; 1167 STRLEN len = cur - buf;
844 1168
845 if (sv) 1169 if (sv)
846 { 1170 {
847 SvGROW (sv, SvCUR (sv) + len + 1); 1171 STRLEN cur = SvCUR (sv);
1172
1173 if (SvLEN (sv) <= cur + len)
1174 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1175
848 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1176 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
849 SvCUR_set (sv, SvCUR (sv) + len); 1177 SvCUR_set (sv, SvCUR (sv) + len);
850 } 1178 }
851 else 1179 else
852 sv = newSVpvn (buf, len); 1180 sv = newSVpvn (buf, len);
… …
937 is_nv = 1; 1265 is_nv = 1;
938 } 1266 }
939 1267
940 if (!is_nv) 1268 if (!is_nv)
941 { 1269 {
1270 int len = dec->cur - start;
1271
942 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1272 // special case the rather common 1..5-digit-int case
943 if (*start == '-') 1273 if (*start == '-')
944 switch (dec->cur - start) 1274 switch (len)
945 { 1275 {
946 case 2: return newSViv (-( start [1] - '0' * 1)); 1276 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
947 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1277 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
948 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1278 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
949 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1279 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1280 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
950 } 1281 }
951 else 1282 else
952 switch (dec->cur - start) 1283 switch (len)
953 { 1284 {
954 case 1: return newSViv ( start [0] - '0' * 1); 1285 case 1: return newSViv ( start [0] - '0' * 1);
955 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1286 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
956 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1287 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
957 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1288 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1289 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
958 } 1290 }
959 1291
960 { 1292 {
961 UV uv; 1293 UV uv;
962 int numtype = grok_number (start, dec->cur - start, &uv); 1294 int numtype = grok_number (start, len, &uv);
963 if (numtype & IS_NUMBER_IN_UV) 1295 if (numtype & IS_NUMBER_IN_UV)
964 if (numtype & IS_NUMBER_NEG) 1296 if (numtype & IS_NUMBER_NEG)
965 { 1297 {
966 if (uv < (UV)IV_MIN) 1298 if (uv < (UV)IV_MIN)
967 return newSViv (-(IV)uv); 1299 return newSViv (-(IV)uv);
968 } 1300 }
969 else 1301 else
970 return newSVuv (uv); 1302 return newSVuv (uv);
971
972 // here would likely be the place for bigint support
973 } 1303 }
974 }
975 1304
976 // if we ever support bigint or bigfloat, this is the place for bigfloat 1305 len -= *start == '-' ? 1 : 0;
1306
1307 // does not fit into IV or UV, try NV
1308 if (len <= NV_DIG)
1309 // fits into NV without loss of precision
1310 return newSVnv (json_atof (start));
1311
1312 // everything else fails, convert it to a string
1313 return newSVpvn (start, dec->cur - start);
1314 }
1315
1316 // loss of precision here
977 return newSVnv (Atof (start)); 1317 return newSVnv (json_atof (start));
978 1318
979fail: 1319fail:
980 return 0; 1320 return 0;
981} 1321}
982 1322
… …
1006 if (*dec->cur == ']') 1346 if (*dec->cur == ']')
1007 { 1347 {
1008 ++dec->cur; 1348 ++dec->cur;
1009 break; 1349 break;
1010 } 1350 }
1011 1351
1012 if (*dec->cur != ',') 1352 if (*dec->cur != ',')
1013 ERR (", or ] expected while parsing array"); 1353 ERR (", or ] expected while parsing array");
1014 1354
1015 ++dec->cur; 1355 ++dec->cur;
1356
1357 decode_ws (dec);
1358
1359 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1360 {
1361 ++dec->cur;
1362 break;
1363 }
1016 } 1364 }
1017 1365
1018 DEC_DEC_DEPTH; 1366 DEC_DEC_DEPTH;
1019 return newRV_noinc ((SV *)av); 1367 return newRV_noinc ((SV *)av);
1020 1368
… …
1036 if (*dec->cur == '}') 1384 if (*dec->cur == '}')
1037 ++dec->cur; 1385 ++dec->cur;
1038 else 1386 else
1039 for (;;) 1387 for (;;)
1040 { 1388 {
1041 decode_ws (dec); EXPECT_CH ('"'); 1389 EXPECT_CH ('"');
1042 1390
1043 // heuristic: assume that 1391 // heuristic: assume that
1044 // a) decode_str + hv_store_ent are abysmally slow. 1392 // a) decode_str + hv_store_ent are abysmally slow.
