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 1.17 by root, Sun Mar 25 02:46:41 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>
9#include <limits.h>
10#include <float.h>
11#include <inttypes.h>
8 12
13#if defined(__BORLANDC__) || defined(_MSC_VER)
14# define snprintf _snprintf // C compilers have this in stdio.h
15#endif
16
17// some old perls do not have this, try to make it work, no
18// guarantees, though. if it breaks, you get to keep the pieces.
19#ifndef UTF8_MAXBYTES
20# define UTF8_MAXBYTES 13
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)
36
9#define F_ASCII 0x00000001 37#define F_ASCII 0x00000001UL
38#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 39#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 40#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 41#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 42#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 43#define F_SPACE_AFTER 0x00000040UL
15#define F_ALLOW_NONREF 0x00000080 44#define F_ALLOW_NONREF 0x00000100UL
16#define F_SHRINK 0x00000100 45#define F_SHRINK 0x00000200UL
17 46#define F_ALLOW_BLESSED 0x00000400UL
18// F_SKIPINVALID? 47#define F_CONV_BLESSED 0x00000800UL
19// F_EXECCODEREF? 48#define F_RELAXED 0x00001000UL
20// F_SELFCONVERT? <=> { &__class__ => } 49#define F_ALLOW_UNKNOWN 0x00002000UL
50#define F_ALLOW_TAGS 0x00004000UL
51#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
21 52
22#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 53#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
23#define F_DEFAULT 0
24 54
25#define INIT_SIZE 32 // initial scalar size to be allocated 55#define INIT_SIZE 32 // initial scalar size to be allocated
26#define INDENT_STEP 3 // spaces per indentation level 56#define INDENT_STEP 3 // spaces per indentation level
27 57
28#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
29#define SHORT_STRING_LEN 512 // special-case strings of up to this size 58#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
59
60#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
30 61
31#define SB do { 62#define SB do {
32#define SE } while (0) 63#define SE } while (0)
33 64
34static HV *json_stash; // JSON::XS:: 65#if __GNUC__ >= 3
66# define expect(expr,value) __builtin_expect ((expr), (value))
67# define INLINE static inline
68#else
69# define expect(expr,value) (expr)
70# define INLINE static
71#endif
72
73#define expect_false(expr) expect ((expr) != 0, 0)
74#define expect_true(expr) expect ((expr) != 0, 1)
75
76#define IN_RANGE_INC(type,val,beg,end) \
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}
35 130
36///////////////////////////////////////////////////////////////////////////// 131/////////////////////////////////////////////////////////////////////////////
37// utility functions 132// utility functions
38 133
39static UV * 134INLINE SV *
40SvJSON (SV *sv) 135get_bool (const char *name)
41{ 136{
42 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 137 SV *sv = get_sv (name, 1);
43 croak ("object is not of type JSON::XS");
44 138
45 return &SvUVX (SvRV (sv)); 139 SvREADONLY_on (sv);
46} 140 SvREADONLY_on (SvRV (sv));
47 141
48static void 142 return sv;
143}
144
145INLINE void
49shrink (SV *sv) 146shrink (SV *sv)
50{ 147{
51 sv_utf8_downgrade (sv, 1); 148 sv_utf8_downgrade (sv, 1);
149
52 if (SvLEN (sv) > SvCUR (sv) + 1) 150 if (SvLEN (sv) > SvCUR (sv) + 1)
53 { 151 {
54#ifdef SvPV_shrink_to_cur 152#ifdef SvPV_shrink_to_cur
55 SvPV_shrink_to_cur (sv); 153 SvPV_shrink_to_cur (sv);
56#elif defined (SvPV_renew) 154#elif defined (SvPV_renew)
57 SvPV_renew (sv, SvCUR (sv) + 1); 155 SvPV_renew (sv, SvCUR (sv) + 1);
58#endif 156#endif
59 } 157 }
60} 158}
61 159
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
62// 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
63// case of an error. 186// case of an error.
64// we special-case "safe" characters from U+80 .. U+7FF, 187// we special-case "safe" characters from U+80 .. U+7FF,
65// but use the very good perl function to parse anything else. 188// but use the very good perl function to parse anything else.
66// note that we never call this function for a ascii codepoints 189// note that we never call this function for a ascii codepoints
67static UV 190INLINE UV
68decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 191decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
69{ 192{
70 if (s[0] > 0xdf || s[0] < 0xc2) 193 if (expect_true (len >= 2
71 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 194 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
72 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 195 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
73 { 196 {
74 *clen = 2; 197 *clen = 2;
75 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 198 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
76 } 199 }
77 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 }
303 }
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))
78 return (UV)-1; 359 return 1;
360
361 scalar = SvRV (scalar);
362
363 if (SvTYPE (scalar) >= SVt_PVMG)
364 {
365 if (SvSTASH (scalar) == bool_stash)
366 return 1;
367
368 if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0)
369 return 1;
370 }
371
372 return 0;
79} 373}
80 374
81///////////////////////////////////////////////////////////////////////////// 375/////////////////////////////////////////////////////////////////////////////
82// encoder 376// encoder
83 377
85typedef struct 379typedef struct
86{ 380{
87 char *cur; // SvPVX (sv) + current output position 381 char *cur; // SvPVX (sv) + current output position
88 char *end; // SvEND (sv) 382 char *end; // SvEND (sv)
89 SV *sv; // result scalar 383 SV *sv; // result scalar
90 UV flags; // F_* 384 JSON json;
91 int indent; // indentation level 385 U32 indent; // indentation level
92 int max_depth; // max. recursion level 386 UV limit; // escape character values >= this value when encoding
93} enc_t; 387} enc_t;
94 388
