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 1.46 by root, Mon Jun 25 22:11:39 2007 UTC vs.
Revision 1.131 by root, Thu Aug 17 01:42:20 2017 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
… …
25#define F_SPACE_BEFORE 0x00000020UL 42#define F_SPACE_BEFORE 0x00000020UL
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 // NYI 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
35 50#define F_ALLOW_TAGS 0x00004000UL
36#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) 51#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
37#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE ))
38 52
39#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 53#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
40#define F_DEFAULT (9UL << S_MAXDEPTH)
41 54
42#define INIT_SIZE 32 // initial scalar size to be allocated 55#define INIT_SIZE 32 // initial scalar size to be allocated
43#define INDENT_STEP 3 // spaces per indentation level 56#define INDENT_STEP 3 // spaces per indentation level
44 57
45#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
46 59
60#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
61
47#define SB do { 62#define SB do {
48#define SE } while (0) 63#define SE } while (0)
49 64
50#if __GNUC__ >= 3 65#if __GNUC__ >= 3
51# define expect(expr,value) __builtin_expect ((expr),(value)) 66# define expect(expr,value) __builtin_expect ((expr), (value))
52# define inline inline 67# define INLINE static inline
53#else 68#else
54# define expect(expr,value) (expr) 69# define expect(expr,value) (expr)
55# define inline static 70# define INLINE static
56#endif 71#endif
57 72
58#define expect_false(expr) expect ((expr) != 0, 0) 73#define expect_false(expr) expect ((expr) != 0, 0)
59#define expect_true(expr) expect ((expr) != 0, 1) 74#define expect_true(expr) expect ((expr) != 0, 1)
60 75
61static HV *json_stash, *json_boolean_stash; // JSON::XS:: 76#define IN_RANGE_INC(type,val,beg,end) \
62static 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)
108
109typedef struct {
110 U32 flags;
111 U32 max_depth;
112 STRLEN max_size;
113
114 SV *cb_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;
122} JSON;
123
124INLINE void
125json_init (JSON *json)
126{
127 Zero (json, 1, JSON);
128 json->max_depth = 512;
129}
63 130
64///////////////////////////////////////////////////////////////////////////// 131/////////////////////////////////////////////////////////////////////////////
65// utility functions 132// utility functions
66 133
67static UV * 134INLINE SV *
68SvJSON (SV *sv) 135get_bool (const char *name)
69{ 136{
70 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 137 SV *sv = get_sv (name, 1);
71 croak ("object is not of type JSON::XS");
72 138
73 return &SvUVX (SvRV (sv)); 139 SvREADONLY_on (sv);
74} 140 SvREADONLY_on (SvRV (sv));
75 141
76static void 142 return sv;
143}
144
145INLINE void
77shrink (SV *sv) 146shrink (SV *sv)
78{ 147{
79 sv_utf8_downgrade (sv, 1); 148 sv_utf8_downgrade (sv, 1);
149
80 if (SvLEN (sv) > SvCUR (sv) + 1) 150 if (SvLEN (sv) > SvCUR (sv) + 1)
81 { 151 {
82#ifdef SvPV_shrink_to_cur 152#ifdef SvPV_shrink_to_cur
83 SvPV_shrink_to_cur (sv); 153 SvPV_shrink_to_cur (sv);
84#elif defined (SvPV_renew) 154#elif defined (SvPV_renew)
85 SvPV_renew (sv, SvCUR (sv) + 1); 155 SvPV_renew (sv, SvCUR (sv) + 1);
86#endif 156#endif
87 } 157 }
88} 158}
89 159
160/* adds two STRLENs together, slow, and with paranoia */
161STRLEN
162strlen_sum (STRLEN l1, STRLEN l2)
163{
164 size_t sum = l1 + l2;
165
166 if (sum < (size_t)l2 || sum != (size_t)(STRLEN)sum)
167 croak ("JSON::XS: string size overflow");
168
169 return sum;
170}
171
172/* similar to SvGROW, but somewhat safer and guarantees exponential realloc strategy */
173static char *
174json_sv_grow (SV *sv, size_t len1, size_t len2)
175{
176 len1 = strlen_sum (len1, len2);
177 len1 = strlen_sum (len1, len1 >> 1);
178
179 if (len1 > 4096 - 24)
180 len1 = (len1 | 4095) - 24;
181
182 return SvGROW (sv, len1);
183}
184
90// decode an utf-8 character and return it, or (UV)-1 in 185// decode an utf-8 character and return it, or (UV)-1 in
91// case of an error. 186// case of an error.
92// we special-case "safe" characters from U+80 .. U+7FF, 187// we special-case "safe" characters from U+80 .. U+7FF,
93// but use the very good perl function to parse anything else. 188// but use the very good perl function to parse anything else.
94// note that we never call this function for a ascii codepoints 189// note that we never call this function for a ascii codepoints
95inline UV 190INLINE UV
96decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 191decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
97{ 192{
98 if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) 193 if (expect_true (len >= 2
99 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 194 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
100 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 195 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
101 { 196 {
102 *clen = 2; 197 *clen = 2;
103 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 198 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
104 } 199 }
105 else 200 else
201 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
202}
203
204// likewise for encoding, also never called for ascii codepoints
205// this function takes advantage of this fact, although current gccs
206// seem to optimise the check for >= 0x80 away anyways
207INLINE unsigned char *
208encode_utf8 (unsigned char *s, UV ch)
209{
210 if (expect_false (ch < 0x000080))
211 *s++ = ch;
212 else if (expect_true (ch < 0x000800))
213 *s++ = 0xc0 | ( ch >> 6),
214 *s++ = 0x80 | ( ch & 0x3f);
215 else if ( ch < 0x010000)
216 *s++ = 0xe0 | ( ch >> 12),
217 *s++ = 0x80 | ((ch >> 6) & 0x3f),
218 *s++ = 0x80 | ( ch & 0x3f);
219 else if ( ch < 0x110000)
220 *s++ = 0xf0 | ( ch >> 18),
221 *s++ = 0x80 | ((ch >> 12) & 0x3f),
222 *s++ = 0x80 | ((ch >> 6) & 0x3f),
223 *s++ = 0x80 | ( ch & 0x3f);
224
225 return s;
226}
227
228// convert offset pointer to character index, sv must be string
229static STRLEN
230ptr_to_index (SV *sv, char *offset)
231{
232 return SvUTF8 (sv)
233 ? utf8_distance (offset, SvPVX (sv))
234 : offset - SvPVX (sv);
235}
236
237/////////////////////////////////////////////////////////////////////////////
238// fp hell
239
240// scan a group of digits, and a trailing exponent
241static void
242json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
243{
244 UV uaccum = 0;
245 int eaccum = 0;
246
247 // if we recurse too deep, skip all remaining digits
248 // to avoid a stack overflow attack
249 if (expect_false (--maxdepth <= 0))
250 while (((U8)*s - '0') < 10)
251 ++s;
252
253 for (;;)
254 {
255 U8 dig = (U8)*s - '0';
256
257 if (expect_false (dig >= 10))
258 {
259 if (dig == (U8)((U8)'.' - (U8)'0'))
260 {
261 ++s;
262 json_atof_scan1 (s, accum, expo, 1, maxdepth);
263 }
264 else if ((dig | ' ') == 'e' - '0')
265 {
266 int exp2 = 0;
267 int neg = 0;
268
269 ++s;
270
271 if (*s == '-')
272 {
273 ++s;
274 neg = 1;
275 }
276 else if (*s == '+')
277 ++s;
278
279 while ((dig = (U8)*s - '0') < 10)
280 exp2 = exp2 * 10 + *s++ - '0';
281
282 *expo += neg ? -exp2 : exp2;
283 }
284
285 break;
286 }
287
288 ++s;
289
290 uaccum = uaccum * 10 + dig;
291 ++eaccum;
292
293 // if we have too many digits, then recurse for more
294 // we actually do this for rather few digits
295 if (uaccum >= (UV_MAX - 9) / 10)
296 {
297 if (postdp) *expo -= eaccum;
298 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
299 if (postdp) *expo += eaccum;
300
301 break;
302 }
106 { 303 }
107 *clen = (STRLEN)-1; 304
305 // this relies greatly on the quality of the pow ()
306 // implementation of the platform, but a good
307 // implementation is hard to beat.