1045 // b) most hash keys are short, simple ascii text. 1393 // b) most hash keys are short, simple ascii text.
1046 // => try to "fast-match" such strings to avoid 1394 // => try to "fast-match" such strings to avoid
… …
1050 char *p = dec->cur; 1398 char *p = dec->cur;
1051 char *e = p + 24; // only try up to 24 bytes 1399 char *e = p + 24; // only try up to 24 bytes
1052 1400
1053 for (;;) 1401 for (;;)
1054 { 1402 {
1055 // the >= 0x80 is true on most architectures 1403 // the >= 0x80 is false on most architectures
1056 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1404 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1057 { 1405 {
1058 // slow path, back up and use decode_str 1406 // slow path, back up and use decode_str
1059 SV *key = decode_str (dec); 1407 SV *key = decode_str (dec);
1060 if (!key) 1408 if (!key)
1061 goto fail; 1409 goto fail;
1062 1410
1063 decode_ws (dec); EXPECT_CH (':'); 1411 decode_ws (dec); EXPECT_CH (':');
1064 1412
1413 decode_ws (dec);
1065 value = decode_sv (dec); 1414 value = decode_sv (dec);
1066 if (!value) 1415 if (!value)
1067 { 1416 {
1068 SvREFCNT_dec (key); 1417 SvREFCNT_dec (key);
1069 goto fail; 1418 goto fail;
… …
1081 int len = p - key; 1430 int len = p - key;
1082 dec->cur = p + 1; 1431 dec->cur = p + 1;
1083 1432
1084 decode_ws (dec); EXPECT_CH (':'); 1433 decode_ws (dec); EXPECT_CH (':');
1085 1434
1435 decode_ws (dec);
1086 value = decode_sv (dec); 1436 value = decode_sv (dec);
1087 if (!value) 1437 if (!value)
1088 goto fail; 1438 goto fail;
1089 1439
1090 hv_store (hv, key, len, value, 0); 1440 hv_store (hv, key, len, value, 0);
… …
1106 1456
1107 if (*dec->cur != ',') 1457 if (*dec->cur != ',')
1108 ERR (", or } expected while parsing object/hash"); 1458 ERR (", or } expected while parsing object/hash");
1109 1459
1110 ++dec->cur; 1460 ++dec->cur;
1461
1462 decode_ws (dec);
1463
1464 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1465 {
1466 ++dec->cur;
1467 break;
1468 }
1111 } 1469 }
1112 1470
1113 DEC_DEC_DEPTH; 1471 DEC_DEC_DEPTH;
1114 sv = newRV_noinc ((SV *)hv); 1472 sv = newRV_noinc ((SV *)hv);
1115 1473
… …
1122 1480
1123 hv_iterinit (hv); 1481 hv_iterinit (hv);
1124 he = hv_iternext (hv); 1482 he = hv_iternext (hv);
1125 hv_iterinit (hv); 1483 hv_iterinit (hv);
1126 1484
1127 // the next line creates a mortal sv each time its called. 1485 // the next line creates a mortal sv each time it's called.
1128 // might want to optimise this for common cases. 1486 // might want to optimise this for common cases.