95static void 389INLINE void
96need (enc_t *enc, STRLEN len) 390need (enc_t *enc, STRLEN len)
97{ 391{
98 if (enc->cur + len >= enc->end) 392 if (expect_false ((uintptr_t)(enc->end - enc->cur) < len))
99 { 393 {
100 STRLEN cur = enc->cur - SvPVX (enc->sv); 394 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
101 SvGROW (enc->sv, cur + len + 1); 395 char *buf = json_sv_grow (enc->sv, cur, len);
102 enc->cur = SvPVX (enc->sv) + cur; 396 enc->cur = buf + cur;
103 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 397 enc->end = buf + SvLEN (enc->sv) - 1;
104 } 398 }
105} 399}
106 400
107static void 401INLINE void
108encode_ch (enc_t *enc, char ch) 402encode_ch (enc_t *enc, char ch)
109{ 403{
110 need (enc, 1); 404 need (enc, 1);
111 *enc->cur++ = ch; 405 *enc->cur++ = ch;
112} 406}
120 414
121 while (str < end) 415 while (str < end)
122 { 416 {
123 unsigned char ch = *(unsigned char *)str; 417 unsigned char ch = *(unsigned char *)str;
124 418
125 if (ch >= 0x20 && ch < 0x80) // most common case 419 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
126 { 420 {
127 if (ch == '"') // but with slow exceptions 421 if (expect_false (ch == '"')) // but with slow exceptions
128 { 422 {
129 need (enc, len += 1); 423 need (enc, len + 1);
130 *enc->cur++ = '\\'; 424 *enc->cur++ = '\\';
131 *enc->cur++ = '"'; 425 *enc->cur++ = '"';
132 } 426 }
133 else if (ch == '\\') 427 else if (expect_false (ch == '\\'))
134 { 428 {
135 need (enc, len += 1); 429 need (enc, len + 1);
136 *enc->cur++ = '\\'; 430 *enc->cur++ = '\\';
137 *enc->cur++ = '\\'; 431 *enc->cur++ = '\\';
138 } 432 }
139 else 433 else
140 *enc->cur++ = ch; 434 *enc->cur++ = ch;
143 } 437 }
144 else 438 else
145 { 439 {
146 switch (ch) 440 switch (ch)
147 { 441 {
148 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;
149 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;
150 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;
151 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;
152 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;
153 447
154 default: 448 default:
155 { 449 {
156 STRLEN clen; 450 STRLEN clen;
157 UV uch; 451 UV uch;
158 452
159 if (is_utf8) 453 if (is_utf8)
160 { 454 {
161 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
162 uch = decode_utf8 (str, end - str, &clen); 455 uch = decode_utf8 (str, end - str, &clen);
163 if (clen == (STRLEN)-1) 456 if (clen == (STRLEN)-1)
164 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 457 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
165 } 458 }
166 else 459 else
167 { 460 {
168 uch = ch; 461 uch = ch;
169 clen = 1; 462 clen = 1;
170 } 463 }
171 464
172 if (uch > 0x10FFFFUL) 465 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
173 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
174
175 if (uch < 0x80 || enc->flags & F_ASCII)
176 { 466 {
177 if (uch > 0xFFFFUL) 467 if (uch >= 0x10000UL)
178 { 468 {
469 if (uch >= 0x110000UL)
470 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
471
179 need (enc, len += 11); 472 need (enc, len + 11);
180 sprintf (enc->cur, "\\u%04x\\u%04x", 473 sprintf (enc->cur, "\\u%04x\\u%04x",
181 (int)((uch - 0x10000) / 0x400 + 0xD800), 474 (int)((uch - 0x10000) / 0x400 + 0xD800),
182 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 475 (int)((uch - 0x10000) % 0x400 + 0xDC00));
183 enc->cur += 12; 476 enc->cur += 12;
184 } 477 }
185 else 478 else
186 { 479 {
187 static char hexdigit [16] = "0123456789abcdef";
188 need (enc, len += 5); 480 need (enc, len + 5);
189 *enc->cur++ = '\\'; 481 *enc->cur++ = '\\';
190 *enc->cur++ = 'u'; 482 *enc->cur++ = 'u';
191 *enc->cur++ = hexdigit [ uch >> 12 ]; 483 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
192 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 484 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
193 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 485 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
194 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 486 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
195 } 487 }
196 488
489 str += clen;
490 }
491 else if (enc->json.flags & F_LATIN1)
492 {
493 *enc->cur++ = uch;
197 str += clen; 494 str += clen;
198 } 495 }
199 else if (is_utf8) 496 else if (is_utf8)
200 { 497 {
201 need (enc, len += clen); 498 need (enc, len + clen);
202 do 499 do
203 { 500 {
204 *enc->cur++ = *str++; 501 *enc->cur++ = *str++;
205 } 502 }
206 while (--clen); 503 while (--clen);
207 } 504 }
208 else 505 else
209 { 506 {
210 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 507 need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed
211 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 508 enc->cur = encode_utf8 (enc->cur, uch);
212 ++str; 509 ++str;
213 } 510 }
214 } 511 }
215 } 512 }
216 } 513 }
217 514
218 --len; 515 --len;
219 } 516 }
220} 517}
221 518
222static void 519INLINE void
223encode_indent (enc_t *enc) 520encode_indent (enc_t *enc)
224{ 521{
225 if (enc->flags & F_INDENT) 522 if (enc->json.flags & F_INDENT)
226 { 523 {
227 int spaces = enc->indent * INDENT_STEP; 524 int spaces = enc->indent * INDENT_STEP;
228 525
229 need (enc, spaces); 526 need (enc, spaces);
230 memset (enc->cur, ' ', spaces); 527 memset (enc->cur, ' ', spaces);
231 enc->cur += spaces; 528 enc->cur += spaces;
232 } 529 }
233} 530}
234 531
235static void 532INLINE void
236encode_space (enc_t *enc) 533encode_space (enc_t *enc)
237{ 534{
238 need (enc, 1); 535 need (enc, 1);
239 encode_ch (enc, ' '); 536 encode_ch (enc, ' ');
240} 537}
241 538
242static void 539INLINE void
243encode_nl (enc_t *enc) 540encode_nl (enc_t *enc)
244{ 541{
245 if (enc->flags & F_INDENT) 542 if (enc->json.flags & F_INDENT)
246 { 543 {
247 need (enc, 1); 544 need (enc, 1);
248 encode_ch (enc, '\n'); 545 encode_ch (enc, '\n');
249 } 546 }
250} 547}
251 548
252static void 549INLINE void
253encode_comma (enc_t *enc) 550encode_comma (enc_t *enc)
254{ 551{
255 encode_ch (enc, ','); 552 encode_ch (enc, ',');
256 553