308 // (IEEE 754 conformant ones are required to be exact)
309 if (postdp) *expo -= eaccum;
310 *accum += uaccum * Perl_pow (10., *expo);
311 *expo += eaccum;
312}
313
314static NV
315json_atof (const char *s)
316{
317 NV accum = 0.;
318 int expo = 0;
319 int neg = 0;
320
321 if (*s == '-')
322 {
323 ++s;
324 neg = 1;
325 }
326
327 // a recursion depth of ten gives us >>500 bits
328 json_atof_scan1 (s, &accum, &expo, 0, 10);
329
330 return neg ? -accum : accum;
331}
332
333// target of scalar reference is bool? -1 == nope, 0 == false, 1 == true
334static int
335ref_bool_type (SV *sv)
336{
337 svtype svt = SvTYPE (sv);
338
339 if (svt < SVt_PVAV)
340 {
341 STRLEN len = 0;
342 char *pv = svt ? SvPV (sv, len) : 0;
343
344 if (len == 1)
345 if (*pv == '1')
346 return 1;
347 else if (*pv == '0')
348 return 0;
349 }
350
351 return -1;
352}
353
354// returns whether scalar is not a reference in the sense of allow_nonref
355static int
356json_nonref (SV *scalar)
357{
358 if (!SvROK (scalar))
359 return 1;
360
361 scalar = SvRV (scalar);
362
363 if (SvTYPE (scalar) >= SVt_PVMG)
364 {
365 if (SvSTASH (scalar) == bool_stash)
108 return (UV)-1; 366 return 1;
367
368 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
369 return 1;
109 } 370 }
371
372 return 0;
110} 373}
111 374
112///////////////////////////////////////////////////////////////////////////// 375/////////////////////////////////////////////////////////////////////////////
113// encoder 376// encoder
114 377
… …
116typedef struct 379typedef struct
117{ 380{
118 char *cur; // SvPVX (sv) + current output position 381 char *cur; // SvPVX (sv) + current output position
119 char *end; // SvEND (sv) 382 char *end; // SvEND (sv)
120 SV *sv; // result scalar 383 SV *sv; // result scalar
121 U32 flags; // F_* 384 JSON json;
122 U32 indent; // indentation level 385 U32 indent; // indentation level
123 U32 maxdepth; // max. indentation/recursion level 386 UV limit; // escape character values >= this value when encoding
124} enc_t; 387} enc_t;
125 388
126inline void 389INLINE void
127need (enc_t *enc, STRLEN len) 390need (enc_t *enc, STRLEN len)
128{ 391{
129 if (expect_false (enc->cur + len >= enc->end)) 392 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
130 { 393 {
131 STRLEN cur = enc->cur - SvPVX (enc->sv); 394 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
132 SvGROW (enc->sv, cur + len + 1); 395 char *buf = json_sv_grow (enc->sv, cur, len);
133 enc->cur = SvPVX (enc->sv) + cur; 396 enc->cur = buf + cur;
134 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 397 enc->end = buf + SvLEN (enc->sv) - 1;
135 } 398 }
136} 399}
137 400
138inline void 401INLINE void
139encode_ch (enc_t *enc, char ch) 402encode_ch (enc_t *enc, char ch)
140{ 403{
141 need (enc, 1); 404 need (enc, 1);
142 *enc->cur++ = ch; 405 *enc->cur++ = ch;
143} 406}
… …
155 418
156 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case 419 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
157 { 420 {
158 if (expect_false (ch == '"')) // but with slow exceptions 421 if (expect_false (ch == '"')) // but with slow exceptions
159 { 422 {
160 need (enc, len += 1); 423 need (enc, len + 1);
161 *enc->cur++ = '\\'; 424 *enc->cur++ = '\\';
162 *enc->cur++ = '"'; 425 *enc->cur++ = '"';
163 } 426 }
164 else if (expect_false (ch == '\\')) 427 else if (expect_false (ch == '\\'))
165 { 428 {
166 need (enc, len += 1); 429 need (enc, len + 1);
167 *enc->cur++ = '\\'; 430 *enc->cur++ = '\\';
168 *enc->cur++ = '\\'; 431 *enc->cur++ = '\\';
169 } 432 }
170 else 433 else
171 *enc->cur++ = ch; 434 *enc->cur++ = ch;
… …
174 } 437 }
175 else 438 else
176 { 439 {
177 switch (ch) 440 switch (ch)
178 { 441 {
179 case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; 442 case '\010': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break;
180 case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; 443 case '\011': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break;
181 case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; 444 case '\012': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break;
182 case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; 445 case '\014': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break;
183 case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; 446 case '\015': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break;
184 447
185 default: 448 default:
186 { 449 {
187 STRLEN clen; 450 STRLEN clen;
188 UV uch; 451 UV uch;
… …
197 { 460 {
198 uch = ch; 461 uch = ch;
199 clen = 1; 462 clen = 1;
200 } 463 }
201 464
202 if (uch > 0x10FFFFUL) 465 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
203 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
204
205 if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF))
206 { 466 {
207 if (uch > 0xFFFFUL) 467 if (uch >= 0x10000UL)
208 { 468 {
469 if (uch >= 0x110000UL)
470 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
471
209 need (enc, len += 11); 472 need (enc, len + 11);
210 sprintf (enc->cur, "\\u%04x\\u%04x", 473 sprintf (enc->cur, "\\u%04x\\u%04x",
211 (int)((uch - 0x10000) / 0x400 + 0xD800), 474 (int)((uch - 0x10000) / 0x400 + 0xD800),
212 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 475 (int)((uch - 0x10000) % 0x400 + 0xDC00));
213 enc->cur += 12; 476 enc->cur += 12;
214 } 477 }
215 else 478 else
216 { 479 {
217 static char hexdigit [16] = "0123456789abcdef";
218 need (enc, len += 5); 480 need (enc, len + 5);
219 *enc->cur++ = '\\'; 481 *enc->cur++ = '\\';
220 *enc->cur++ = 'u'; 482 *enc->cur++ = 'u';
221 *enc->cur++ = hexdigit [ uch >> 12 ]; 483 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
222 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 484 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
223 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 485 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
224 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 486 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
225 } 487 }
226 488
227 str += clen; 489 str += clen;
228 } 490 }
229 else if (enc->flags & F_LATIN1) 491 else if (enc->json.flags & F_LATIN1)
230 { 492 {
231 *enc->cur++ = uch; 493 *enc->cur++ = uch;
232 str += clen; 494 str += clen;
233 } 495 }
234 else if (is_utf8) 496 else if (is_utf8)
235 { 497 {
236 need (enc, len += clen); 498 need (enc, len + clen);
237 do 499 do
238 { 500 {
239 *enc->cur++ = *str++; 501 *enc->cur++ = *str++;
240 } 502 }
241 while (--clen); 503 while (--clen);
242 } 504 }
243 else 505 else
244 { 506 {
245 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed 507 need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed
246 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 508 enc->cur = encode_utf8 (enc->cur, uch);
247 ++str; 509 ++str;
248 } 510 }
249 } 511 }
250 } 512 }
251 } 513 }
252 514
253 --len; 515 --len;
254 } 516 }
255} 517}
256 518
257inline void 519INLINE void
258encode_indent (enc_t *enc) 520encode_indent (enc_t *enc)
259{ 521{
260 if (enc->flags & F_INDENT) 522 if (enc->json.flags & F_INDENT)
261 { 523 {
262 int spaces = enc->indent * INDENT_STEP; 524 int spaces = enc->indent * INDENT_STEP;
263 525
264 need (enc, spaces); 526 need (enc, spaces);
265 memset (enc->cur, ' ', spaces); 527 memset (enc->cur, ' ', spaces);
266 enc->cur += spaces; 528 enc->cur += spaces;
267 } 529 }
268} 530}
269 531
270inline void 532INLINE void
271encode_space (enc_t *enc) 533encode_space (enc_t *enc)
272{ 534{
273 need (enc, 1); 535 need (enc, 1);
274 encode_ch (enc, ' '); 536 encode_ch (enc, ' ');
275} 537}
276 538
277inline void 539INLINE void
278encode_nl (enc_t *enc) 540encode_nl (enc_t *enc)
279{ 541{
280 if (enc->flags & F_INDENT) 542 if (enc->json.flags & F_INDENT)
281 { 543 {
282 need (enc, 1); 544 need (enc, 1);
283 encode_ch (enc, '\n'); 545 encode_ch (enc, '\n');
284 } 546 }
285} 547}
286 548
287inline void 549INLINE void
288encode_comma (enc_t *enc) 550encode_comma (enc_t *enc)
289{ 551{
290 encode_ch (enc, ','); 552 encode_ch (enc, ',');
291 553
292 if (enc->flags & F_INDENT) 554 if (enc->json.flags & F_INDENT)
293 encode_nl (enc); 555 encode_nl (enc);
294 else if (enc->flags & F_SPACE_AFTER) 556 else if (enc->json.flags & F_SPACE_AFTER)
295 encode_space (enc); 557 encode_space (enc);
296} 558}
297 559
298static void encode_sv (enc_t *enc, SV *sv); 560static void encode_sv (enc_t *enc, SV *sv);
299 561
300static void 562static void
301encode_av (enc_t *enc, AV *av) 563encode_av (enc_t *enc, AV *av)
302{ 564{
303 int i, len = av_len (av); 565 int i, len = av_len (av);
304 566
305 if (enc->indent >= enc->maxdepth) 567 if (enc->indent >= enc->json.max_depth)
306 croak ("data structure too deep (hit recursion limit)"); 568 croak (ERR_NESTING_EXCEEDED);
307 569
308 encode_ch (enc, '['); encode_nl (enc); 570 encode_ch (enc, '[');
309 ++enc->indent;
310 571
572 if (len >= 0)
573 {
574 encode_nl (enc); ++enc->indent;
575
311 for (i = 0; i <= len; ++i) 576 for (i = 0; i <= len; ++i)
312 { 577 {
578 SV **svp = av_fetch (av, i, 0);
579
313 encode_indent (enc); 580 encode_indent (enc);