1129 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); 1487 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1130 1488
1131 if (cb) 1489 if (cb)
1132 { 1490 {
1133 dSP; 1491 dSP;
1134 int count; 1492 int count;
1135 1493
1136 ENTER; SAVETMPS; PUSHMARK (SP); 1494 ENTER; SAVETMPS;
1495 SAVESTACK_POS ();
1496 PUSHMARK (SP);
1137 XPUSHs (HeVAL (he)); 1497 XPUSHs (HeVAL (he));
1498 sv_2mortal (sv);
1138 1499
1139 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1500 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1140 1501
1141 if (count == 1) 1502 if (count == 1)
1142 { 1503 {
1143 sv = newSVsv (POPs); 1504 sv = newSVsv (POPs);
1144 FREETMPS; LEAVE; 1505 FREETMPS; LEAVE;
1145 return sv; 1506 return sv;
1146 } 1507 }
1147 1508
1509 SvREFCNT_inc (sv);
1148 FREETMPS; LEAVE; 1510 FREETMPS; LEAVE;
1149 } 1511 }
1150 } 1512 }
1151 1513
1152 if (dec->json.cb_object) 1514 if (dec->json.cb_object)
1153 { 1515 {
1154 dSP; 1516 dSP;
1155 int count; 1517 int count;
1156 1518
1157 ENTER; SAVETMPS; PUSHMARK (SP); 1519 ENTER; SAVETMPS;
1520 SAVESTACK_POS ();
1521 PUSHMARK (SP);
1158 XPUSHs (sv_2mortal (sv)); 1522 XPUSHs (sv_2mortal (sv));
1159 1523
1160 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; 1524 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1161 1525
1162 if (count == 1) 1526 if (count == 1)
… …
1178 DEC_DEC_DEPTH; 1542 DEC_DEC_DEPTH;
1179 return 0; 1543 return 0;
1180} 1544}
1181 1545
1182static SV * 1546static SV *
1547decode_tag (dec_t *dec)
1548{
1549 SV *tag = 0;
1550 SV *val = 0;
1551
1552 if (!(dec->json.flags & F_ALLOW_TAGS))
1553 ERR ("malformed JSON string, neither array, object, number, string or atom");
1554
1555 ++dec->cur;
1556
1557 decode_ws (dec);
1558
1559 tag = decode_sv (dec);
1560 if (!tag)
1561 goto fail;
1562
1563 if (!SvPOK (tag))
1564 ERR ("malformed JSON string, (tag) must be a string");
1565
1566 decode_ws (dec);
1567
1568 if (*dec->cur != ')')
1569 ERR (") expected after tag");
1570
1571 ++dec->cur;
1572
1573 decode_ws (dec);
1574
1575 val = decode_sv (dec);
1576 if (!val)
1577 goto fail;
1578
1579 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1580 ERR ("malformed JSON string, tag value must be an array");
1581
1582 {
1583 AV *av = (AV *)SvRV (val);
1584 int i, len = av_len (av) + 1;
1585 HV *stash = gv_stashsv (tag, 0);
1586 SV *sv;
1587
1588 if (!stash)
1589 ERR ("cannot decode perl-object (package does not exist)");
1590
1591 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1592
1593 if (!method)
1594 ERR ("cannot decode perl-object (package does not have a THAW method)");
1595
1596 dSP;
1597
1598 ENTER; SAVETMPS;
1599 PUSHMARK (SP);
1600 EXTEND (SP, len + 2);
1601 // we re-bless the reference to get overload and other niceties right
1602 PUSHs (tag);
1603 PUSHs (sv_json);
1604
1605 for (i = 0; i < len; ++i)
1606 PUSHs (*av_fetch (av, i, 1));
1607
1608 PUTBACK;
1609 call_sv ((SV *)GvCV (method), G_SCALAR);
1610 SPAGAIN;
1611
1612 SvREFCNT_dec (tag);
1613 SvREFCNT_dec (val);
1614 sv = SvREFCNT_inc (POPs);
1615
1616 PUTBACK;
1617
1618 FREETMPS; LEAVE;
1619
1620 return sv;
1621 }
1622
1623fail:
1624 SvREFCNT_dec (tag);
1625 SvREFCNT_dec (val);
1626 return 0;
1627}
1628
1629static SV *
1183decode_sv (dec_t *dec) 1630decode_sv (dec_t *dec)
1184{ 1631{
1185 decode_ws (dec);
1186
1187 // the beauty of JSON: you need exactly one character lookahead 1632 // the beauty of JSON: you need exactly one character lookahead
1188 // to parse anything. 1633 // to parse everything.