257 if (enc->flags & F_INDENT) 554 if (enc->json.flags & F_INDENT)
258 encode_nl (enc); 555 encode_nl (enc);
259 else if (enc->flags & F_SPACE_AFTER) 556 else if (enc->json.flags & F_SPACE_AFTER)
260 encode_space (enc); 557 encode_space (enc);
261} 558}
262 559
263static void encode_sv (enc_t *enc, SV *sv); 560static void encode_sv (enc_t *enc, SV *sv);
264 561
265static void 562static void
266encode_av (enc_t *enc, AV *av) 563encode_av (enc_t *enc, AV *av)
267{ 564{
268 int i, len = av_len (av); 565 int i, len = av_len (av);
269 566
270 encode_ch (enc, '['); encode_nl (enc); 567 if (enc->indent >= enc->json.max_depth)
271 ++enc->indent; 568 croak (ERR_NESTING_EXCEEDED);
272 569
570 encode_ch (enc, '[');
571
572 if (len >= 0)
573 {
574 encode_nl (enc); ++enc->indent;
575
273 for (i = 0; i <= len; ++i) 576 for (i = 0; i <= len; ++i)
274 { 577 {
578 SV **svp = av_fetch (av, i, 0);
579
275 encode_indent (enc); 580 encode_indent (enc);
276 encode_sv (enc, *av_fetch (av, i, 0));
277 581
582 if (svp)
583 encode_sv (enc, *svp);
584 else
585 encode_str (enc, "null", 4, 0);
586
278 if (i < len) 587 if (i < len)
279 encode_comma (enc); 588 encode_comma (enc);
280 } 589 }
281 590
591 encode_nl (enc); --enc->indent; encode_indent (enc);
592 }
593
282 encode_nl (enc); 594 encode_ch (enc, ']');
283
284 --enc->indent;
285 encode_indent (enc); encode_ch (enc, ']');
286} 595}
287 596
288static void 597static void
289encode_he (enc_t *enc, HE *he) 598encode_hk (enc_t *enc, HE *he)
290{ 599{
291 encode_ch (enc, '"'); 600 encode_ch (enc, '"');
292 601
293 if (HeKLEN (he) == HEf_SVKEY) 602 if (HeKLEN (he) == HEf_SVKEY)
294 { 603 {
295 SV *sv = HeSVKEY (he); 604 SV *sv = HeSVKEY (he);
296 STRLEN len; 605 STRLEN len;
297 char *str; 606 char *str;
298 607
299 SvGETMAGIC (sv); 608 SvGETMAGIC (sv);
300 str = SvPV (sv, len); 609 str = SvPV (sv, len);
301 610
302 encode_str (enc, str, len, SvUTF8 (sv)); 611 encode_str (enc, str, len, SvUTF8 (sv));
303 } 612 }
304 else 613 else
305 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 614 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
306 615
307 encode_ch (enc, '"'); 616 encode_ch (enc, '"');
308 617
309 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 618 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
310 encode_ch (enc, ':'); 619 encode_ch (enc, ':');
311 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 620 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
312 encode_sv (enc, HeVAL (he));
313} 621}
314 622
315// compare hash entries, used when all keys are bytestrings 623// compare hash entries, used when all keys are bytestrings
316static int 624static int
317he_cmp_fast (const void *a_, const void *b_) 625he_cmp_fast (const void *a_, const void *b_)
322 HE *b = *(HE **)b_; 630 HE *b = *(HE **)b_;
323 631
324 STRLEN la = HeKLEN (a); 632 STRLEN la = HeKLEN (a);
325 STRLEN lb = HeKLEN (b); 633 STRLEN lb = HeKLEN (b);
326 634
327 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 635 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
328 cmp = la - lb; 636 cmp = lb - la;
329 637
330 return cmp; 638 return cmp;
331} 639}
332 640
333// 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
334static int 642static int
335he_cmp_slow (const void *a, const void *b) 643he_cmp_slow (const void *a, const void *b)
336{ 644{
337 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 645 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
338} 646}
339 647
340static void 648static void
341encode_hv (enc_t *enc, HV *hv) 649encode_hv (enc_t *enc, HV *hv)
342{ 650{
343 int count, i; 651 HE *he;
344 652
345 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 653 if (enc->indent >= enc->json.max_depth)
654 croak (ERR_NESTING_EXCEEDED);
346 655
347 if ((count = hv_iterinit (hv))) 656 encode_ch (enc, '{');
348 { 657
349 // for canonical output we have to sort by keys first 658 // for canonical output we have to sort by keys first
350 // actually, this is mostly due to the stupid so-called 659 // actually, this is mostly due to the stupid so-called
351 // security workaround added somewhere in 5.8.x. 660 // security workaround added somewhere in 5.8.x
352 // that randomises hash orderings 661 // that randomises hash orderings
353 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))
354 { 667 {
355 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode 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 {
356 int fast = 1; 682 int i, fast = 1;
683 HE *hes_stack [STACK_HES];
684 HE **hes = hes_stack;
685
686 // allocate larger arrays on the heap
687 if (count > STACK_HES)
688 {
689 SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes)));
690 hes = (HE **)SvPVX (sv);
691 }
357 692
358 i = 0; 693 i = 0;
359 while ((he = hv_iternext (hv))) 694 while ((he = hv_iternext (hv)))
360 { 695 {
361 hes [i++] = he; 696 hes [i++] = he;
383 718
384 FREETMPS; 719 FREETMPS;
385 LEAVE; 720 LEAVE;
386 } 721 }
387 722
388 for (i = 0; i < count; ++i) 723 encode_nl (enc); ++enc->indent;
724
725 while (count--)
389 { 726 {
390 encode_indent (enc); 727 encode_indent (enc);
728 he = hes [count];
391 encode_he (enc, hes [i]); 729 encode_hk (enc, he);
730 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
392 731
393 if (i < count - 1) 732 if (count)
394 encode_comma (enc); 733 encode_comma (enc);
395 } 734 }
396 735
397 encode_nl (enc); 736 encode_nl (enc); --enc->indent; encode_indent (enc);
398 } 737 }
738 }
739 else
740 {
741 if (hv_iterinit (hv) || SvMAGICAL (hv))
742 if ((he = hv_iternext (hv)))
743 {
744 encode_nl (enc); ++enc->indent;
745
746 for (;;)
747 {
748 encode_indent (enc);
749 encode_hk (enc, he);
750 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
751
752 if (!(he = hv_iternext (hv)))
753 break;
754
755 encode_comma (enc);
756 }
757
758 encode_nl (enc); --enc->indent; encode_indent (enc);
759 }
760 }
761
762 encode_ch (enc, '}');
763}
764
765// encode objects, arrays and special \0=false and \1=true values.