314 encode_sv (enc, *av_fetch (av, i, 0));
315 581
582 if (svp)
583 encode_sv (enc, *svp);
584 else
585 encode_str (enc, "null", 4, 0);
586
316 if (i < len) 587 if (i < len)
317 encode_comma (enc); 588 encode_comma (enc);
318 } 589 }
319 590
591 encode_nl (enc); --enc->indent; encode_indent (enc);
592 }
593
320 encode_nl (enc); 594 encode_ch (enc, ']');
321
322 --enc->indent;
323 encode_indent (enc); encode_ch (enc, ']');
324} 595}
325 596
326static void 597static void
327encode_he (enc_t *enc, HE *he) 598encode_hk (enc_t *enc, HE *he)
328{ 599{
329 encode_ch (enc, '"'); 600 encode_ch (enc, '"');
330 601
331 if (HeKLEN (he) == HEf_SVKEY) 602 if (HeKLEN (he) == HEf_SVKEY)
332 { 603 {
333 SV *sv = HeSVKEY (he); 604 SV *sv = HeSVKEY (he);
334 STRLEN len; 605 STRLEN len;
335 char *str; 606 char *str;
336 607
337 SvGETMAGIC (sv); 608 SvGETMAGIC (sv);
338 str = SvPV (sv, len); 609 str = SvPV (sv, len);
339 610
340 encode_str (enc, str, len, SvUTF8 (sv)); 611 encode_str (enc, str, len, SvUTF8 (sv));
341 } 612 }
342 else 613 else
343 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 614 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
344 615
345 encode_ch (enc, '"'); 616 encode_ch (enc, '"');
346 617
347 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 618 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
348 encode_ch (enc, ':'); 619 encode_ch (enc, ':');
349 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 620 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
350 encode_sv (enc, HeVAL (he));
351} 621}
352 622
353// compare hash entries, used when all keys are bytestrings 623// compare hash entries, used when all keys are bytestrings
354static int 624static int
355he_cmp_fast (const void *a_, const void *b_) 625he_cmp_fast (const void *a_, const void *b_)
… …
360 HE *b = *(HE **)b_; 630 HE *b = *(HE **)b_;
361 631
362 STRLEN la = HeKLEN (a); 632 STRLEN la = HeKLEN (a);
363 STRLEN lb = HeKLEN (b); 633 STRLEN lb = HeKLEN (b);
364 634
365 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 635 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
366 cmp = la - lb; 636 cmp = lb - la;
367 637
368 return cmp; 638 return cmp;
369} 639}
370 640
371// compare hash entries, used when some keys are sv's or utf-x 641// compare hash entries, used when some keys are sv's or utf-x
372static int 642static int
373he_cmp_slow (const void *a, const void *b) 643he_cmp_slow (const void *a, const void *b)
374{ 644{
375 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 645 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
376} 646}
377 647
378static void 648static void
379encode_hv (enc_t *enc, HV *hv) 649encode_hv (enc_t *enc, HV *hv)
380{ 650{
381 int count, i; 651 HE *he;
382 652
383 if (enc->indent >= enc->maxdepth) 653 if (enc->indent >= enc->json.max_depth)
384 croak ("data structure too deep (hit recursion limit)"); 654 croak (ERR_NESTING_EXCEEDED);
385 655
386 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 656 encode_ch (enc, '{');
387 657
388 if ((count = hv_iterinit (hv)))
389 {
390 // for canonical output we have to sort by keys first 658 // for canonical output we have to sort by keys first
391 // actually, this is mostly due to the stupid so-called 659 // actually, this is mostly due to the stupid so-called
392 // security workaround added somewhere in 5.8.x. 660 // security workaround added somewhere in 5.8.x
393 // that randomises hash orderings 661 // that randomises hash orderings
394 if (enc->flags & F_CANONICAL) 662 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
663 {
664 int count = hv_iterinit (hv);
665
666 if (SvMAGICAL (hv))
395 { 667 {
668 // need to count by iterating. could improve by dynamically building the vector below
669 // but I don't care for the speed of this special case.
670 // note also that we will run into undefined behaviour when the two iterations
671 // do not result in the same count, something I might care for in some later release.
672
673 count = 0;
674 while (hv_iternext (hv))
675 ++count;
676
677 hv_iterinit (hv);
678 }
679
680 if (count)
681 {
396 int fast = 1; 682 int i, fast = 1;
397 HE *he; 683 HE *hes_stack [STACK_HES];
398#if defined(__BORLANDC__) || defined(_MSC_VER) 684 HE **hes = hes_stack;
399 HE **hes = _alloca (count * sizeof (HE)); 685
400#else 686 // allocate larger arrays on the heap
401 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode 687 if (count > STACK_HES)
402#endif 688 {
689 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
690 hes = (HE **)SvPVX (sv);
691 }
403 692
404 i = 0; 693 i = 0;
405 while ((he = hv_iternext (hv))) 694 while ((he = hv_iternext (hv)))
406 { 695 {
407 hes [i++] = he; 696 hes [i++] = he;
… …
429 718
430 FREETMPS; 719 FREETMPS;
431 LEAVE; 720 LEAVE;
432 } 721 }
433 722
434 for (i = 0; i < count; ++i) 723 encode_nl (enc); ++enc->indent;
724
725 while (count--)
435 { 726 {
436 encode_indent (enc); 727 encode_indent (enc);
728 he = hes [count];
437 encode_he (enc, hes [i]); 729 encode_hk (enc, he);
730 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
438 731
439 if (i < count - 1) 732 if (count)
440 encode_comma (enc); 733 encode_comma (enc);
441 } 734 }
442 735
443 encode_nl (enc); 736 encode_nl (enc); --enc->indent; encode_indent (enc);
444 } 737 }
738 }
445 else 739 else
740 {
741 if (hv_iterinit (hv) || SvMAGICAL (hv))
742 if ((he = hv_iternext (hv)))
446 { 743 {
447 HE *he = hv_iternext (hv); 744 encode_nl (enc); ++enc->indent;
448 745
449 for (;;) 746 for (;;)
450 { 747 {
451 encode_indent (enc); 748 encode_indent (enc);
452 encode_he (enc, he); 749 encode_hk (enc, he);
750 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
453 751
454 if (!(he = hv_iternext (hv))) 752 if (!(he = hv_iternext (hv)))
455 break; 753 break;
456 754
457 encode_comma (enc); 755 encode_comma (enc);
458 } 756 }
459 757
460 encode_nl (enc); 758 encode_nl (enc); --enc->indent; encode_indent (enc);
461 } 759 }
462 } 760 }
463 761
464 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 762 encode_ch (enc, '}');
465} 763}
466 764
467// encode objects, arrays and special \0=false and \1=true values. 765// encode objects, arrays and special \0=false and \1=true values.
468static void 766static void
469encode_rv (enc_t *enc, SV *sv) 767encode_rv (enc_t *enc, SV *sv)
470{ 768{
471 svtype svt; 769 svtype svt;
770 GV *method;
472 771
473 SvGETMAGIC (sv); 772 SvGETMAGIC (sv);
474 svt = SvTYPE (sv); 773 svt = SvTYPE (sv);
475 774
476 if (expect_false (SvOBJECT (sv))) 775 if (expect_false (SvOBJECT (sv)))
477 { 776 {
478 if (SvSTASH (sv) == json_boolean_stash) 777 HV *stash = SvSTASH (sv);
778
779 if (stash == bool_stash)
479 { 780 {
480 if (SvIV (sv) == 0) 781 if (SvIV (sv))
782 encode_str (enc, "true", 4, 0);
783 else
481 encode_str (enc, "false", 5, 0); 784 encode_str (enc, "false", 5, 0);
482 else
483 encode_str (enc, "true", 4, 0);
484 } 785 }
786 else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0)))
787 {
788 int count;
789 dSP;
790
791 ENTER; SAVETMPS;
792 SAVESTACK_POS ();
793 PUSHMARK (SP);
794 EXTEND (SP, 2);
795 // we re-bless the reference to get overload and other niceties right
796 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
797 PUSHs (sv_json);
798
799 PUTBACK;
800 count = call_sv ((SV *)GvCV (method), G_ARRAY);
801 SPAGAIN;
802
803 // catch this surprisingly common error
804 if (SvROK (TOPs) && SvRV (TOPs) == sv)
805 croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
806
807 encode_ch (enc, '(');
808 encode_ch (enc, '"');
809 encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash));
810 encode_ch (enc, '"');
811 encode_ch (enc, ')');
812 encode_ch (enc, '[');
813
814 while (count)
815 {
816 encode_sv (enc, SP[1 - count--]);
817
818 if (count)
819 encode_ch (enc, ',');
820 }
821
822 encode_ch (enc, ']');
823
824 FREETMPS; LEAVE;
825 }
826 else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0)))
827 {
828 dSP;
829
830 ENTER; SAVETMPS;
831 PUSHMARK (SP);
832 // we re-bless the reference to get overload and other niceties right
833 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
834
835 // calling with G_SCALAR ensures that we always get a 1 return value
836 PUTBACK;
837 call_sv ((SV *)GvCV (method), G_SCALAR);
838 SPAGAIN;
839
840 // catch this surprisingly common error
841 if (SvROK (TOPs) && SvRV (TOPs) == sv)
842 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
843
844 sv = POPs;
845 PUTBACK;
846
847 encode_sv (enc, sv);
848
849 FREETMPS; LEAVE;
850 }
851 else if (enc->json.flags & F_ALLOW_BLESSED)
852 encode_str (enc, "null", 4, 0);
485 else 853 else
486 {
487#if 0
488 if (0 && sv_derived_from (rv, "JSON::Literal"))
489 {
490 // not yet
491 }
492#endif
493 if (enc->flags & F_CONV_BLESSED)
494 {
495 // we re-bless the reference to get overload and other niceties right
496 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1);
497
498 if (to_json)
499 {
500 dSP;
501 ENTER;
502 SAVETMPS;
503 PUSHMARK (SP);
504 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
505
506 // calling with G_SCALAR ensures that we always get a 1 reutrn value
507 // check anyways.