1189 switch (*dec->cur) 1634 switch (*dec->cur)
1190 { 1635 {
1191 case '"': ++dec->cur; return decode_str (dec); 1636 case '"': ++dec->cur; return decode_str (dec);
1192 case '[': ++dec->cur; return decode_av (dec); 1637 case '[': ++dec->cur; return decode_av (dec);
1193 case '{': ++dec->cur; return decode_hv (dec); 1638 case '{': ++dec->cur; return decode_hv (dec);
1639 case '(': return decode_tag (dec);
1194 1640
1195 case '-': 1641 case '-':
1196 case '0': case '1': case '2': case '3': case '4': 1642 case '0': case '1': case '2': case '3': case '4':
1197 case '5': case '6': case '7': case '8': case '9': 1643 case '5': case '6': case '7': case '8': case '9':
1198 return decode_num (dec); 1644 return decode_num (dec);
1199 1645
1200 case 't': 1646 case 't':
1201 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1647 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1202 { 1648 {
1203 dec->cur += 4; 1649 dec->cur += 4;
1204 return SvREFCNT_inc (json_true); 1650#if JSON_SLOW
1651 bool_true = get_bool ("Types::Serialiser::true");
1652#endif
1653 return newSVsv (bool_true);
1205 } 1654 }
1206 else 1655 else
1207 ERR ("'true' expected"); 1656 ERR ("'true' expected");
1208 1657
1209 break; 1658 break;
1210 1659
1211 case 'f': 1660 case 'f':
1212 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1661 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1213 { 1662 {
1214 dec->cur += 5; 1663 dec->cur += 5;
1664#if JSON_SLOW
1665 bool_false = get_bool ("Types::Serialiser::false");
1666#endif
1215 return SvREFCNT_inc (json_false); 1667 return newSVsv (bool_false);
1216 } 1668 }
1217 else 1669 else
1218 ERR ("'false' expected"); 1670 ERR ("'false' expected");
1219 1671
1220 break; 1672 break;
… …
1229 ERR ("'null' expected"); 1681 ERR ("'null' expected");
1230 1682
1231 break; 1683 break;
1232 1684
1233 default: 1685 default:
1234 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1686 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1235 break; 1687 break;
1236 } 1688 }
1237 1689
1238fail: 1690fail:
1239 return 0; 1691 return 0;
1240} 1692}
1241 1693
1242static SV * 1694static SV *
1243decode_json (SV *string, JSON *json, UV *offset_return) 1695decode_json (SV *string, JSON *json, STRLEN *offset_return)
1244{ 1696{
1245 dec_t dec; 1697 dec_t dec;
1246 UV offset;
1247 SV *sv; 1698 SV *sv;
1248 1699
1700 /* work around bugs in 5.10 where manipulating magic values
1701 * makes perl ignore the magic in subsequent accesses.
1702 * also make a copy of non-PV values, to get them into a clean
1703 * state (SvPV should do that, but it's buggy, see below).
1704 *
1705 * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
1706 * as reported by Reini Urban.
1707 */
1249 SvGETMAGIC (string); 1708 /*SvGETMAGIC (string);*/
1709 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1710 string = sv_2mortal (newSVsv (string));
1711
1250 SvUPGRADE (string, SVt_PV); 1712 SvUPGRADE (string, SVt_PV);
1251 1713
1252 if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) 1714 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1715 * assertion with -DDEBUGGING, although SvCUR is documented to
1716 * return the xpv_cur field which certainly exists after upgrading.
1717 * according to nicholas clark, calling SvPOK fixes this.
1718 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1719 * and hope for the best.
1720 * Damnit, SvPV_nolen still trips over yet another assertion. This
1721 * assertion business is seriously broken, try yet another workaround
1722 * for the broken -DDEBUGGING.
1723 */
1724 {
1725#ifdef DEBUGGING
1726 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1727#else
1728 STRLEN offset = SvCUR (string);
1729#endif
1730
1731 if (offset > json->max_size && json->max_size)
1253 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1732 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1254 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); 1733 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1734 }
1255 1735