766static void
767encode_rv (enc_t *enc, SV *sv)
768{
769 svtype svt;
770 GV *method;
771
772 SvGETMAGIC (sv);
773 svt = SvTYPE (sv);
774
775 if (expect_false (SvOBJECT (sv)))
776 {
777 HV *stash = SvSTASH (sv);
778
779 if (stash == bool_stash)
780 {
781 if (SvIV (sv))
782 encode_str (enc, "true", 4, 0);
783 else
784 encode_str (enc, "false", 5, 0);
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);
399 else 853 else
400 { 854 croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)",
401 SV *sv; 855 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
402 HE *he = hv_iternext (hv); 856 }
403 857 else if (svt == SVt_PVHV)
404 for (;;) 858 encode_hv (enc, (HV *)sv);
405 { 859 else if (svt == SVt_PVAV)
406 encode_indent (enc); 860 encode_av (enc, (AV *)sv);
407 encode_he (enc, he); 861 else if (svt < SVt_PVAV)
408
409 if (!(he = hv_iternext (hv)))
410 break;
411
412 encode_comma (enc);
413 }
414
415 encode_nl (enc);
416 }
417 } 862 {
863 int bool_type = ref_bool_type (sv);
418 864
419 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 865 if (bool_type == 1)
866 encode_str (enc, "true", 4, 0);
867 else if (bool_type == 0)
868 encode_str (enc, "false", 5, 0);
869 else if (enc->json.flags & F_ALLOW_UNKNOWN)
870 encode_str (enc, "null", 4, 0);
871 else
872 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
873 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
874 }
875 else if (enc->json.flags & F_ALLOW_UNKNOWN)
876 encode_str (enc, "null", 4, 0);
877 else
878 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
879 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
420} 880}
421 881
422static void 882static void
423encode_sv (enc_t *enc, SV *sv) 883encode_sv (enc_t *enc, SV *sv)
424{ 884{
432 encode_str (enc, str, len, SvUTF8 (sv)); 892 encode_str (enc, str, len, SvUTF8 (sv));
433 encode_ch (enc, '"'); 893 encode_ch (enc, '"');
434 } 894 }
435 else if (SvNOKp (sv)) 895 else if (SvNOKp (sv))
436 { 896 {
897 // trust that perl will do the right thing w.r.t. JSON syntax.
437 need (enc, NV_DIG + 32); 898 need (enc, NV_DIG + 32);
438 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 899 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
439 enc->cur += strlen (enc->cur); 900 enc->cur += strlen (enc->cur);
440 } 901 }
441 else if (SvIOKp (sv)) 902 else if (SvIOKp (sv))
442 { 903 {
904 // we assume we can always read an IV as a UV and vice versa
905 // we assume two's complement
906 // we assume no aliasing issues in the union
907 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
908 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
909 {
910 // optimise the "small number case"
911 // code will likely be branchless and use only a single multiplication
912 // works for numbers up to 59074
913 I32 i = SvIVX (sv);
914 U32 u;
915 char digit, nz = 0;
916
443 need (enc, 64); 917 need (enc, 6);
918
919 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
920 u = i < 0 ? -i : i;
921
922 // convert to 4.28 fixed-point representation
923 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
924
925 // now output digit by digit, each time masking out the integer part
926 // and multiplying by 5 while moving the decimal point one to the right,
927 // resulting in a net multiplication by 10.
928 // we always write the digit to memory but conditionally increment
929 // the pointer, to enable the use of conditional move instructions.
930 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
931 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
932 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
933 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
934 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
935 }
936 else
937 {
938 // large integer, use the (rather slow) snprintf way.
939 need (enc, IVUV_MAXCHARS);
444 enc->cur += 940 enc->cur +=
445 SvIsUV(sv) 941 SvIsUV(sv)
446 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 942 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
447 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 943 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
944 }
448 } 945 }
449 else if (SvROK (sv)) 946 else if (SvROK (sv))
450 { 947 encode_rv (enc, SvRV (sv));
451 SV *rv = SvRV (sv); 948 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
452
453 if (enc->indent >= enc->max_depth)
454 croak ("data structure too deep (hit recursion limit)");
455
456 switch (SvTYPE (rv))
457 {
458 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
459 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
460
461 default:
462 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
463 SvPV_nolen (sv));
464 }
465 }
466 else if (!SvOK (sv))
467 encode_str (enc, "null", 4, 0); 949 encode_str (enc, "null", 4, 0);
468 else 950 else
469 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",
470 SvPV_nolen (sv), SvFLAGS (sv)); 952 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
471} 953}
472 954
473static SV * 955static SV *
474encode_json (SV *scalar, UV flags) 956encode_json (SV *scalar, JSON *json)
475{ 957{
958 enc_t enc;
959
476 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 960 if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar))
477 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)");
478 962
479 enc_t enc; 963 enc.json = *json;
480 enc.flags = flags;
481 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 964 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
482 enc.cur = SvPVX (enc.sv); 965 enc.cur = SvPVX (enc.sv);
483 enc.end = SvEND (enc.sv); 966 enc.end = SvEND (enc.sv);
484 enc.indent = 0; 967 enc.indent = 0;
485 enc.max_depth = 0x7fffffffUL; 968 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
969 : enc.json.flags & F_LATIN1 ? 0x000100UL
970 : 0x110000UL;
486 971
487 SvPOK_only (enc.sv); 972 SvPOK_only (enc.sv);