508 PUTBACK;
509 assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR));
510 SPAGAIN;
511
512 encode_sv (enc, POPs);
513
514 FREETMPS;
515 LEAVE;
516 }
517 else if (enc->flags & F_ALLOW_BLESSED)
518 encode_str (enc, "null", 4, 0);
519 else
520 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
521 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
522 }
523 else if (enc->flags & F_ALLOW_BLESSED)
524 encode_str (enc, "null", 4, 0);
525 else
526 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", 854 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
527 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 855 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
528 }
529 } 856 }
530 else if (svt == SVt_PVHV) 857 else if (svt == SVt_PVHV)
531 encode_hv (enc, (HV *)sv); 858 encode_hv (enc, (HV *)sv);
532 else if (svt == SVt_PVAV) 859 else if (svt == SVt_PVAV)
533 encode_av (enc, (AV *)sv); 860 encode_av (enc, (AV *)sv);
534 else if (svt < SVt_PVAV) 861 else if (svt < SVt_PVAV)
535 { 862 {
536 if (SvNIOK (sv) && SvIV (sv) == 0) 863 int bool_type = ref_bool_type (sv);
864
865 if (bool_type == 1)
866 encode_str (enc, "true", 4, 0);
867 else if (bool_type == 0)
537 encode_str (enc, "false", 5, 0); 868 encode_str (enc, "false", 5, 0);
538 else if (SvNIOK (sv) && SvIV (sv) == 1) 869 else if (enc->json.flags & F_ALLOW_UNKNOWN)
539 encode_str (enc, "true", 4, 0); 870 encode_str (enc, "null", 4, 0);
540 else 871 else
541 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", 872 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
542 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 873 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
543 } 874 }
875 else if (enc->json.flags & F_ALLOW_UNKNOWN)
876 encode_str (enc, "null", 4, 0);
544 else 877 else
545 croak ("encountered %s, but JSON can only represent references to arrays or hashes", 878 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
546 SvPV_nolen (sv_2mortal (newRV_inc (sv)))); 879 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
547} 880}
548 881
… …
566 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 899 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
567 enc->cur += strlen (enc->cur); 900 enc->cur += strlen (enc->cur);
568 } 901 }
569 else if (SvIOKp (sv)) 902 else if (SvIOKp (sv))
570 { 903 {
571 // we assume we can always read an IV as a UV 904 // we assume we can always read an IV as a UV and vice versa
572 if (SvUV (sv) & ~(UV)0x7fff) 905 // we assume two's complement
573 { 906 // we assume no aliasing issues in the union
574 // large integer, use the (rather slow) snprintf way. 907 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
575 need (enc, sizeof (UV) * 3); 908 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
576 enc->cur +=
577 SvIsUV(sv)
578 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
579 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
580 }
581 else
582 { 909 {
583 // optimise the "small number case" 910 // optimise the "small number case"
584 // code will likely be branchless and use only a single multiplication 911 // code will likely be branchless and use only a single multiplication
912 // works for numbers up to 59074
585 I32 i = SvIV (sv); 913 I32 i = SvIVX (sv);
586 U32 u; 914 U32 u;
587 char digit, nz = 0; 915 char digit, nz = 0;
588 916
589 need (enc, 6); 917 need (enc, 6);
590 918
… …
596 924
597 // now output digit by digit, each time masking out the integer part 925 // now output digit by digit, each time masking out the integer part
598 // and multiplying by 5 while moving the decimal point one to the right, 926 // and multiplying by 5 while moving the decimal point one to the right,
599 // resulting in a net multiplication by 10. 927 // resulting in a net multiplication by 10.
600 // we always write the digit to memory but conditionally increment 928 // we always write the digit to memory but conditionally increment
601 // the pointer, to ease the usage of conditional move instructions. 929 // the pointer, to enable the use of conditional move instructions.
602 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 930 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
603 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 931 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
604 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 932 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
605 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 933 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
606 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 934 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
607 } 935 }
936 else
937 {
938 // large integer, use the (rather slow) snprintf way.
939 need (enc, IVUV_MAXCHARS);
940 enc->cur +=
941 SvIsUV(sv)
942 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
943 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
944 }
608 } 945 }
609 else if (SvROK (sv)) 946 else if (SvROK (sv))
610 encode_rv (enc, SvRV (sv)); 947 encode_rv (enc, SvRV (sv));
611 else if (!SvOK (sv)) 948 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
612 encode_str (enc, "null", 4, 0); 949 encode_str (enc, "null", 4, 0);
613 else 950 else
614 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 951 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data",
615 SvPV_nolen (sv), SvFLAGS (sv)); 952 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
616} 953}
617 954
618static SV * 955static SV *
619encode_json (SV *scalar, U32 flags) 956encode_json (SV *scalar, JSON *json)
620{ 957{
621 enc_t enc; 958 enc_t enc;
622 959
623 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 960 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
624 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 961 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
625 962
626 enc.flags = flags; 963 enc.json = *json;
627 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 964 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
628 enc.cur = SvPVX (enc.sv); 965 enc.cur = SvPVX (enc.sv);
629 enc.end = SvEND (enc.sv); 966 enc.end = SvEND (enc.sv);
630 enc.indent = 0; 967 enc.indent = 0;
631 enc.maxdepth = DEC_DEPTH (flags); 968 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
969 : enc.json.flags & F_LATIN1 ? 0x000100UL
970 : 0x110000UL;
632 971
633 SvPOK_only (enc.sv); 972 SvPOK_only (enc.sv);
634 encode_sv (&enc, scalar); 973 encode_sv (&enc, scalar);
974 encode_nl (&enc);
635 975
636 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 976 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
637 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 977 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
638 978
639 if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) 979 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
640 SvUTF8_on (enc.sv); 980 SvUTF8_on (enc.sv);
641 981
642 if (enc.flags & F_SHRINK) 982 if (enc.json.flags & F_SHRINK)
643 shrink (enc.sv); 983 shrink (enc.sv);
644 984
645 return enc.sv; 985 return enc.sv;
646} 986}
647 987
… …
652typedef struct 992typedef struct
653{ 993{
654 char *cur; // current parser pointer 994 char *cur; // current parser pointer
655 char *end; // end of input string 995 char *end; // end of input string
656 const char *err; // parse error, if != 0 996 const char *err; // parse error, if != 0
657 U32 flags; // F_* 997 JSON json;
658 U32 depth; // recursion depth 998 U32 depth; // recursion depth
659 U32 maxdepth; // recursion depth limit 999 U32 maxdepth; // recursion depth limit
660} dec_t; 1000} dec_t;
661 1001
662inline void 1002INLINE void
1003decode_comment (dec_t *dec)
1004{
1005 // only '#'-style comments allowed a.t.m.