1256 if (json->flags & F_UTF8) 1736 if (DECODE_WANTS_OCTETS (json))
1257 sv_utf8_downgrade (string, 0); 1737 sv_utf8_downgrade (string, 0);
1258 else 1738 else
1259 sv_utf8_upgrade (string); 1739 sv_utf8_upgrade (string);
1260 1740
1261 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1741 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1262 1742
1263 dec.json = *json; 1743 dec.json = *json;
1264 dec.cur = SvPVX (string); 1744 dec.cur = SvPVX (string);
1265 dec.end = SvEND (string); 1745 dec.end = SvEND (string);
1266 dec.err = 0; 1746 dec.err = 0;
1267 dec.depth = 0; 1747 dec.depth = 0;
1268 dec.maxdepth = DEC_DEPTH (dec.json.flags);
1269 1748
1270 if (dec.json.cb_object || dec.json.cb_sk_object) 1749 if (dec.json.cb_object || dec.json.cb_sk_object)
1271 dec.json.flags |= F_HOOK; 1750 dec.json.flags |= F_HOOK;
1272 1751
1273 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1752 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1753
1754 decode_ws (&dec);
1274 sv = decode_sv (&dec); 1755 sv = decode_sv (&dec);
1275 1756
1276 if (!(offset_return || !sv)) 1757 if (offset_return)
1758 *offset_return = dec.cur - SvPVX (string);
1759 else if (sv)
1277 { 1760 {
1278 // check for trailing garbage 1761 // check for trailing garbage
1279 decode_ws (&dec); 1762 decode_ws (&dec);
1280 1763
1281 if (*dec.cur) 1764 if (*dec.cur)
1282 { 1765 {
1283 dec.err = "garbage after JSON object"; 1766 dec.err = "garbage after JSON object";
1284 SvREFCNT_dec (sv); 1767 SvREFCNT_dec (sv);
1285 sv = 0; 1768 sv = 0;
1286 } 1769 }
1287 }
1288
1289 if (offset_return || !sv)
1290 {
1291 offset = dec.json.flags & F_UTF8
1292 ? dec.cur - SvPVX (string)
1293 : utf8_distance (dec.cur, SvPVX (string));
1294
1295 if (offset_return)
1296 *offset_return = offset;
1297 } 1770 }
1298 1771
1299 if (!sv) 1772 if (!sv)
1300 { 1773 {
1301 SV *uni = sv_newmortal (); 1774 SV *uni = sv_newmortal ();
… …
1307 SAVEVPTR (PL_curcop); 1780 SAVEVPTR (PL_curcop);
1308 PL_curcop = &cop; 1781 PL_curcop = &cop;
1309 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1782 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1310 LEAVE; 1783 LEAVE;
1311 1784
1312 croak ("%s, at character offset %d [\"%s\"]", 1785 croak ("%s, at character offset %d (before \"%s\")",
1313 dec.err, 1786 dec.err,
1314 (int)offset, 1787 (int)ptr_to_index (string, dec.cur),
1315 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1788 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1316 } 1789 }
1317 1790
1318 sv = sv_2mortal (sv); 1791 sv = sv_2mortal (sv);
1319 1792
1320 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1793 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1321 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1794 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1322 1795
1323 return sv; 1796 return sv;
1797}
1798
1799/////////////////////////////////////////////////////////////////////////////
1800// incremental parser
1801
1802static void
1803incr_parse (JSON *self)
1804{
1805 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1806
1807 // the state machine here is a bit convoluted and could be simplified a lot
1808 // but this would make it slower, so...
1809
1810 for (;;)
1811 {
1812 //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
1813 switch (self->incr_mode)
1814 {
1815 // only used for initial whitespace skipping
1816 case INCR_M_WS:
1817 for (;;)
1818 {
1819 if (*p > 0x20)
1820 {
1821 if (*p == '#')
1822 {
1823 self->incr_mode = INCR_M_C0;
1824 goto incr_m_c;
1825 }
1826 else
1827 {
1828 self->incr_mode = INCR_M_JSON;
1829 goto incr_m_json;
1830 }
1831 }
1832 else if (!*p)
1833 goto interrupt;
1834
1835 ++p;
1836 }
1837
1838 // skip a single char inside a string (for \\-processing)
1839 case INCR_M_BS:
1840 if (!*p)
1841 goto interrupt;
1842
1843 ++p;
1844 self->incr_mode = INCR_M_STR;
1845 goto incr_m_str;
1846
1847 // inside #-style comments
1848 case INCR_M_C0:
1849 case INCR_M_C1:
1850 incr_m_c:
1851 for (;;)
1852 {
1853 if (*p == '\n')
1854 {
1855 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1856 break;