488 encode_sv (&enc, scalar); 973 encode_sv (&enc, scalar);
974 encode_nl (&enc);
489 975
976 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
977 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
978
490 if (!(flags & (F_ASCII | F_UTF8))) 979 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
491 SvUTF8_on (enc.sv); 980 SvUTF8_on (enc.sv);
492 981
493 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
494
495 if (enc.flags & F_SHRINK) 982 if (enc.json.flags & F_SHRINK)
496 shrink (enc.sv); 983 shrink (enc.sv);
497 984
498 return enc.sv; 985 return enc.sv;
499} 986}
500 987
505typedef struct 992typedef struct
506{ 993{
507 char *cur; // current parser pointer 994 char *cur; // current parser pointer
508 char *end; // end of input string 995 char *end; // end of input string
509 const char *err; // parse error, if != 0 996 const char *err; // parse error, if != 0
510 UV flags; // F_* 997 JSON json;
998 U32 depth; // recursion depth
999 U32 maxdepth; // recursion depth limit
511} dec_t; 1000} dec_t;
512 1001
513static 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
514decode_ws (dec_t *dec) 1012decode_ws (dec_t *dec)
515{ 1013{
516 for (;;) 1014 for (;;)
517 { 1015 {
518 char ch = *dec->cur; 1016 char ch = *dec->cur;
519 1017
520 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 }
521 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 1030 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
522 break; 1031 break; // parse error, but let higher level handle it, gives better error messages
523 1032
524 ++dec->cur; 1033 ++dec->cur;
525 } 1034 }
526} 1035}
527 1036
528#define ERR(reason) SB dec->err = reason; goto fail; SE 1037#define ERR(reason) SB dec->err = reason; goto fail; SE
1038
529#define EXPECT_CH(ch) SB \ 1039#define EXPECT_CH(ch) SB \
530 if (*dec->cur != ch) \ 1040 if (*dec->cur != ch) \
531 ERR (# ch " expected"); \ 1041 ERR (# ch " expected"); \
532 ++dec->cur; \ 1042 ++dec->cur; \
533 SE 1043 SE
534 1044
1045#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
1046#define DEC_DEC_DEPTH --dec->depth
1047
535static SV *decode_sv (dec_t *dec); 1048static SV *decode_sv (dec_t *dec);
536 1049
537static signed char decode_hexdigit[256]; 1050static signed char decode_hexdigit[256];
538 1051
539static UV 1052static UV
540decode_4hex (dec_t *dec) 1053decode_4hex (dec_t *dec)
541{ 1054{
542 signed char d1, d2, d3, d4; 1055 signed char d1, d2, d3, d4;
543 unsigned char *cur = (unsigned char *)dec->cur; 1056 unsigned char *cur = (unsigned char *)dec->cur;
544 1057
545 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 1058 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
546 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 1059 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
547 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 1060 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
548 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 1061 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
549 1062
550 dec->cur += 4; 1063 dec->cur += 4;
551 1064
552 return ((UV)d1) << 12 1065 return ((UV)d1) << 12
553 | ((UV)d2) << 8 1066 | ((UV)d2) << 8
561static SV * 1074static SV *
562decode_str (dec_t *dec) 1075decode_str (dec_t *dec)
563{ 1076{
564 SV *sv = 0; 1077 SV *sv = 0;
565 int utf8 = 0; 1078 int utf8 = 0;
1079 char *dec_cur = dec->cur;
566 1080
567 do 1081 do
568 { 1082 {
569 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 1083 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
570 char *cur = buf; 1084 char *cur = buf;
571 1085
572 do 1086 do
573 { 1087 {
574 unsigned char ch = *(unsigned char *)dec->cur++; 1088 unsigned char ch = *(unsigned char *)dec_cur++;
575 1089
576 if (ch == '"') 1090 if (expect_false (ch == '"'))
577 { 1091 {
578 --dec->cur; 1092 --dec_cur;
579 break; 1093 break;
580 } 1094 }
581 else if (ch == '\\') 1095 else if (expect_false (ch == '\\'))
582 { 1096 {
583 switch (*dec->cur) 1097 switch (*dec_cur)
584 { 1098 {
585 case '\\': 1099 case '\\':
586 case '/': 1100 case '/':
587 case '"': *cur++ = *dec->cur++; break; 1101 case '"': *cur++ = *dec_cur++; break;
588 1102
589 case 'b': ++dec->cur; *cur++ = '\010'; break; 1103 case 'b': ++dec_cur; *cur++ = '\010'; break;
590 case 't': ++dec->cur; *cur++ = '\011'; break; 1104 case 't': ++dec_cur; *cur++ = '\011'; break;
591 case 'n': ++dec->cur; *cur++ = '\012'; break; 1105 case 'n': ++dec_cur; *cur++ = '\012'; break;
592 case 'f': ++dec->cur; *cur++ = '\014'; break; 1106 case 'f': ++dec_cur; *cur++ = '\014'; break;
593 case 'r': ++dec->cur; *cur++ = '\015'; break; 1107 case 'r': ++dec_cur; *cur++ = '\015'; break;
594 1108
595 case 'u': 1109 case 'u':
596 { 1110 {
597 UV lo, hi; 1111 UV lo, hi;
598 ++dec->cur; 1112 ++dec_cur;
599 1113
1114 dec->cur = dec_cur;
600 hi = decode_4hex (dec); 1115 hi = decode_4hex (dec);
1116 dec_cur = dec->cur;
601 if (hi == (UV)-1) 1117 if (hi == (UV)-1)
602 goto fail; 1118 goto fail;
603 1119
604 // possibly a surrogate pair 1120 // possibly a surrogate pair
605 if (hi >= 0xd800) 1121 if (hi >= 0xd800)
606 if (hi < 0xdc00) 1122 if (hi < 0xdc00)
607 { 1123 {
608 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 1124 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
609 ERR ("missing low surrogate character in surrogate pair"); 1125 ERR ("missing low surrogate character in surrogate pair");
610 1126
611 dec->cur += 2; 1127 dec_cur += 2;
612 1128
1129 dec->cur = dec_cur;
613 lo = decode_4hex (dec); 1130 lo = decode_4hex (dec);
1131 dec_cur = dec->cur;
614 if (lo == (UV)-1) 1132 if (lo == (UV)-1)
615 goto fail; 1133 goto fail;
616 1134
617 if (lo < 0xdc00 || lo >= 0xe000) 1135 if (lo < 0xdc00 || lo >= 0xe000)
618 ERR ("surrogate pair expected"); 1136 ERR ("surrogate pair expected");
624 1142
625 if (hi >= 0x80) 1143 if (hi >= 0x80)
626 { 1144 {
627 utf8 = 1; 1145 utf8 = 1;
628 1146