1006
1007 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
1008 ++dec->cur;
1009}
1010
1011INLINE void
663decode_ws (dec_t *dec) 1012decode_ws (dec_t *dec)
664{ 1013{
665 for (;;) 1014 for (;;)
666 { 1015 {
667 char ch = *dec->cur; 1016 char ch = *dec->cur;
668 1017
669 if (ch > 0x20 1018 if (ch > 0x20)
1019 {
1020 if (expect_false (ch == '#'))
1021 {
1022 if (dec->json.flags & F_RELAXED)
1023 decode_comment (dec);
1024 else
1025 break;
1026 }
1027 else
1028 break;
1029 }
670 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 1030 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
671 break; 1031 break; // parse error, but let higher level handle it, gives better error messages
672 1032
673 ++dec->cur; 1033 ++dec->cur;
674 } 1034 }
675} 1035}
676 1036
… …
680 if (*dec->cur != ch) \ 1040 if (*dec->cur != ch) \
681 ERR (# ch " expected"); \ 1041 ERR (# ch " expected"); \
682 ++dec->cur; \ 1042 ++dec->cur; \
683 SE 1043 SE
684 1044
685#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") 1045#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
686#define DEC_DEC_DEPTH --dec->depth 1046#define DEC_DEC_DEPTH --dec->depth
687 1047
688static SV *decode_sv (dec_t *dec); 1048static SV *decode_sv (dec_t *dec);
689 1049
690static signed char decode_hexdigit[256]; 1050static signed char decode_hexdigit[256];
… …
782 1142
783 if (hi >= 0x80) 1143 if (hi >= 0x80)
784 { 1144 {
785 utf8 = 1; 1145 utf8 = 1;
786 1146
787 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1147 cur = encode_utf8 (cur, hi);
788 } 1148 }
789 else 1149 else
790 *cur++ = hi; 1150 *cur++ = hi;
791 } 1151 }
792 break; 1152 break;
… …
794 default: 1154 default:
795 --dec_cur; 1155 --dec_cur;
796 ERR ("illegal backslash escape sequence in string"); 1156 ERR ("illegal backslash escape sequence in string");
797 } 1157 }
798 } 1158 }
799 else if (expect_true (ch >= 0x20 && ch <= 0x7f)) 1159 else if (expect_true (ch >= 0x20 && ch < 0x80))
800 *cur++ = ch; 1160 *cur++ = ch;
801 else if (ch >= 0x80) 1161 else if (ch >= 0x80)
802 { 1162 {
803 STRLEN clen; 1163 STRLEN clen;
804 UV uch;
805 1164
806 --dec_cur; 1165 --dec_cur;
807 1166
808 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1167 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
809 if (clen == (STRLEN)-1) 1168 if (clen == (STRLEN)-1)
810 ERR ("malformed UTF-8 character in JSON string"); 1169 ERR ("malformed UTF-8 character in JSON string");
811 1170
812 do 1171 do
813 *cur++ = *dec_cur++; 1172 *cur++ = *dec_cur++;
814 while (--clen); 1173 while (--clen);
815 1174
816 utf8 = 1; 1175 utf8 = 1;
817 } 1176 }
1177 else if (ch == '\t' && dec->json.flags & F_RELAXED)
1178 *cur++ = ch;
818 else 1179 else
819 { 1180 {
820 --dec_cur; 1181 --dec_cur;
821 1182
822 if (!ch) 1183 if (!ch)
… …
830 { 1191 {
831 STRLEN len = cur - buf; 1192 STRLEN len = cur - buf;
832 1193
833 if (sv) 1194 if (sv)
834 { 1195 {
835 SvGROW (sv, SvCUR (sv) + len + 1); 1196 STRLEN cur = SvCUR (sv);
1197
1198 if (SvLEN (sv) - cur <= len)
1199 json_sv_grow (sv, cur, len);
1200
836 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1201 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
837 SvCUR_set (sv, SvCUR (sv) + len); 1202 SvCUR_set (sv, SvCUR (sv) + len);
838 } 1203 }
839 else 1204 else
840 sv = newSVpvn (buf, len); 1205 sv = newSVpvn (buf, len);
… …
925 is_nv = 1; 1290 is_nv = 1;
926 } 1291 }
927 1292
928 if (!is_nv) 1293 if (!is_nv)
929 { 1294 {
1295 int len = dec->cur - start;
1296
930 // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so 1297 // special case the rather common 1..5-digit-int case
931 if (*start == '-') 1298 if (*start == '-')
932 switch (dec->cur - start) 1299 switch (len)
933 { 1300 {
934 case 2: return newSViv (-( start [1] - '0' * 1)); 1301 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
935 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1302 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
936 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1303 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
937 case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1304 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1305 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
938 } 1306 }
939 else 1307 else
940 switch (dec->cur - start) 1308 switch (len)
941 { 1309 {
942 case 1: return newSViv ( start [0] - '0' * 1); 1310 case 1: return newSViv ( start [0] - '0' * 1);
943 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1311 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
944 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1312 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
945 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1313 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1314 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
946 } 1315 }
947 1316
948 { 1317 {
949 UV uv; 1318 UV uv;
950 int numtype = grok_number (start, dec->cur - start, &uv); 1319 int numtype = grok_number (start, len, &uv);
951 if (numtype & IS_NUMBER_IN_UV) 1320 if (numtype & IS_NUMBER_IN_UV)
952 if (numtype & IS_NUMBER_NEG) 1321 if (numtype & IS_NUMBER_NEG)
953 { 1322 {
954 if (uv < (UV)IV_MIN) 1323 if (uv < (UV)IV_MIN)
955 return newSViv (-(IV)uv); 1324 return newSViv (-(IV)uv);
956 } 1325 }
957 else 1326 else
958 return newSVuv (uv); 1327 return newSVuv (uv);
959
960 // here would likely be the place for bigint support
961 } 1328 }
962 }
963 1329
964 // if we ever support bigint or bigfloat, this is the place for bigfloat 1330 len -= *start == '-' ? 1 : 0;
1331
1332 // does not fit into IV or UV, try NV
1333 if (len <= NV_DIG)
1334 // fits into NV without loss of precision
1335 return newSVnv (json_atof (start));
1336
1337 // everything else fails, convert it to a string
1338 return newSVpvn (start, dec->cur - start);
1339 }
1340
1341 // loss of precision here
965 return newSVnv (Atof (start)); 1342 return newSVnv (json_atof (start));
966 1343
967fail: 1344fail:
968 return 0; 1345 return 0;
969} 1346}
970 1347
… …
994 if (*dec->cur == ']') 1371 if (*dec->cur == ']')
995 { 1372 {
996 ++dec->cur; 1373 ++dec->cur;
997 break; 1374 break;
998 } 1375 }
999 1376
1000 if (*dec->cur != ',') 1377 if (*dec->cur != ',')
1001 ERR (", or ] expected while parsing array"); 1378 ERR (", or ] expected while parsing array");
1002 1379
1003 ++dec->cur; 1380 ++dec->cur;
1381
1382 decode_ws (dec);
1383
1384 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1385 {
1386 ++dec->cur;
1387 break;
1388 }
1004 } 1389 }
1005 1390
1006 DEC_DEC_DEPTH; 1391 DEC_DEC_DEPTH;
1007 return newRV_noinc ((SV *)av); 1392 return newRV_noinc ((SV *)av);
1008 1393
… …
1013} 1398}
1014 1399
1015static SV * 1400static SV *
1016decode_hv (dec_t *dec) 1401decode_hv (dec_t *dec)
1017{ 1402{
1403 SV *sv;
1018 HV *hv = newHV (); 1404 HV *hv = newHV ();
1019 1405
1020 DEC_INC_DEPTH; 1406 DEC_INC_DEPTH;
1021 decode_ws (dec); 1407 decode_ws (dec);
1022 1408
1023 if (*dec->cur == '}') 1409 if (*dec->cur == '}')
1024 ++dec->cur; 1410 ++dec->cur;
1025 else 1411 else
1026 for (;;) 1412 for (;;)
1027 { 1413 {
1028 decode_ws (dec); EXPECT_CH ('"'); 1414 EXPECT_CH ('"');
1029 1415
1030 // heuristic: assume that 1416 // heuristic: assume that
1031 // a) decode_str + hv_store_ent are abysmally slow 1417 // a) decode_str + hv_store_ent are abysmally slow.
1032 // b) most hash keys are short, simple ascii text 1418 // b) most hash keys are short, simple ascii text.
1033 // so try to "fast-match" such strings to avoid 1419 // => try to "fast-match" such strings to avoid
1034 // the overhead of hv_store_ent. 1420 // the overhead of decode_str + hv_store_ent.
1035 { 1421 {
1036 SV *value; 1422 SV *value;
1037 char *p = dec->cur; 1423 char *p = dec->cur;
1038 char *e = p + 24; // only try up to 24 bytes 1424 char *e = p + 24; // only try up to 24 bytes
1039 1425
1040 for (;;) 1426 for (;;)
1041 { 1427 {
1428 // the >= 0x80 is false on most architectures
1042 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1429 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1043 { 1430 {
1044 // slow path, back up and use decode_str 1431 // slow path, back up and use decode_str
1045 SV *key = decode_str (dec); 1432 SV *key = decode_str (dec);
1046 if (!key) 1433 if (!key)
1047 goto fail; 1434 goto fail;
1048 1435
1049 decode_ws (dec); EXPECT_CH (':'); 1436 decode_ws (dec); EXPECT_CH (':');
1050 1437
1438 decode_ws (dec);
1051 value = decode_sv (dec); 1439 value = decode_sv (dec);
1052 if (!value) 1440 if (!value)
1053 { 1441 {
1054 SvREFCNT_dec (key); 1442 SvREFCNT_dec (key);
1055 goto fail; 1443 goto fail;
… …
1067 int len = p - key; 1455 int len = p - key;
1068 dec->cur = p + 1; 1456 dec->cur = p + 1;
1069 1457
1070 decode_ws (dec); EXPECT_CH (':'); 1458 decode_ws (dec); EXPECT_CH (':');
1071 1459
1460 decode_ws (dec);
1072 value = decode_sv (dec); 1461 value = decode_sv (dec);
1073 if (!value) 1462 if (!value)
1074 goto fail; 1463 goto fail;
1075 1464
1076 hv_store (hv, key, len, value, 0); 1465 hv_store (hv, key, len, value, 0);
… …
1092 1481
1093 if (*dec->cur != ',') 1482 if (*dec->cur != ',')
1094 ERR (", or } expected while parsing object/hash"); 1483 ERR (", or } expected while parsing object/hash");
1095 1484
1096 ++dec->cur; 1485 ++dec->cur;
1486
1487 decode_ws (dec);
1488
1489 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1490 {
1491 ++dec->cur;
1492 break;
1493 }
1097 } 1494 }
1098 1495
1099 DEC_DEC_DEPTH; 1496 DEC_DEC_DEPTH;
1100 return newRV_noinc ((SV *)hv); 1497 sv = newRV_noinc ((SV *)hv);
1498
1499 // check filter callbacks
1500 if (dec->json.flags & F_HOOK)
1501 {
1502 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1503 {
1504 HE *cb, *he;
1505
1506 hv_iterinit (hv);
1507 he = hv_iternext (hv);
1508 hv_iterinit (hv);
1509
1510 // the next line creates a mortal sv each time it's called.