1857 }
1858 else if (!*p)
1859 goto interrupt;
1860
1861 ++p;
1862 }
1863
1864 break;
1865
1866 // inside a string
1867 case INCR_M_STR:
1868 incr_m_str:
1869 for (;;)
1870 {
1871 if (*p == '"')
1872 {
1873 ++p;
1874 self->incr_mode = INCR_M_JSON;
1875
1876 if (!self->incr_nest)
1877 goto interrupt;
1878
1879 goto incr_m_json;
1880 }
1881 else if (*p == '\\')
1882 {
1883 ++p; // "virtually" consumes character after \
1884
1885 if (!*p) // if at end of string we have to switch modes
1886 {
1887 self->incr_mode = INCR_M_BS;
1888 goto interrupt;
1889 }
1890 }
1891 else if (!*p)
1892 goto interrupt;
1893
1894 ++p;
1895 }
1896
1897 // after initial ws, outside string
1898 case INCR_M_JSON:
1899 incr_m_json:
1900 for (;;)
1901 {
1902 switch (*p++)
1903 {
1904 case 0:
1905 --p;
1906 goto interrupt;
1907
1908 case 0x09:
1909 case 0x0a:
1910 case 0x0d:
1911 case 0x20:
1912 if (!self->incr_nest)
1913 {
1914 --p; // do not eat the whitespace, let the next round do it
1915 goto interrupt;
1916 }
1917 break;
1918
1919 case '"':
1920 self->incr_mode = INCR_M_STR;
1921 goto incr_m_str;
1922
1923 case '[':
1924 case '{':
1925 case '(':
1926 if (++self->incr_nest > self->max_depth)
1927 croak (ERR_NESTING_EXCEEDED);
1928 break;
1929
1930 case ']':
1931 case '}':
1932 if (--self->incr_nest <= 0)
1933 goto interrupt;
1934 break;
1935
1936 case ')':
1937 --self->incr_nest;
1938 break;
1939
1940 case '#':
1941 self->incr_mode = INCR_M_C1;
1942 goto incr_m_c;
1943 }
1944 }
1945 }
1946
1947 modechange:
1948 ;
1949 }
1950
1951interrupt:
1952 self->incr_pos = p - SvPVX (self->incr_text);
1953 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1954 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1324} 1955}
1325 1956
1326///////////////////////////////////////////////////////////////////////////// 1957/////////////////////////////////////////////////////////////////////////////
1327// XS interface functions 1958// XS interface functions
1328 1959
… …
1337 i >= '0' && i <= '9' ? i - '0' 1968 i >= '0' && i <= '9' ? i - '0'
1338 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1969 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1339 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1970 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1340 : -1; 1971 : -1;
1341 1972
1342 json_stash = gv_stashpv ("JSON::XS" , 1); 1973 json_stash = gv_stashpv ("JSON::XS" , 1);
1343 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1974 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1975 bool_true = get_bool ("Types::Serialiser::true");
1976 bool_false = get_bool ("Types::Serialiser::false");
1344 1977
1345 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1978 sv_json = newSVpv ("JSON", 0);
1346 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1979 SvREADONLY_on (sv_json);
1980
1981 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1347} 1982}
1348 1983
1349PROTOTYPES: DISABLE 1984PROTOTYPES: DISABLE
1985
1986void CLONE (...)
1987 CODE:
1988 json_stash = 0;
1989 bool_stash = 0;
1350 1990
1351void new (char *klass) 1991void new (char *klass)
1352 PPCODE: 1992 PPCODE:
1353{ 1993{
1354 SV *pv = NEWSV (0, sizeof (JSON)); 1994 SV *pv = NEWSV (0, sizeof (JSON));
1355 SvPOK_only (pv); 1995 SvPOK_only (pv);
1356 Zero (SvPVX (pv), 1, JSON); 1996 json_init ((JSON *)SvPVX (pv));
1357 ((JSON *)SvPVX (pv))->flags = F_DEFAULT; 1997 XPUSHs (sv_2mortal (sv_bless (
1358 XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash))); 1998 newRV_noinc (pv),
1999 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
2000 )));
1359} 2001}
1360 2002
1361void ascii (JSON *self, int enable = 1) 2003void ascii (JSON *self, int enable = 1)
1362 ALIAS: 2004 ALIAS:
1363 ascii = F_ASCII 2005 ascii = F_ASCII
… …
1370 pretty = F_PRETTY 2012 pretty = F_PRETTY
1371 allow_nonref = F_ALLOW_NONREF 2013 allow_nonref = F_ALLOW_NONREF
1372 shrink = F_SHRINK 2014 shrink = F_SHRINK
1373 allow_blessed = F_ALLOW_BLESSED 2015 allow_blessed = F_ALLOW_BLESSED