629 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 1147 cur = encode_utf8 (cur, hi);
630 } 1148 }
631 else 1149 else
632 *cur++ = hi; 1150 *cur++ = hi;
633 } 1151 }
634 break; 1152 break;
635 1153
636 default: 1154 default:
637 --dec->cur; 1155 --dec_cur;
638 ERR ("illegal backslash escape sequence in string"); 1156 ERR ("illegal backslash escape sequence in string");
639 } 1157 }
640 } 1158 }
641 else if (ch >= 0x20 && ch <= 0x7f) 1159 else if (expect_true (ch >= 0x20 && ch < 0x80))
642 *cur++ = ch; 1160 *cur++ = ch;
643 else if (ch >= 0x80) 1161 else if (ch >= 0x80)
644 { 1162 {
645 --dec->cur;
646
647 STRLEN clen; 1163 STRLEN clen;
1164
1165 --dec_cur;
1166
648 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 1167 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
649 if (clen == (STRLEN)-1) 1168 if (clen == (STRLEN)-1)
650 ERR ("malformed UTF-8 character in JSON string"); 1169 ERR ("malformed UTF-8 character in JSON string");
651 1170
652 do 1171 do
653 {
654 *cur++ = *dec->cur++; 1172 *cur++ = *dec_cur++;
655 }
656 while (--clen); 1173 while (--clen);
657 1174
658 utf8 = 1; 1175 utf8 = 1;
659 } 1176 }
660 else if (!ch) 1177 else if (ch == '\t' && dec->json.flags & F_RELAXED)
661 ERR ("unexpected end of string while parsing json string"); 1178 *cur++ = ch;
662 else 1179 else
1180 {
1181 --dec_cur;
1182
1183 if (!ch)
1184 ERR ("unexpected end of string while parsing JSON string");
1185 else
663 ERR ("invalid character encountered"); 1186 ERR ("invalid character encountered while parsing JSON string");
664 1187 }
665 } 1188 }
666 while (cur < buf + SHORT_STRING_LEN); 1189 while (cur < buf + SHORT_STRING_LEN);
667 1190
1191 {
668 STRLEN len = cur - buf; 1192 STRLEN len = cur - buf;
669 1193
670 if (sv) 1194 if (sv)
671 { 1195 {
672 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
673 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1201 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
674 SvCUR_set (sv, SvCUR (sv) + len); 1202 SvCUR_set (sv, SvCUR (sv) + len);
675 } 1203 }
676 else 1204 else
677 sv = newSVpvn (buf, len); 1205 sv = newSVpvn (buf, len);
678 } 1206 }
1207 }
679 while (*dec->cur != '"'); 1208 while (*dec_cur != '"');
680 1209
681 ++dec->cur; 1210 ++dec_cur;
682 1211
683 if (sv) 1212 if (sv)
684 { 1213 {
685 SvPOK_only (sv); 1214 SvPOK_only (sv);
686 *SvEND (sv) = 0; 1215 *SvEND (sv) = 0;
689 SvUTF8_on (sv); 1218 SvUTF8_on (sv);
690 } 1219 }
691 else 1220 else
692 sv = newSVpvn ("", 0); 1221 sv = newSVpvn ("", 0);
693 1222
1223 dec->cur = dec_cur;
694 return sv; 1224 return sv;
695 1225
696fail: 1226fail:
1227 dec->cur = dec_cur;
697 return 0; 1228 return 0;
698} 1229}
699 1230
700static SV * 1231static SV *
701decode_num (dec_t *dec) 1232decode_num (dec_t *dec)
759 is_nv = 1; 1290 is_nv = 1;
760 } 1291 }
761 1292
762 if (!is_nv) 1293 if (!is_nv)
763 { 1294 {
764 UV uv; 1295 int len = dec->cur - start;
765 int numtype = grok_number (start, dec->cur - start, &uv); 1296
766 if (numtype & IS_NUMBER_IN_UV) 1297 // special case the rather common 1..5-digit-int case
767 if (numtype & IS_NUMBER_NEG) 1298 if (*start == '-')
1299 switch (len)
768 { 1300 {
769 if (uv < (UV)IV_MIN) 1301 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
770 return newSViv (-(IV)uv); 1302 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1303 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
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));
771 } 1306 }
1307 else
1308 switch (len)
1309 {
1310 case 1: return newSViv ( start [0] - '0' * 1);
1311 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1312 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
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);
1315 }
1316
1317 {
1318 UV uv;
1319 int numtype = grok_number (start, len, &uv);
1320 if (numtype & IS_NUMBER_IN_UV)
1321 if (numtype & IS_NUMBER_NEG)
1322 {
1323 if (uv < (UV)IV_MIN)
1324 return newSViv (-(IV)uv);
1325 }
772 else 1326 else
773 return newSVuv (uv); 1327 return newSVuv (uv);
774 } 1328 }
775 1329
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
776 return newSVnv (Atof (start)); 1342 return newSVnv (json_atof (start));
777 1343
778fail: 1344fail:
779 return 0; 1345 return 0;
780} 1346}
781 1347
782static SV * 1348static SV *
783decode_av (dec_t *dec) 1349decode_av (dec_t *dec)
784{ 1350{
785 AV *av = newAV (); 1351 AV *av = newAV ();
786 1352
1353 DEC_INC_DEPTH;
787 decode_ws (dec); 1354 decode_ws (dec);
1355
788 if (*dec->cur == ']') 1356 if (*dec->cur == ']')
789 ++dec->cur; 1357 ++dec->cur;
790 else 1358 else
791 for (;;) 1359 for (;;)
792 { 1360 {
803 if (*dec->cur == ']') 1371 if (*dec->cur == ']')
804 { 1372 {
805 ++dec->cur; 1373 ++dec->cur;
806 break; 1374 break;
807 } 1375 }
808 1376
809 if (*dec->cur != ',') 1377 if (*dec->cur != ',')
810 ERR (", or ] expected while parsing array"); 1378 ERR (", or ] expected while parsing array");
811 1379
812 ++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 }
813 } 1389 }
814 1390
1391 DEC_DEC_DEPTH;
815 return newRV_noinc ((SV *)av); 1392 return newRV_noinc ((SV *)av);
816 1393
817fail: 1394fail:
818 SvREFCNT_dec (av); 1395 SvREFCNT_dec (av);
1396 DEC_DEC_DEPTH;
819 return 0; 1397 return 0;
820} 1398}
821 1399
822static SV * 1400static SV *
823decode_hv (dec_t *dec) 1401decode_hv (dec_t *dec)
824{ 1402{
1403 SV *sv;
825 HV *hv = newHV (); 1404 HV *hv = newHV ();
826 1405
1406 DEC_INC_DEPTH;
827 decode_ws (dec); 1407 decode_ws (dec);
1408
828 if (*dec->cur == '}') 1409 if (*dec->cur == '}')
829 ++dec->cur; 1410 ++dec->cur;
830 else 1411 else
831 for (;;) 1412 for (;;)
832 { 1413 {
1414 EXPECT_CH ('"');
1415
1416 // heuristic: assume that
1417 // a) decode_str + hv_store_ent are abysmally slow.
1418 // b) most hash keys are short, simple ascii text.
1419 // => try to "fast-match" such strings to avoid
1420 // the overhead of decode_str + hv_store_ent.