1511 // might want to optimise this for common cases.
1512 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1513
1514 if (cb)
1515 {
1516 dSP;
1517 int count;
1518
1519 ENTER; SAVETMPS;
1520 SAVESTACK_POS ();
1521 PUSHMARK (SP);
1522 XPUSHs (HeVAL (he));
1523 sv_2mortal (sv);
1524
1525 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1526
1527 if (count == 1)
1528 {
1529 sv = newSVsv (POPs);
1530 FREETMPS; LEAVE;
1531 return sv;
1532 }
1533
1534 SvREFCNT_inc (sv);
1535 FREETMPS; LEAVE;
1536 }
1537 }
1538
1539 if (dec->json.cb_object)
1540 {
1541 dSP;
1542 int count;
1543
1544 ENTER; SAVETMPS;
1545 SAVESTACK_POS ();
1546 PUSHMARK (SP);
1547 XPUSHs (sv_2mortal (sv));
1548
1549 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1550
1551 if (count == 1)
1552 {
1553 sv = newSVsv (POPs);
1554 FREETMPS; LEAVE;
1555 return sv;
1556 }
1557
1558 SvREFCNT_inc (sv);
1559 FREETMPS; LEAVE;
1560 }
1561 }
1562
1563 return sv;
1101 1564
1102fail: 1565fail:
1103 SvREFCNT_dec (hv); 1566 SvREFCNT_dec (hv);
1104 DEC_DEC_DEPTH; 1567 DEC_DEC_DEPTH;
1105 return 0; 1568 return 0;
1106} 1569}
1107 1570
1108static SV * 1571static SV *
1572decode_tag (dec_t *dec)
1573{
1574 SV *tag = 0;
1575 SV *val = 0;
1576
1577 if (!(dec->json.flags & F_ALLOW_TAGS))
1578 ERR ("malformed JSON string, neither array, object, number, string or atom");
1579
1580 ++dec->cur;
1581
1582 decode_ws (dec);
1583
1584 tag = decode_sv (dec);
1585 if (!tag)
1586 goto fail;
1587
1588 if (!SvPOK (tag))
1589 ERR ("malformed JSON string, (tag) must be a string");
1590
1591 decode_ws (dec);
1592
1593 if (*dec->cur != ')')
1594 ERR (") expected after tag");
1595
1596 ++dec->cur;
1597
1598 decode_ws (dec);
1599
1600 val = decode_sv (dec);
1601 if (!val)
1602 goto fail;
1603
1604 if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV)
1605 ERR ("malformed JSON string, tag value must be an array");
1606
1607 {
1608 AV *av = (AV *)SvRV (val);
1609 int i, len = av_len (av) + 1;
1610 HV *stash = gv_stashsv (tag, 0);
1611 SV *sv;
1612
1613 if (!stash)
1614 ERR ("cannot decode perl-object (package does not exist)");
1615
1616 GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
1617
1618 if (!method)
1619 ERR ("cannot decode perl-object (package does not have a THAW method)");
1620
1621 dSP;
1622
1623 ENTER; SAVETMPS;
1624 PUSHMARK (SP);
1625 EXTEND (SP, len + 2);
1626 // we re-bless the reference to get overload and other niceties right
1627 PUSHs (tag);
1628 PUSHs (sv_json);
1629
1630 for (i = 0; i < len; ++i)
1631 PUSHs (*av_fetch (av, i, 1));
1632
1633 PUTBACK;
1634 call_sv ((SV *)GvCV (method), G_SCALAR);
1635 SPAGAIN;
1636
1637 SvREFCNT_dec (tag);
1638 SvREFCNT_dec (val);
1639 sv = SvREFCNT_inc (POPs);
1640
1641 PUTBACK;
1642
1643 FREETMPS; LEAVE;
1644
1645 return sv;
1646 }
1647
1648fail:
1649 SvREFCNT_dec (tag);
1650 SvREFCNT_dec (val);
1651 return 0;
1652}
1653
1654static SV *
1109decode_sv (dec_t *dec) 1655decode_sv (dec_t *dec)
1110{ 1656{
1111 decode_ws (dec);
1112
1113 // the beauty of JSON: you need exactly one character lookahead 1657 // the beauty of JSON: you need exactly one character lookahead
1114 // to parse anything. 1658 // to parse everything.
1115 switch (*dec->cur) 1659 switch (*dec->cur)
1116 { 1660 {
1117 case '"': ++dec->cur; return decode_str (dec); 1661 case '"': ++dec->cur; return decode_str (dec);
1118 case '[': ++dec->cur; return decode_av (dec); 1662 case '[': ++dec->cur; return decode_av (dec);
1119 case '{': ++dec->cur; return decode_hv (dec); 1663 case '{': ++dec->cur; return decode_hv (dec);
1664 case '(': return decode_tag (dec);
1120 1665
1121 case '-': 1666 case '-':
1122 case '0': case '1': case '2': case '3': case '4': 1667 case '0': case '1': case '2': case '3': case '4':
1123 case '5': case '6': case '7': case '8': case '9': 1668 case '5': case '6': case '7': case '8': case '9':
1124 return decode_num (dec); 1669 return decode_num (dec);
1125 1670
1126 case 't': 1671 case 't':
1127 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1672 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1128 { 1673 {
1129 dec->cur += 4; 1674 dec->cur += 4;
1130 return SvREFCNT_inc (json_true); 1675#if JSON_SLOW
1676 bool_true = get_bool ("Types::Serialiser::true");
1677#endif
1678 return newSVsv (bool_true);
1131 } 1679 }
1132 else 1680 else
1133 ERR ("'true' expected"); 1681 ERR ("'true' expected");
1134 1682
1135 break; 1683 break;
1136 1684
1137 case 'f': 1685 case 'f':
1138 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1686 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1139 { 1687 {
1140 dec->cur += 5; 1688 dec->cur += 5;
1689#if JSON_SLOW
1690 bool_false = get_bool ("Types::Serialiser::false");
1691#endif
1141 return SvREFCNT_inc (json_false); 1692 return newSVsv (bool_false);
1142 } 1693 }
1143 else 1694 else
1144 ERR ("'false' expected"); 1695 ERR ("'false' expected");
1145 1696
1146 break; 1697 break;
… …
1155 ERR ("'null' expected"); 1706 ERR ("'null' expected");
1156 1707
1157 break; 1708 break;
1158 1709
1159 default: 1710 default:
1160 ERR ("malformed JSON string, neither array, object, number, string or atom"); 1711 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
1161 break; 1712 break;
1162 } 1713 }
1163 1714
1164fail: 1715fail:
1165 return 0; 1716 return 0;
1166} 1717}
1167 1718
1168static SV * 1719static SV *
1169decode_json (SV *string, U32 flags, UV *offset_return) 1720decode_json (SV *string, JSON *json, STRLEN *offset_return)
1170{ 1721{
1171 dec_t dec; 1722 dec_t dec;
1172 UV offset;
1173 SV *sv; 1723 SV *sv;
1174 1724
1725 /* work around bugs in 5.10 where manipulating magic values
1726 * makes perl ignore the magic in subsequent accesses.
1727 * also make a copy of non-PV values, to get them into a clean
1728 * state (SvPV should do that, but it's buggy, see below).
1729 *
1730 * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16),
1731 * as reported by Reini Urban.
1732 */
1175 SvGETMAGIC (string); 1733 /*SvGETMAGIC (string);*/
1734 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1735 string = sv_2mortal (newSVsv (string));
1736
1176 SvUPGRADE (string, SVt_PV); 1737 SvUPGRADE (string, SVt_PV);
1177 1738
1178 if (flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (flags)) 1739 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1740 * assertion with -DDEBUGGING, although SvCUR is documented to
1741 * return the xpv_cur field which certainly exists after upgrading.
1742 * according to nicholas clark, calling SvPOK fixes this.
1743 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1744 * and hope for the best.
1745 * Damnit, SvPV_nolen still trips over yet another assertion. This
1746 * assertion business is seriously broken, try yet another workaround
1747 * for the broken -DDEBUGGING.