1374 convert_blessed = F_CONV_BLESSED 2016 convert_blessed = F_CONV_BLESSED
2017 relaxed = F_RELAXED
2018 allow_unknown = F_ALLOW_UNKNOWN
2019 allow_tags = F_ALLOW_TAGS
1375 PPCODE: 2020 PPCODE:
1376{ 2021{
1377 if (enable) 2022 if (enable)
1378 self->flags |= ix; 2023 self->flags |= ix;
1379 else 2024 else
1380 self->flags &= ~ix; 2025 self->flags &= ~ix;
1381 2026
1382 XPUSHs (ST (0)); 2027 XPUSHs (ST (0));
1383} 2028}
1384 2029
1385void max_depth (JSON *self, UV max_depth = 0x80000000UL) 2030void get_ascii (JSON *self)
2031 ALIAS:
2032 get_ascii = F_ASCII
2033 get_latin1 = F_LATIN1
2034 get_utf8 = F_UTF8
2035 get_indent = F_INDENT
2036 get_canonical = F_CANONICAL
2037 get_space_before = F_SPACE_BEFORE
2038 get_space_after = F_SPACE_AFTER
2039 get_allow_nonref = F_ALLOW_NONREF
2040 get_shrink = F_SHRINK
2041 get_allow_blessed = F_ALLOW_BLESSED
2042 get_convert_blessed = F_CONV_BLESSED
2043 get_relaxed = F_RELAXED
2044 get_allow_unknown = F_ALLOW_UNKNOWN
2045 get_allow_tags = F_ALLOW_TAGS
1386 PPCODE: 2046 PPCODE:
1387{ 2047 XPUSHs (boolSV (self->flags & ix));
1388 UV log2 = 0;
1389 2048
1390 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; 2049void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1391 2050 PPCODE:
1392 while ((1UL << log2) < max_depth) 2051 self->max_depth = max_depth;
1393 ++log2;
1394
1395 self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1396
1397 XPUSHs (ST (0)); 2052 XPUSHs (ST (0));
1398}
1399 2053
2054U32 get_max_depth (JSON *self)
2055 CODE:
2056 RETVAL = self->max_depth;
2057 OUTPUT:
2058 RETVAL
2059
1400void max_size (JSON *self, UV max_size = 0) 2060void max_size (JSON *self, U32 max_size = 0)
1401 PPCODE: 2061 PPCODE:
1402{ 2062 self->max_size = max_size;
1403 UV log2 = 0;
1404
1405 if (max_size > 0x80000000UL) max_size = 0x80000000UL;
1406 if (max_size == 1) max_size = 2;
1407
1408 while ((1UL << log2) < max_size)
1409 ++log2;
1410
1411 self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1412
1413 XPUSHs (ST (0)); 2063 XPUSHs (ST (0));
1414} 2064
2065int get_max_size (JSON *self)
2066 CODE:
2067 RETVAL = self->max_size;
2068 OUTPUT:
2069 RETVAL
1415 2070
1416void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) 2071void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1417 PPCODE: 2072 PPCODE:
1418{ 2073{
1419 SvREFCNT_dec (self->cb_object); 2074 SvREFCNT_dec (self->cb_object);
… …
1423} 2078}
1424 2079
1425void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) 2080void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1426 PPCODE: 2081 PPCODE:
1427{ 2082{
1428 if (!self->cb_sk_object) 2083 if (!self->cb_sk_object)
1429 self->cb_sk_object = newHV (); 2084 self->cb_sk_object = newHV ();
1430 2085
1431 if (SvOK (cb)) 2086 if (SvOK (cb))
1432 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); 2087 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1433 else 2088 else
… …
1444 XPUSHs (ST (0)); 2099 XPUSHs (ST (0));
1445} 2100}
1446 2101
1447void encode (JSON *self, SV *scalar) 2102void encode (JSON *self, SV *scalar)
1448 PPCODE: 2103 PPCODE:
1449 XPUSHs (encode_json (scalar, self)); 2104 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2105 XPUSHs (scalar);
1450 2106
1451void decode (JSON *self, SV *jsonstr) 2107void decode (JSON *self, SV *jsonstr)
1452 PPCODE: 2108 PPCODE:
1453 XPUSHs (decode_json (jsonstr, self, 0)); 2109 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2110 XPUSHs (jsonstr);
1454 2111
1455void decode_prefix (JSON *self, SV *jsonstr) 2112void decode_prefix (JSON *self, SV *jsonstr)
1456 PPCODE: 2113 PPCODE:
1457{ 2114{
1458 UV offset; 2115 SV *sv;
2116 STRLEN offset;
2117 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
1459 EXTEND (SP, 2); 2118 EXTEND (SP, 2);
1460 PUSHs (decode_json (jsonstr, self, &offset)); 2119 PUSHs (sv);
1461 PUSHs (sv_2mortal (newSVuv (offset))); 2120 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
2121}
2122
2123void incr_parse (JSON *self, SV *jsonstr = 0)
2124 PPCODE:
2125{
2126 if (!self->incr_text)
2127 self->incr_text = newSVpvn ("", 0);
2128