1421 {
833 SV *key, *value; 1422 SV *value;
1423 char *p = dec->cur;
1424 char *e = p + 24; // only try up to 24 bytes
834 1425
835 decode_ws (dec); EXPECT_CH ('"'); 1426 for (;;)
836
837 key = decode_str (dec);
838 if (!key)
839 goto fail;
840
841 decode_ws (dec); EXPECT_CH (':');
842
843 value = decode_sv (dec);
844 if (!value)
845 { 1427 {
1428 // the >= 0x80 is false on most architectures
1429 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1430 {
1431 // slow path, back up and use decode_str
1432 SV *key = decode_str (dec);
1433 if (!key)
1434 goto fail;
1435
1436 decode_ws (dec); EXPECT_CH (':');
1437
1438 decode_ws (dec);
1439 value = decode_sv (dec);
1440 if (!value)
1441 {
1442 SvREFCNT_dec (key);
1443 goto fail;
1444 }
1445
1446 hv_store_ent (hv, key, value, 0);
846 SvREFCNT_dec (key); 1447 SvREFCNT_dec (key);
1448
1449 break;
1450 }
1451 else if (*p == '"')
1452 {
1453 // fast path, got a simple key
1454 char *key = dec->cur;
1455 int len = p - key;
1456 dec->cur = p + 1;
1457
1458 decode_ws (dec); EXPECT_CH (':');
1459
1460 decode_ws (dec);
1461 value = decode_sv (dec);
1462 if (!value)
847 goto fail; 1463 goto fail;
1464
1465 hv_store (hv, key, len, value, 0);
1466
1467 break;
1468 }
1469
1470 ++p;
848 } 1471 }
849 1472 }
850 //TODO: optimise
851 hv_store_ent (hv, key, value, 0);
852 1473
853 decode_ws (dec); 1474 decode_ws (dec);
854 1475
855 if (*dec->cur == '}') 1476 if (*dec->cur == '}')
856 { 1477 {
860 1481
861 if (*dec->cur != ',') 1482 if (*dec->cur != ',')
862 ERR (", or } expected while parsing object/hash"); 1483 ERR (", or } expected while parsing object/hash");
863 1484
864 ++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 }
865 } 1494 }
866 1495
1496 DEC_DEC_DEPTH;
867 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;
868 1564
869fail: 1565fail:
870 SvREFCNT_dec (hv); 1566 SvREFCNT_dec (hv);
1567 DEC_DEC_DEPTH;
1568 return 0;
1569}
1570
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);
871 return 0; 1651 return 0;
872} 1652}
873 1653
874static SV * 1654static SV *
875decode_sv (dec_t *dec) 1655decode_sv (dec_t *dec)
876{ 1656{
877 decode_ws (dec); 1657 // the beauty of JSON: you need exactly one character lookahead
1658 // to parse everything.
878 switch (*dec->cur) 1659 switch (*dec->cur)
879 { 1660 {
880 case '"': ++dec->cur; return decode_str (dec); 1661 case '"': ++dec->cur; return decode_str (dec);
881 case '[': ++dec->cur; return decode_av (dec); 1662 case '[': ++dec->cur; return decode_av (dec);
882 case '{': ++dec->cur; return decode_hv (dec); 1663 case '{': ++dec->cur; return decode_hv (dec);
1664 case '(': return decode_tag (dec);
883 1665
884 case '-': 1666 case '-':
885 case '0': case '1': case '2': case '3': case '4': 1667 case '0': case '1': case '2': case '3': case '4':
886 case '5': case '6': case '7': case '8': case '9': 1668 case '5': case '6': case '7': case '8': case '9':
887 return decode_num (dec); 1669 return decode_num (dec);
888 1670
889 case 't': 1671 case 't':
890 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1672 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
891 { 1673 {
892 dec->cur += 4; 1674 dec->cur += 4;
1675#if JSON_SLOW
1676 bool_true = get_bool ("Types::Serialiser::true");
1677#endif
893 return newSViv (1); 1678 return newSVsv (bool_true);
894 } 1679 }
895 else 1680 else
896 ERR ("'true' expected"); 1681 ERR ("'true' expected");
897 1682
898 break; 1683 break;
899 1684
900 case 'f': 1685 case 'f':
901 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1686 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
902 { 1687 {
903 dec->cur += 5; 1688 dec->cur += 5;
1689#if JSON_SLOW
1690 bool_false = get_bool ("Types::Serialiser::false");
1691#endif
904 return newSViv (0); 1692 return newSVsv (bool_false);
905 } 1693 }
906 else 1694 else
907 ERR ("'false' expected"); 1695 ERR ("'false' expected");
908 1696
909 break; 1697 break;
918 ERR ("'null' expected"); 1706 ERR ("'null' expected");
919 1707
920 break; 1708 break;
921 1709
922 default: 1710 default:
923 ERR ("malformed json string, neither array, object, number, string or atom"); 1711 ERR ("malformed JSON string, neither tag, array, object, number, string or atom");
924 break; 1712 break;
925 } 1713 }
926 1714
927fail: 1715fail:
928 return 0; 1716 return 0;
929} 1717}
930 1718
931static SV * 1719static SV *
932decode_json (SV *string, UV flags) 1720decode_json (SV *string, JSON *json, STRLEN *offset_return)
933{ 1721{
1722 dec_t dec;
934 SV *sv; 1723 SV *sv;
935 1724
936 if (flags & F_UTF8) 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 */
1733 /*SvGETMAGIC (string);*/
1734 if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string))
1735 string = sv_2mortal (newSVsv (string));
1736
1737 SvUPGRADE (string, SVt_PV);
1738
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)
1757 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1758 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1759 }
1760
1761 if (DECODE_WANTS_OCTETS (json))
937 sv_utf8_downgrade (string, 0); 1762 sv_utf8_downgrade (string, 0);
938 else 1763 else
939 sv_utf8_upgrade (string); 1764 sv_utf8_upgrade (string);
940 1765
941 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1766 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
942 1767
943 dec_t dec; 1768 dec.json = *json;
944 dec.flags = flags;
945 dec.cur = SvPVX (string); 1769 dec.cur = SvPVX (string);
946 dec.end = SvEND (string); 1770 dec.end = SvEND (string);
947 dec.err = 0; 1771 dec.err = 0;
1772 dec.depth = 0;
948 1773
1774 if (dec.json.cb_object || dec.json.cb_sk_object)
1775 dec.json.flags |= F_HOOK;
1776
1777 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1778
1779 decode_ws (&dec);
949 sv = decode_sv (&dec); 1780 sv = decode_sv (&dec);