1748 */
1749 {
1750#ifdef DEBUGGING
1751 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1752#else
1753 STRLEN offset = SvCUR (string);
1754#endif
1755
1756 if (offset > json->max_size && json->max_size)
1179 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1757 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1180 (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (flags)); 1758 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1759 }
1181 1760
1182 if (flags & F_UTF8) 1761 if (DECODE_WANTS_OCTETS (json))
1183 sv_utf8_downgrade (string, 0); 1762 sv_utf8_downgrade (string, 0);
1184 else 1763 else
1185 sv_utf8_upgrade (string); 1764 sv_utf8_upgrade (string);
1186 1765
1187 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1766 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
1188 1767
1189 dec.flags = flags; 1768 dec.json = *json;
1190 dec.cur = SvPVX (string); 1769 dec.cur = SvPVX (string);
1191 dec.end = SvEND (string); 1770 dec.end = SvEND (string);
1192 dec.err = 0; 1771 dec.err = 0;
1193 dec.depth = 0; 1772 dec.depth = 0;
1194 dec.maxdepth = DEC_DEPTH (dec.flags); 1773
1774 if (dec.json.cb_object || dec.json.cb_sk_object)
1775 dec.json.flags |= F_HOOK;
1195 1776
1196 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1777 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1778
1779 decode_ws (&dec);
1197 sv = decode_sv (&dec); 1780 sv = decode_sv (&dec);
1198 1781
1199 if (!(offset_return || !sv)) 1782 if (offset_return)
1783 *offset_return = dec.cur - SvPVX (string);
1784 else if (sv)
1200 { 1785 {
1201 // check for trailing garbage 1786 // check for trailing garbage
1202 decode_ws (&dec); 1787 decode_ws (&dec);
1203 1788
1204 if (*dec.cur) 1789 if (*dec.cur)
1205 { 1790 {
1206 dec.err = "garbage after JSON object"; 1791 dec.err = "garbage after JSON object";
1207 SvREFCNT_dec (sv); 1792 SvREFCNT_dec (sv);
1208 sv = 0; 1793 sv = 0;
1209 } 1794 }
1210 }
1211
1212 if (offset_return || !sv)
1213 {
1214 offset = dec.flags & F_UTF8
1215 ? dec.cur - SvPVX (string)
1216 : utf8_distance (dec.cur, SvPVX (string));
1217
1218 if (offset_return)
1219 *offset_return = offset;
1220 } 1795 }
1221 1796
1222 if (!sv) 1797 if (!sv)
1223 { 1798 {
1224 SV *uni = sv_newmortal (); 1799 SV *uni = sv_newmortal ();
… …
1230 SAVEVPTR (PL_curcop); 1805 SAVEVPTR (PL_curcop);
1231 PL_curcop = &cop; 1806 PL_curcop = &cop;
1232 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1807 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1233 LEAVE; 1808 LEAVE;
1234 1809
1235 croak ("%s, at character offset %d [\"%s\"]", 1810 croak ("%s, at character offset %d (before \"%s\")",
1236 dec.err, 1811 dec.err,
1237 (int)offset, 1812 (int)ptr_to_index (string, dec.cur),
1238 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1813 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1239 } 1814 }
1240 1815
1241 sv = sv_2mortal (sv); 1816 sv = sv_2mortal (sv);
1242 1817
1243 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1818 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
1244 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1819 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
1245 1820
1246 return sv; 1821 return sv;
1822}
1823
1824/////////////////////////////////////////////////////////////////////////////
1825// incremental parser
1826
1827static void
1828incr_parse (JSON *self)
1829{
1830 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1831
1832 // the state machine here is a bit convoluted and could be simplified a lot
1833 // but this would make it slower, so...
1834
1835 for (;;)
1836 {
1837 //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
1838 switch (self->incr_mode)
1839 {
1840 // only used for initial whitespace skipping
1841 case INCR_M_WS:
1842 for (;;)
1843 {
1844 if (*p > 0x20)
1845 {
1846 if (*p == '#')
1847 {
1848 self->incr_mode = INCR_M_C0;
1849 goto incr_m_c;
1850 }
1851 else
1852 {
1853 self->incr_mode = INCR_M_JSON;
1854 goto incr_m_json;
1855 }
1856 }
1857 else if (!*p)
1858 goto interrupt;
1859
1860 ++p;
1861 }
1862
1863 // skip a single char inside a string (for \\-processing)
1864 case INCR_M_BS:
1865 if (!*p)
1866 goto interrupt;
1867
1868 ++p;
1869 self->incr_mode = INCR_M_STR;
1870 goto incr_m_str;
1871
1872 // inside #-style comments
1873 case INCR_M_C0:
1874 case INCR_M_C1:
1875 incr_m_c:
1876 for (;;)
1877 {
1878 if (*p == '\n')
1879 {
1880 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1881 break;
1882 }
1883 else if (!*p)
1884 goto interrupt;
1885
1886 ++p;
1887 }
1888
1889 break;
1890
1891 // inside a string
1892 case INCR_M_STR:
1893 incr_m_str:
1894 for (;;)
1895 {
1896 if (*p == '"')
1897 {
1898 ++p;
1899 self->incr_mode = INCR_M_JSON;
1900
1901 if (!self->incr_nest)
1902 goto interrupt;
1903
1904 goto incr_m_json;
1905 }
1906 else if (*p == '\\')
1907 {
1908 ++p; // "virtually" consumes character after \
1909
1910 if (!*p) // if at end of string we have to switch modes
1911 {
1912 self->incr_mode = INCR_M_BS;
1913 goto interrupt;
1914 }
1915 }
1916 else if (!*p)
1917 goto interrupt;
1918
1919 ++p;
1920 }
1921
1922 // after initial ws, outside string
1923 case INCR_M_JSON:
1924 incr_m_json:
1925 for (;;)
1926 {
1927 switch (*p++)
1928 {
1929 case 0:
1930 --p;
1931 goto interrupt;
1932
1933 case 0x09:
1934 case 0x0a:
1935 case 0x0d:
1936 case 0x20:
1937 if (!self->incr_nest)
1938 {
1939 --p; // do not eat the whitespace, let the next round do it
1940 goto interrupt;
1941 }
1942 break;
1943
1944 case '"':
1945 self->incr_mode = INCR_M_STR;
1946 goto incr_m_str;
1947
1948 case '[':
1949 case '{':
1950 case '(':
1951 if (++self->incr_nest > self->max_depth)
1952 croak (ERR_NESTING_EXCEEDED);
1953 break;
1954
1955 case ']':
1956 case '}':
1957 if (--self->incr_nest <= 0)
1958 goto interrupt;
1959 break;
1960
1961 case ')':
1962 --self->incr_nest;
1963 break;
1964
1965 case '#':
1966 self->incr_mode = INCR_M_C1;
1967 goto incr_m_c;
1968 }
1969 }
1970 }
1971
1972 modechange:
1973 ;
1974 }
1975
1976interrupt:
1977 self->incr_pos = p - SvPVX (self->incr_text);
1978 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1979 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1247} 1980}
1248 1981
1249///////////////////////////////////////////////////////////////////////////// 1982/////////////////////////////////////////////////////////////////////////////
1250// XS interface functions 1983// XS interface functions
1251 1984
… …
1260 i >= '0' && i <= '9' ? i - '0' 1993 i >= '0' && i <= '9' ? i - '0'
1261 : i >= 'a' && i <= 'f' ? i - 'a' + 10 1994 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1262 : i >= 'A' && i <= 'F' ? i - 'A' + 10 1995 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1263 : -1; 1996 : -1;
1264 1997
1265 json_stash = gv_stashpv ("JSON::XS" , 1); 1998 json_stash = gv_stashpv ("JSON::XS" , 1);
1266 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1999 bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
2000 bool_true = get_bool ("Types::Serialiser::true");
2001 bool_false = get_bool ("Types::Serialiser::false");
1267 2002
1268 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 2003 sv_json = newSVpv ("JSON", 0);
1269 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 2004 SvREADONLY_on (sv_json);
2005
2006 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1270} 2007}
1271 2008
1272PROTOTYPES: DISABLE 2009PROTOTYPES: DISABLE
1273 2010
1274SV *new (char *dummy) 2011void CLONE (...)
1275 CODE: 2012 CODE:
1276 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 2013 json_stash = 0;
1277 OUTPUT: 2014 bool_stash = 0;
1278 RETVAL
1279 2015
2016void new (char *klass)
2017 PPCODE:
2018{
2019 SV *pv = NEWSV (0, sizeof (JSON));
2020 SvPOK_only (pv);
2021 json_init ((JSON *)SvPVX (pv));
2022 XPUSHs (sv_2mortal (sv_bless (
2023 newRV_noinc (pv),
2024 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
2025 )));
2026}
2027
1280SV *ascii (SV *self, int enable = 1) 2028void ascii (JSON *self, int enable = 1)
1281 ALIAS: 2029 ALIAS:
1282 ascii = F_ASCII 2030 ascii = F_ASCII
1283 latin1 = F_LATIN1 2031 latin1 = F_LATIN1
1284 utf8 = F_UTF8 2032 utf8 = F_UTF8
1285 indent = F_INDENT 2033 indent = F_INDENT
… …
1289 pretty = F_PRETTY 2037 pretty = F_PRETTY
1290 allow_nonref = F_ALLOW_NONREF 2038 allow_nonref = F_ALLOW_NONREF
1291 shrink = F_SHRINK 2039 shrink = F_SHRINK
1292 allow_blessed = F_ALLOW_BLESSED 2040 allow_blessed = F_ALLOW_BLESSED
1293 convert_blessed = F_CONV_BLESSED 2041 convert_blessed = F_CONV_BLESSED
2042 relaxed = F_RELAXED
2043 allow_unknown = F_ALLOW_UNKNOWN
2044 allow_tags = F_ALLOW_TAGS
2045 PPCODE:
2046{
2047 if (enable)
2048 self->flags |= ix;
2049 else
2050 self->flags &= ~ix;
2051
2052 XPUSHs (ST (0));
2053}
2054
2055void get_ascii (JSON *self)
2056 ALIAS:
2057 get_ascii = F_ASCII
2058 get_latin1 = F_LATIN1
2059 get_utf8 = F_UTF8
2060 get_indent = F_INDENT
2061 get_canonical = F_CANONICAL
2062 get_space_before = F_SPACE_BEFORE
2063 get_space_after = F_SPACE_AFTER
2064 get_allow_nonref = F_ALLOW_NONREF
2065 get_shrink = F_SHRINK
2066 get_allow_blessed = F_ALLOW_BLESSED
2067 get_convert_blessed = F_CONV_BLESSED
2068 get_relaxed = F_RELAXED
2069 get_allow_unknown = F_ALLOW_UNKNOWN
2070 get_allow_tags = F_ALLOW_TAGS
2071 PPCODE:
2072 XPUSHs (boolSV (self->flags & ix));
2073
2074void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
2075 PPCODE:
2076 self->max_depth = max_depth;
2077 XPUSHs (ST (0));
2078
2079U32 get_max_depth (JSON *self)
1294 CODE: 2080 CODE:
1295{ 2081 RETVAL = self->max_depth;
1296 UV *uv = SvJSON (self);
1297 if (enable)
1298 *uv |= ix;
1299 else
1300 *uv &= ~ix;
1301
1302 RETVAL = newSVsv (self);
1303}
1304 OUTPUT: 2082 OUTPUT:
1305 RETVAL 2083 RETVAL
1306 2084
1307SV *max_depth (SV *self, UV max_depth = 0x80000000UL) 2085void max_size (JSON *self, U32 max_size = 0)
2086 PPCODE:
2087 self->max_size = max_size;
2088 XPUSHs (ST (0));
2089
2090int get_max_size (JSON *self)
1308 CODE: 2091 CODE:
1309{ 2092 RETVAL = self->max_size;
1310 UV *uv = SvJSON (self);
1311 UV log2 = 0;
1312
1313 if (max_depth > 0x80000000UL) max_depth = 0x80000000UL;
1314
1315 while ((1UL << log2) < max_depth)
1316 ++log2;
1317
1318 *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH);
1319
1320 RETVAL = newSVsv (self);
1321}
1322 OUTPUT: 2093 OUTPUT:
1323 RETVAL 2094 RETVAL
1324 2095
1325SV *max_size (SV *self, UV max_size = 0) 2096void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
2097 PPCODE:
2098{
2099 SvREFCNT_dec (self->cb_object);
2100 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
2101
2102 XPUSHs (ST (0));
2103}
2104
2105void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
2106 PPCODE:
2107{
2108 if (!self->cb_sk_object)
2109 self->cb_sk_object = newHV ();
2110
2111 if (SvOK (cb))
2112 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
2113 else
2114 {
2115 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
2116
2117 if (!HvKEYS (self->cb_sk_object))
2118 {
2119 SvREFCNT_dec (self->cb_sk_object);
2120 self->cb_sk_object = 0;
2121 }
2122 }
2123
2124 XPUSHs (ST (0));
2125}
2126
2127void encode (JSON *self, SV *scalar)
2128 PPCODE:
2129 PUTBACK; scalar = encode_json (scalar, self); SPAGAIN;
2130 XPUSHs (scalar);
2131
2132void decode (JSON *self, SV *jsonstr)
2133 PPCODE:
2134 PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN;
2135 XPUSHs (jsonstr);
2136
2137void decode_prefix (JSON *self, SV *jsonstr)
2138 PPCODE:
2139{
2140 SV *sv;
2141 STRLEN offset;
2142 PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN;
2143 EXTEND (SP, 2);
2144 PUSHs (sv);
2145 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset))));
2146}
2147
2148void incr_parse (JSON *self, SV *jsonstr = 0)
2149 PPCODE:
2150{
2151 if (!self->incr_text)
2152 self->incr_text = newSVpvn ("", 0);
2153
2154 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
2155 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
2156 if (DECODE_WANTS_OCTETS (self))
2157 {
2158 if (self->incr_pos)
2159 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
2160 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
2161
2162 sv_utf8_downgrade (self->incr_text, 0);
2163 }
2164 else
2165 {
2166 sv_utf8_upgrade (self->incr_text);
2167
2168 if (self->incr_pos)
2169 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
2170 - (U8 *)SvPVX (self->incr_text);
2171 }
2172
2173 // append data, if any
2174 if (jsonstr)
2175 {
2176 /* make sure both strings have same encoding */
2177 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
2178 if (SvUTF8 (jsonstr))
2179 sv_utf8_downgrade (jsonstr, 0);
2180 else
2181 sv_utf8_upgrade (jsonstr);
2182
2183 /* and then just blindly append */
2184 {
2185 STRLEN len;
2186 const char *str = SvPV (jsonstr, len);
2187 STRLEN cur = SvCUR (self->incr_text);
2188
2189 if (SvLEN (self->incr_text) - cur <= len)
2190 json_sv_grow (self->incr_text, cur, len);
2191
2192 Move (str, SvEND (self->incr_text), len, char);
2193 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
2194 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
2195 }
2196 }
2197
2198 if (GIMME_V != G_VOID)
2199 do
2200 {
2201 SV *sv;
2202 STRLEN offset;
2203
2204 if (!INCR_DONE (self))
2205 {
2206 incr_parse (self);
2207
2208 if (self->incr_pos > self->max_size && self->max_size)
2209 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
2210 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
2211
2212 if (!INCR_DONE (self))
2213 {
2214 // as an optimisation, do not accumulate white space in the incr buffer
2215 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2216 {
2217 self->incr_pos = 0;
2218 SvCUR_set (self->incr_text, 0);
2219 }
2220
2221 break;
2222 }
2223 }
2224
2225 PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN;
2226 XPUSHs (sv);
2227
2228 self->incr_pos -= offset;
2229 self->incr_nest = 0;
2230 self->incr_mode = 0;
2231
2232 sv_chop (self->incr_text, SvPVX (self->incr_text) + offset);
2233 }
2234 while (GIMME_V == G_ARRAY);
2235}
2236
2237SV *incr_text (JSON *self)
2238 ATTRS: lvalue
1326 CODE: 2239 CODE:
1327{ 2240{
1328 UV *uv = SvJSON (self); 2241 if (self->incr_pos)
1329 UV log2 = 0; 2242 croak ("incr_text can not be called when the incremental parser already started parsing");
1330 2243
1331 if (max_size > 0x80000000UL) max_size = 0x80000000UL; 2244 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1332 if (max_size == 1) max_size = 2;
1333
1334 while ((1UL << log2) < max_size)
1335 ++log2;
1336
1337 *uv = *uv & ~F_MAXSIZE | (log2 << S_MAXSIZE);
1338
1339 RETVAL = newSVsv (self);
1340} 2245}
1341 OUTPUT: 2246 OUTPUT:
1342 RETVAL 2247 RETVAL
1343 2248
2249void incr_skip (JSON *self)
2250 CODE:
2251{
2252 if (self->incr_pos)
2253 {
2254 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
2255 self->incr_pos = 0;
2256 self->incr_nest = 0;
2257 self->incr_mode = 0;
2258 }
2259}
2260
2261void incr_reset (JSON *self)
2262 CODE:
2263{
2264 SvREFCNT_dec (self->incr_text);
2265 self->incr_text = 0;
2266 self->incr_pos = 0;
2267 self->incr_nest = 0;
2268 self->incr_mode = 0;
2269}
2270
2271void DESTROY (JSON *self)
2272 CODE:
2273 SvREFCNT_dec (self->cb_sk_object);
2274 SvREFCNT_dec (self->cb_object);
2275 SvREFCNT_dec (self->incr_text);
2276
2277PROTOTYPES: ENABLE
2278
1344void encode (SV *self, SV *scalar) 2279void encode_json (SV *scalar)
1345 PPCODE: 2280 PPCODE:
1346 XPUSHs (encode_json (scalar, *SvJSON (self))); 2281{
2282 JSON json;
2283 json_init (&json);
2284 json.flags |= F_UTF8;
2285 PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN;
2286 XPUSHs (scalar);
2287}
1347 2288
1348void decode (SV *self, SV *jsonstr) 2289void decode_json (SV *jsonstr)
1349 PPCODE: 2290 PPCODE:
1350 XPUSHs (decode_json (jsonstr, *SvJSON (self), 0));
1351
1352void decode_prefix (SV *self, SV *jsonstr)
1353 PPCODE:
1354{ 2291{
1355 UV offset; 2292 JSON json;
1356 EXTEND (SP, 2); 2293 json_init (&json);
1357 PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); 2294 json.flags |= F_UTF8;
1358 PUSHs (sv_2mortal (newSVuv (offset))); 2295 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2296 XPUSHs (jsonstr);
1359} 2297}
1360 2298
1361PROTOTYPES: ENABLE
1362
1363void to_json (SV *scalar)
1364 ALIAS:
1365 objToJson = 0
1366 PPCODE:
1367 XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8));
1368
1369void from_json (SV *jsonstr)
1370 ALIAS:
1371 jsonToObj = 0
1372 PPCODE:
1373 XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));
1374

Diff Legend

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