2129 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2130 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2131 if (DECODE_WANTS_OCTETS (self))
2132 {
2133 if (self->incr_pos)
2134 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2135 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2136
2137 sv_utf8_downgrade (self->incr_text, 0);
2138 }
2139 else
2140 {
2141 sv_utf8_upgrade (self->incr_text);
2142
2143 if (self->incr_pos)
2144 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2145 - (U8 *)SvPVX (self->incr_text);
2146 }
2147
2148 // append data, if any
2149 if (jsonstr)
2150 {
2151 /* make sure both strings have same encoding */
2152 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2153 if (SvUTF8 (jsonstr))
2154 sv_utf8_downgrade (jsonstr, 0);
2155 else
2156 sv_utf8_upgrade (jsonstr);
2157
2158 /* and then just blindly append */
2159 {
2160 STRLEN len;
2161 const char *str = SvPV (jsonstr, len);
2162 STRLEN cur = SvCUR (self->incr_text);
2163
2164 if (SvLEN (self->incr_text) <= cur + len)
2165 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
2166
2167 Move (str, SvEND (self->incr_text), len, char);
2168 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2169 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2170 }
2171 }
2172
2173 if (GIMME_V != G_VOID)
2174 do
2175 {
2176 SV *sv;
2177 STRLEN offset;
2178
2179 if (!INCR_DONE (self))
2180 {
2181 incr_parse (self);
2182
2183 if (self->incr_pos > self->max_size && self->max_size)
2184 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2185 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2186
2187 if (!INCR_DONE (self))
2188 {
2189 // as an optimisation, do not accumulate white space in the incr buffer
2190 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2191 {
2192 self->incr_pos = 0;
2193 SvCUR_set (self->incr_text, 0);
2194 }
2195
2196 break;
2197 }
2198 }
2199
2200 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2201 XPUSHs (sv);
2202
2203 self->incr_pos -= offset;
2204 self->incr_nest = 0;
2205 self->incr_mode = 0;
2206
2207 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
2208 }
2209 while (GIMME_V == G_ARRAY);
2210}
2211
2212SV *incr_text (JSON *self)
2213 ATTRS: lvalue
2214 CODE:
2215{
2216 if (self->incr_pos)
2217 croak ("incr_text can not be called when the incremental parser already started parsing");
2218
2219 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
2220}
2221 OUTPUT:
2222 RETVAL
2223
2224void incr_skip (JSON *self)
2225 CODE:
2226{
2227 if (self->incr_pos)
2228 {
2229 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2230 self->incr_pos = 0;
2231 self->incr_nest = 0;
2232 self->incr_mode = 0;
2233 }
2234}
2235
2236void incr_reset (JSON *self)
2237 CODE:
2238{
2239 SvREFCNT_dec (self->incr_text);
2240 self->incr_text = 0;
2241 self->incr_pos = 0;
2242 self->incr_nest = 0;
2243 self->incr_mode = 0;
1462} 2244}
1463 2245
1464void DESTROY (JSON *self) 2246void DESTROY (JSON *self)
1465 CODE: 2247 CODE:
1466 SvREFCNT_dec (self->cb_sk_object); 2248 SvREFCNT_dec (self->cb_sk_object);
1467 SvREFCNT_dec (self->cb_object); 2249 SvREFCNT_dec (self->cb_object);
2250 SvREFCNT_dec (self->incr_text);
1468 2251
1469PROTOTYPES: ENABLE 2252PROTOTYPES: ENABLE
1470 2253
1471void to_json (SV *scalar) 2254void encode_json (SV *scalar)
1472 PPCODE: 2255 PPCODE:
1473{ 2256{
1474 JSON json = { F_DEFAULT | F_UTF8 }; 2257 JSON json;
1475 XPUSHs (encode_json (scalar, &json)); 2258 json_init (&json);
2259 json.flags |= F_UTF8;
2260 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2261 XPUSHs (scalar);
1476} 2262}
1477 2263
1478void from_json (SV *jsonstr) 2264void decode_json (SV *jsonstr)
1479 PPCODE: 2265 PPCODE:
1480{ 2266{
1481 JSON json = { F_DEFAULT | F_UTF8 }; 2267 JSON json;
1482 XPUSHs (decode_json (jsonstr, &json, 0)); 2268 json_init (&json);
2269 json.flags |= F_UTF8;
2270 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2271 XPUSHs (jsonstr);
1483} 2272}
1484 2273

Diff Legend

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