950 1781
1782 if (offset_return)
1783 *offset_return = dec.cur - SvPVX (string);
1784 else if (sv)
1785 {
1786 // check for trailing garbage
1787 decode_ws (&dec);
1788
1789 if (*dec.cur)
1790 {
1791 dec.err = "garbage after JSON object";
1792 SvREFCNT_dec (sv);
1793 sv = 0;
1794 }
1795 }
1796
951 if (!sv) 1797 if (!sv)
952 { 1798 {
953 IV offset = dec.flags & F_UTF8
954 ? dec.cur - SvPVX (string)
955 : utf8_distance (dec.cur, SvPVX (string));
956 SV *uni = sv_newmortal (); 1799 SV *uni = sv_newmortal ();
957 1800
958 // horrible hack to silence warning inside pv_uni_display 1801 // horrible hack to silence warning inside pv_uni_display
959 COP cop = *PL_curcop; 1802 COP cop = *PL_curcop;
960 cop.cop_warnings = pWARN_NONE; 1803 cop.cop_warnings = pWARN_NONE;
962 SAVEVPTR (PL_curcop); 1805 SAVEVPTR (PL_curcop);
963 PL_curcop = &cop; 1806 PL_curcop = &cop;
964 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);
965 LEAVE; 1808 LEAVE;
966 1809
967 croak ("%s, at character offset %d (%s)", 1810 croak ("%s, at character offset %d (before \"%s\")",
968 dec.err, 1811 dec.err,
969 (int)offset, 1812 (int)ptr_to_index (string, dec.cur),
970 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1813 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
971 } 1814 }
972 1815
973 sv = sv_2mortal (sv); 1816 sv = sv_2mortal (sv);
974 1817
975 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1818 if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv))
976 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)");
977 1820
978 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
979} 1980}
980 1981
981///////////////////////////////////////////////////////////////////////////// 1982/////////////////////////////////////////////////////////////////////////////
982// XS interface functions 1983// XS interface functions
983 1984
985 1986
986BOOT: 1987BOOT:
987{ 1988{
988 int i; 1989 int i;
989 1990
990 memset (decode_hexdigit, 0xff, 256);
991 for (i = 10; i--; ) 1991 for (i = 0; i < 256; ++i)
992 decode_hexdigit ['0' + i] = i; 1992 decode_hexdigit [i] =
1993 i >= '0' && i <= '9' ? i - '0'
1994 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1995 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1996 : -1;
993 1997
994 for (i = 7; i--; )
995 {
996 decode_hexdigit ['a' + i] = 10 + i;
997 decode_hexdigit ['A' + i] = 10 + i;
998 }
999
1000 json_stash = gv_stashpv ("JSON::XS", 1); 1998 json_stash = gv_stashpv ("JSON::XS" , 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");
2002
2003 sv_json = newSVpv ("JSON", 0);
2004 SvREADONLY_on (sv_json);
2005
2006 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1001} 2007}
1002 2008
1003PROTOTYPES: DISABLE 2009PROTOTYPES: DISABLE
1004 2010
1005SV *new (char *dummy) 2011void CLONE (...)
1006 CODE: 2012 CODE:
1007 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 2013 json_stash = 0;
2014 bool_stash = 0;
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
2028void ascii (JSON *self, int enable = 1)
2029 ALIAS:
2030 ascii = F_ASCII
2031 latin1 = F_LATIN1
2032 utf8 = F_UTF8
2033 indent = F_INDENT
2034 canonical = F_CANONICAL
2035 space_before = F_SPACE_BEFORE
2036 space_after = F_SPACE_AFTER
2037 pretty = F_PRETTY
2038 allow_nonref = F_ALLOW_NONREF
2039 shrink = F_SHRINK
2040 allow_blessed = F_ALLOW_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)
2080 CODE:
2081 RETVAL = self->max_depth;
1008 OUTPUT: 2082 OUTPUT:
1009 RETVAL 2083 RETVAL
1010 2084
1011SV *ascii (SV *self, int enable = 1) 2085void max_size (JSON *self, U32 max_size = 0)
1012 ALIAS: 2086 PPCODE:
1013 ascii = F_ASCII 2087 self->max_size = max_size;
1014 utf8 = F_UTF8 2088 XPUSHs (ST (0));
1015 indent = F_INDENT 2089
1016 canonical = F_CANONICAL 2090int get_max_size (JSON *self)
1017 space_before = F_SPACE_BEFORE
1018 space_after = F_SPACE_AFTER
1019 pretty = F_PRETTY
1020 allow_nonref = F_ALLOW_NONREF
1021 shrink = F_SHRINK
1022 CODE: 2091 CODE:
1023{ 2092 RETVAL = self->max_size;
1024 UV *uv = SvJSON (self);
1025 if (enable)
1026 *uv |= ix;
1027 else
1028 *uv &= ~ix;
1029
1030 RETVAL = newSVsv (self);
1031}
1032 OUTPUT: 2093 OUTPUT:
1033 RETVAL 2094 RETVAL
1034 2095
1035void encode (SV *self, SV *scalar) 2096void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1036 PPCODE: 2097 PPCODE:
1037 XPUSHs (encode_json (scalar, *SvJSON (self))); 2098{
2099 SvREFCNT_dec (self->cb_object);
2100 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1038 2101
1039void decode (SV *self, SV *jsonstr) 2102 XPUSHs (ST (0));
2103}
2104
2105void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1040 PPCODE: 2106 PPCODE:
1041 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 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
2239 CODE:
2240{
2241 if (self->incr_pos)
2242 croak ("incr_text can not be called when the incremental parser already started parsing");
2243
2244 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
2245}
2246 OUTPUT:
2247 RETVAL
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);
1042 2276
1043PROTOTYPES: ENABLE 2277PROTOTYPES: ENABLE
1044 2278
1045void to_json (SV *scalar) 2279void encode_json (SV *scalar)
1046 ALIAS:
1047 objToJson = 0
1048 PPCODE: 2280 PPCODE:
1049 XPUSHs (encode_json (scalar, F_UTF8)); 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}
1050 2288
1051void from_json (SV *jsonstr) 2289void decode_json (SV *jsonstr)
1052 ALIAS:
1053 jsonToObj = 0
1054 PPCODE: 2290 PPCODE:
1055 XPUSHs (decode_json (jsonstr, F_UTF8)); 2291{
2292 JSON json;
2293 json_init (&json);
2294 json.flags |= F_UTF8;
2295 PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN;
2296 XPUSHs (jsonstr);
2297}
1056 2298

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines