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 1.14 by root, Sat Mar 24 22:55:43 2007 UTC vs.
Revision 1.102 by root, Sat Oct 10 01:48:50 2009 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "assert.h" 5#include <assert.h>
6#include "string.h" 6#include <string.h>
7#include "stdlib.h" 7#include <stdlib.h>
8#include <stdio.h>
9#include <limits.h>
10#include <float.h>
8 11
12#if defined(__BORLANDC__) || defined(_MSC_VER)
13# define snprintf _snprintf // C compilers have this in stdio.h
14#endif
15
16// some old perls do not have this, try to make it work, no
17// guarantees, though. if it breaks, you get to keep the pieces.
18#ifndef UTF8_MAXBYTES
19# define UTF8_MAXBYTES 13
20#endif
21
22// three extra for rounding, sign, and end of string
23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
24
9#define F_ASCII 0x00000001 25#define F_ASCII 0x00000001UL
26#define F_LATIN1 0x00000002UL
10#define F_UTF8 0x00000002 27#define F_UTF8 0x00000004UL
11#define F_INDENT 0x00000004 28#define F_INDENT 0x00000008UL
12#define F_CANONICAL 0x00000008 29#define F_CANONICAL 0x00000010UL
13#define F_SPACE_BEFORE 0x00000010 30#define F_SPACE_BEFORE 0x00000020UL
14#define F_SPACE_AFTER 0x00000020 31#define F_SPACE_AFTER 0x00000040UL
15#define F_ALLOW_NONREF 0x00000080 32#define F_ALLOW_NONREF 0x00000100UL
16#define F_SHRINK 0x00000100 33#define F_SHRINK 0x00000200UL
34#define F_ALLOW_BLESSED 0x00000400UL
35#define F_CONV_BLESSED 0x00000800UL
36#define F_RELAXED 0x00001000UL
37#define F_ALLOW_UNKNOWN 0x00002000UL
38#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing
17 39
18#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER 40#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER
19#define F_DEFAULT 0
20 41
21#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
22#define INDENT_STEP 3 // spaces per indentation level 43#define INDENT_STEP 3 // spaces per indentation level
23 44
24#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character
25#define SHORT_STRING_LEN 512 // special-case strings of up to this size 45#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
26 46
27#define SB do { 47#define SB do {
28#define SE } while (0) 48#define SE } while (0)
29 49
50#if __GNUC__ >= 3
51# define expect(expr,value) __builtin_expect ((expr), (value))
52# define INLINE static inline
53#else
54# define expect(expr,value) (expr)
55# define INLINE static
56#endif
57
58#define expect_false(expr) expect ((expr) != 0, 0)
59#define expect_true(expr) expect ((expr) != 0, 1)
60
61#define IN_RANGE_INC(type,val,beg,end) \
62 ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
63 <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
64
65#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)"
66
67#ifdef USE_ITHREADS
68# define JSON_SLOW 1
69# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1))
70#else
71# define JSON_SLOW 0
72# define JSON_STASH json_stash
73#endif
74
30static HV *json_stash; // JSON::XS:: 75static HV *json_stash, *json_boolean_stash; // JSON::XS::
76static SV *json_true, *json_false;
77
78enum {
79 INCR_M_WS = 0, // initial whitespace skipping, must be 0
80 INCR_M_STR, // inside string
81 INCR_M_BS, // inside backslash
82 INCR_M_JSON // outside anything, count nesting
83};
84
85#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
86
87typedef struct {
88 U32 flags;
89 U32 max_depth;
90 STRLEN max_size;
91
92 SV *cb_object;
93 HV *cb_sk_object;
94
95 // for the incremental parser
96 SV *incr_text; // the source text so far
97 STRLEN incr_pos; // the current offset into the text
98 int incr_nest; // {[]}-nesting level
99 unsigned char incr_mode;
100} JSON;
101
102INLINE void
103json_init (JSON *json)
104{
105 Zero (json, 1, JSON);
106 json->max_depth = 512;
107}
31 108
32///////////////////////////////////////////////////////////////////////////// 109/////////////////////////////////////////////////////////////////////////////
33// utility functions 110// utility functions
34 111
35static UV * 112INLINE SV *
36SvJSON (SV *sv) 113get_bool (const char *name)
37{ 114{
38 if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) 115 SV *sv = get_sv (name, 1);
39 croak ("object is not of type JSON::XS");
40 116
41 return &SvUVX (SvRV (sv)); 117 SvREADONLY_on (sv);
42} 118 SvREADONLY_on (SvRV (sv));
43 119
44static void 120 return sv;
121}
122
123INLINE void
45shrink (SV *sv) 124shrink (SV *sv)
46{ 125{
47 sv_utf8_downgrade (sv, 1); 126 sv_utf8_downgrade (sv, 1);
127
48 if (SvLEN (sv) > SvCUR (sv) + 1) 128 if (SvLEN (sv) > SvCUR (sv) + 1)
49 { 129 {
50#ifdef SvPV_shrink_to_cur 130#ifdef SvPV_shrink_to_cur
51 SvPV_shrink_to_cur (sv); 131 SvPV_shrink_to_cur (sv);
52#elif defined (SvPV_renew) 132#elif defined (SvPV_renew)
… …
58// decode an utf-8 character and return it, or (UV)-1 in 138// decode an utf-8 character and return it, or (UV)-1 in
59// case of an error. 139// case of an error.
60// we special-case "safe" characters from U+80 .. U+7FF, 140// we special-case "safe" characters from U+80 .. U+7FF,
61// but use the very good perl function to parse anything else. 141// but use the very good perl function to parse anything else.
62// note that we never call this function for a ascii codepoints 142// note that we never call this function for a ascii codepoints
63static UV 143INLINE UV
64decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) 144decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
65{ 145{
66 if (s[0] > 0xdf || s[0] < 0xc2) 146 if (expect_true (len >= 2
67 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); 147 && IN_RANGE_INC (char, s[0], 0xc2, 0xdf)
68 else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) 148 && IN_RANGE_INC (char, s[1], 0x80, 0xbf)))
69 { 149 {
70 *clen = 2; 150 *clen = 2;
71 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); 151 return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
72 } 152 }
73 else 153 else
74 return (UV)-1; 154 return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
155}
156
157// likewise for encoding, also never called for ascii codepoints
158// this function takes advantage of this fact, although current gccs
159// seem to optimise the check for >= 0x80 away anyways
160INLINE unsigned char *
161encode_utf8 (unsigned char *s, UV ch)
162{
163 if (expect_false (ch < 0x000080))
164 *s++ = ch;
165 else if (expect_true (ch < 0x000800))
166 *s++ = 0xc0 | ( ch >> 6),
167 *s++ = 0x80 | ( ch & 0x3f);
168 else if ( ch < 0x010000)
169 *s++ = 0xe0 | ( ch >> 12),
170 *s++ = 0x80 | ((ch >> 6) & 0x3f),
171 *s++ = 0x80 | ( ch & 0x3f);
172 else if ( ch < 0x110000)
173 *s++ = 0xf0 | ( ch >> 18),
174 *s++ = 0x80 | ((ch >> 12) & 0x3f),
175 *s++ = 0x80 | ((ch >> 6) & 0x3f),
176 *s++ = 0x80 | ( ch & 0x3f);
177
178 return s;
179}
180
181// convert offset pointer to character index, sv must be string
182static STRLEN
183ptr_to_index (SV *sv, char *offset)
184{
185 return SvUTF8 (sv)
186 ? utf8_distance (offset, SvPVX (sv))
187 : offset - SvPVX (sv);
75} 188}
76 189
77///////////////////////////////////////////////////////////////////////////// 190/////////////////////////////////////////////////////////////////////////////
78// encoder 191// encoder
79 192
… …
81typedef struct 194typedef struct
82{ 195{
83 char *cur; // SvPVX (sv) + current output position 196 char *cur; // SvPVX (sv) + current output position
84 char *end; // SvEND (sv) 197 char *end; // SvEND (sv)
85 SV *sv; // result scalar 198 SV *sv; // result scalar
86 UV flags; // F_* 199 JSON json;
87 int indent; // indentation level 200 U32 indent; // indentation level
88 int max_depth; // max. recursion level 201 UV limit; // escape character values >= this value when encoding
89} enc_t; 202} enc_t;
90 203
91static void 204INLINE void
92need (enc_t *enc, STRLEN len) 205need (enc_t *enc, STRLEN len)
93{ 206{
94 if (enc->cur + len >= enc->end) 207 if (expect_false (enc->cur + len >= enc->end))
95 { 208 {
96 STRLEN cur = enc->cur - SvPVX (enc->sv); 209 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
97 SvGROW (enc->sv, cur + len + 1); 210 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
98 enc->cur = SvPVX (enc->sv) + cur; 211 enc->cur = SvPVX (enc->sv) + cur;
99 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); 212 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
100 } 213 }
101} 214}
102 215
103static void 216INLINE void
104encode_ch (enc_t *enc, char ch) 217encode_ch (enc_t *enc, char ch)
105{ 218{
106 need (enc, 1); 219 need (enc, 1);
107 *enc->cur++ = ch; 220 *enc->cur++ = ch;
108} 221}
… …
116 229
117 while (str < end) 230 while (str < end)
118 { 231 {
119 unsigned char ch = *(unsigned char *)str; 232 unsigned char ch = *(unsigned char *)str;
120 233
121 if (ch >= 0x20 && ch < 0x80) // most common case 234 if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case
122 { 235 {
123 if (ch == '"') // but with slow exceptions 236 if (expect_false (ch == '"')) // but with slow exceptions
124 { 237 {
125 need (enc, len += 1); 238 need (enc, len += 1);
126 *enc->cur++ = '\\'; 239 *enc->cur++ = '\\';
127 *enc->cur++ = '"'; 240 *enc->cur++ = '"';
128 } 241 }
129 else if (ch == '\\') 242 else if (expect_false (ch == '\\'))
130 { 243 {
131 need (enc, len += 1); 244 need (enc, len += 1);
132 *enc->cur++ = '\\'; 245 *enc->cur++ = '\\';
133 *enc->cur++ = '\\'; 246 *enc->cur++ = '\\';
134 } 247 }
… …
152 STRLEN clen; 265 STRLEN clen;
153 UV uch; 266 UV uch;
154 267
155 if (is_utf8) 268 if (is_utf8)
156 { 269 {
157 //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
158 uch = decode_utf8 (str, end - str, &clen); 270 uch = decode_utf8 (str, end - str, &clen);
159 if (clen == (STRLEN)-1) 271 if (clen == (STRLEN)-1)
160 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); 272 croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
161 } 273 }
162 else 274 else
163 { 275 {
164 uch = ch; 276 uch = ch;
165 clen = 1; 277 clen = 1;
166 } 278 }
167 279
168 if (uch > 0x10FFFFUL) 280 if (uch < 0x80/*0x20*/ || uch >= enc->limit)
169 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
170
171 if (uch < 0x80 || enc->flags & F_ASCII)
172 { 281 {
173 if (uch > 0xFFFFUL) 282 if (uch >= 0x10000UL)
174 { 283 {
284 if (uch >= 0x110000UL)
285 croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch);
286
175 need (enc, len += 11); 287 need (enc, len += 11);
176 sprintf (enc->cur, "\\u%04x\\u%04x", 288 sprintf (enc->cur, "\\u%04x\\u%04x",
177 (int)((uch - 0x10000) / 0x400 + 0xD800), 289 (int)((uch - 0x10000) / 0x400 + 0xD800),
178 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 290 (int)((uch - 0x10000) % 0x400 + 0xDC00));
179 enc->cur += 12; 291 enc->cur += 12;
180 } 292 }
181 else 293 else
182 { 294 {
183 static char hexdigit [16] = "0123456789abcdef";
184 need (enc, len += 5); 295 need (enc, len += 5);
185 *enc->cur++ = '\\'; 296 *enc->cur++ = '\\';
186 *enc->cur++ = 'u'; 297 *enc->cur++ = 'u';
187 *enc->cur++ = hexdigit [ uch >> 12 ]; 298 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
188 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 299 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
189 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 300 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
190 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 301 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
191 } 302 }
192 303
304 str += clen;
305 }
306 else if (enc->json.flags & F_LATIN1)
307 {
308 *enc->cur++ = uch;
193 str += clen; 309 str += clen;
194 } 310 }
195 else if (is_utf8) 311 else if (is_utf8)
196 { 312 {
197 need (enc, len += clen); 313 need (enc, len += clen);
… …
201 } 317 }
202 while (--clen); 318 while (--clen);
203 } 319 }
204 else 320 else
205 { 321 {
206 need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed 322 need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed
207 enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); 323 enc->cur = encode_utf8 (enc->cur, uch);
208 ++str; 324 ++str;
209 } 325 }
210 } 326 }
211 } 327 }
212 } 328 }
213 329
214 --len; 330 --len;
215 } 331 }
216} 332}
217 333
218static void 334INLINE void
219encode_indent (enc_t *enc) 335encode_indent (enc_t *enc)
220{ 336{
221 if (enc->flags & F_INDENT) 337 if (enc->json.flags & F_INDENT)
222 { 338 {
223 int spaces = enc->indent * INDENT_STEP; 339 int spaces = enc->indent * INDENT_STEP;
224 340
225 need (enc, spaces); 341 need (enc, spaces);
226 memset (enc->cur, ' ', spaces); 342 memset (enc->cur, ' ', spaces);
227 enc->cur += spaces; 343 enc->cur += spaces;
228 } 344 }
229} 345}
230 346
231static void 347INLINE void
232encode_space (enc_t *enc) 348encode_space (enc_t *enc)
233{ 349{
234 need (enc, 1); 350 need (enc, 1);
235 encode_ch (enc, ' '); 351 encode_ch (enc, ' ');
236} 352}
237 353
238static void 354INLINE void
239encode_nl (enc_t *enc) 355encode_nl (enc_t *enc)
240{ 356{
241 if (enc->flags & F_INDENT) 357 if (enc->json.flags & F_INDENT)
242 { 358 {
243 need (enc, 1); 359 need (enc, 1);
244 encode_ch (enc, '\n'); 360 encode_ch (enc, '\n');
245 } 361 }
246} 362}
247 363
248static void 364INLINE void
249encode_comma (enc_t *enc) 365encode_comma (enc_t *enc)
250{ 366{
251 encode_ch (enc, ','); 367 encode_ch (enc, ',');
252 368
253 if (enc->flags & F_INDENT) 369 if (enc->json.flags & F_INDENT)
254 encode_nl (enc); 370 encode_nl (enc);
255 else if (enc->flags & F_SPACE_AFTER) 371 else if (enc->json.flags & F_SPACE_AFTER)
256 encode_space (enc); 372 encode_space (enc);
257} 373}
258 374
259static void encode_sv (enc_t *enc, SV *sv); 375static void encode_sv (enc_t *enc, SV *sv);
260 376
261static void 377static void
262encode_av (enc_t *enc, AV *av) 378encode_av (enc_t *enc, AV *av)
263{ 379{
264 int i, len = av_len (av); 380 int i, len = av_len (av);
265 381
266 encode_ch (enc, '['); encode_nl (enc); 382 if (enc->indent >= enc->json.max_depth)
267 ++enc->indent; 383 croak (ERR_NESTING_EXCEEDED);
268 384
385 encode_ch (enc, '[');
386
387 if (len >= 0)
388 {
389 encode_nl (enc); ++enc->indent;
390
269 for (i = 0; i <= len; ++i) 391 for (i = 0; i <= len; ++i)
270 { 392 {
393 SV **svp = av_fetch (av, i, 0);
394
271 encode_indent (enc); 395 encode_indent (enc);
272 encode_sv (enc, *av_fetch (av, i, 0));
273 396
397 if (svp)
398 encode_sv (enc, *svp);
399 else
400 encode_str (enc, "null", 4, 0);
401
274 if (i < len) 402 if (i < len)
275 encode_comma (enc); 403 encode_comma (enc);
276 } 404 }
277 405
406 encode_nl (enc); --enc->indent; encode_indent (enc);
407 }
408
278 encode_nl (enc); 409 encode_ch (enc, ']');
279
280 --enc->indent;
281 encode_indent (enc); encode_ch (enc, ']');
282} 410}
283 411
284static void 412static void
285encode_he (enc_t *enc, HE *he) 413encode_hk (enc_t *enc, HE *he)
286{ 414{
287 encode_ch (enc, '"'); 415 encode_ch (enc, '"');
288 416
289 if (HeKLEN (he) == HEf_SVKEY) 417 if (HeKLEN (he) == HEf_SVKEY)
290 { 418 {
… …
300 else 428 else
301 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); 429 encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he));
302 430
303 encode_ch (enc, '"'); 431 encode_ch (enc, '"');
304 432
305 if (enc->flags & F_SPACE_BEFORE) encode_space (enc); 433 if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc);
306 encode_ch (enc, ':'); 434 encode_ch (enc, ':');
307 if (enc->flags & F_SPACE_AFTER ) encode_space (enc); 435 if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc);
308 encode_sv (enc, HeVAL (he));
309} 436}
310 437
311// compare hash entries, used when all keys are bytestrings 438// compare hash entries, used when all keys are bytestrings
312static int 439static int
313he_cmp_fast (const void *a_, const void *b_) 440he_cmp_fast (const void *a_, const void *b_)
… …
318 HE *b = *(HE **)b_; 445 HE *b = *(HE **)b_;
319 446
320 STRLEN la = HeKLEN (a); 447 STRLEN la = HeKLEN (a);
321 STRLEN lb = HeKLEN (b); 448 STRLEN lb = HeKLEN (b);
322 449
323 if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) 450 if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la)))
324 cmp = la - lb; 451 cmp = lb - la;
325 452
326 return cmp; 453 return cmp;
327} 454}
328 455
329// compare hash entries, used when some keys are sv's or utf-x 456// compare hash entries, used when some keys are sv's or utf-x
330static int 457static int
331he_cmp_slow (const void *a, const void *b) 458he_cmp_slow (const void *a, const void *b)
332{ 459{
333 return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); 460 return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a));
334} 461}
335 462
336static void 463static void
337encode_hv (enc_t *enc, HV *hv) 464encode_hv (enc_t *enc, HV *hv)
338{ 465{
339 int count, i; 466 HE *he;
340 467
341 encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; 468 if (enc->indent >= enc->json.max_depth)
469 croak (ERR_NESTING_EXCEEDED);
342 470
343 if ((count = hv_iterinit (hv))) 471 encode_ch (enc, '{');
344 { 472
345 // for canonical output we have to sort by keys first 473 // for canonical output we have to sort by keys first
346 // actually, this is mostly due to the stupid so-called 474 // actually, this is mostly due to the stupid so-called
347 // security workaround added somewhere in 5.8.x. 475 // security workaround added somewhere in 5.8.x
348 // that randomises hash orderings 476 // that randomises hash orderings
349 if (enc->flags & F_CANONICAL) 477 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
478 {
479 int count = hv_iterinit (hv);
480
481 if (SvMAGICAL (hv))
350 { 482 {
483 // need to count by iterating. could improve by dynamically building the vector below
484 // but I don't care for the speed of this special case.
485 // note also that we will run into undefined behaviour when the two iterations
486 // do not result in the same count, something I might care for in some later release.
487
488 count = 0;
489 while (hv_iternext (hv))
490 ++count;
491
492 hv_iterinit (hv);
493 }
494
495 if (count)
496 {
497 int i, fast = 1;
498#if defined(__BORLANDC__) || defined(_MSC_VER)
499 HE **hes = _alloca (count * sizeof (HE));
500#else
351 HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode 501 HE *hes [count]; // if your compiler dies here, you need to enable C99 mode
352 int fast = 1; 502#endif
353 503
354 i = 0; 504 i = 0;
355 while ((he = hv_iternext (hv))) 505 while ((he = hv_iternext (hv)))
356 { 506 {
357 hes [i++] = he; 507 hes [i++] = he;
… …
379 529
380 FREETMPS; 530 FREETMPS;
381 LEAVE; 531 LEAVE;
382 } 532 }
383 533
384 for (i = 0; i < count; ++i) 534 encode_nl (enc); ++enc->indent;
535
536 while (count--)
385 { 537 {
386 encode_indent (enc); 538 encode_indent (enc);
539 he = hes [count];
387 encode_he (enc, hes [i]); 540 encode_hk (enc, he);
541 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
388 542
389 if (i < count - 1) 543 if (count)
390 encode_comma (enc); 544 encode_comma (enc);
391 } 545 }
392 546
547 encode_nl (enc); --enc->indent; encode_indent (enc);
548 }
549 }
550 else
551 {
552 if (hv_iterinit (hv) || SvMAGICAL (hv))
553 if ((he = hv_iternext (hv)))
554 {
555 encode_nl (enc); ++enc->indent;
556
557 for (;;)
558 {
393 encode_nl (enc); 559 encode_indent (enc);
560 encode_hk (enc, he);
561 encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he));
562
563 if (!(he = hv_iternext (hv)))
564 break;
565
566 encode_comma (enc);
567 }
568
569 encode_nl (enc); --enc->indent; encode_indent (enc);
570 }
571 }
572
573 encode_ch (enc, '}');
574}
575
576// encode objects, arrays and special \0=false and \1=true values.
577static void
578encode_rv (enc_t *enc, SV *sv)
579{
580 svtype svt;
581
582 SvGETMAGIC (sv);
583 svt = SvTYPE (sv);
584
585 if (expect_false (SvOBJECT (sv)))
586 {
587 HV *stash = !JSON_SLOW || json_boolean_stash
588 ? json_boolean_stash
589 : gv_stashpv ("JSON::XS::Boolean", 1);
590
591 if (SvSTASH (sv) == stash)
592 {
593 if (SvIV (sv))
594 encode_str (enc, "true", 4, 0);
595 else
596 encode_str (enc, "false", 5, 0);
394 } 597 }
395 else 598 else
396 { 599 {
397 SV *sv; 600#if 0
398 HE *he = hv_iternext (hv); 601 if (0 && sv_derived_from (rv, "JSON::Literal"))
399
400 for (;;)
401 { 602 {
402 encode_indent (enc); 603 // not yet
403 encode_he (enc, he);
404
405 if (!(he = hv_iternext (hv)))
406 break;
407
408 encode_comma (enc);
409 } 604 }
605#endif
606 if (enc->json.flags & F_CONV_BLESSED)
607 {
608 // we re-bless the reference to get overload and other niceties right
609 GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0);
410 610
611 if (to_json)
612 {
613 dSP;
614
615 ENTER; SAVETMPS; PUSHMARK (SP);
616 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv)));
617
618 // calling with G_SCALAR ensures that we always get a 1 return value
619 PUTBACK;
620 call_sv ((SV *)GvCV (to_json), G_SCALAR);
621 SPAGAIN;
622
623 // catch this surprisingly common error
624 if (SvROK (TOPs) && SvRV (TOPs) == sv)
625 croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv)));
626
627 sv = POPs;
628 PUTBACK;
629
411 encode_nl (enc); 630 encode_sv (enc, sv);
631
632 FREETMPS; LEAVE;
633 }
634 else if (enc->json.flags & F_ALLOW_BLESSED)
635 encode_str (enc, "null", 4, 0);
636 else
637 croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it",
638 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
639 }
640 else if (enc->json.flags & F_ALLOW_BLESSED)
641 encode_str (enc, "null", 4, 0);
642 else
643 croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled",
644 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
412 } 645 }
413 } 646 }
647 else if (svt == SVt_PVHV)
648 encode_hv (enc, (HV *)sv);
649 else if (svt == SVt_PVAV)
650 encode_av (enc, (AV *)sv);
651 else if (svt < SVt_PVAV)
652 {
653 STRLEN len = 0;
654 char *pv = svt ? SvPV (sv, len) : 0;
414 655
415 --enc->indent; encode_indent (enc); encode_ch (enc, '}'); 656 if (len == 1 && *pv == '1')
657 encode_str (enc, "true", 4, 0);
658 else if (len == 1 && *pv == '0')
659 encode_str (enc, "false", 5, 0);
660 else if (enc->json.flags & F_ALLOW_UNKNOWN)
661 encode_str (enc, "null", 4, 0);
662 else
663 croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1",
664 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
665 }
666 else if (enc->json.flags & F_ALLOW_UNKNOWN)
667 encode_str (enc, "null", 4, 0);
668 else
669 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
670 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
416} 671}
417 672
418static void 673static void
419encode_sv (enc_t *enc, SV *sv) 674encode_sv (enc_t *enc, SV *sv)
420{ 675{
… …
428 encode_str (enc, str, len, SvUTF8 (sv)); 683 encode_str (enc, str, len, SvUTF8 (sv));
429 encode_ch (enc, '"'); 684 encode_ch (enc, '"');
430 } 685 }
431 else if (SvNOKp (sv)) 686 else if (SvNOKp (sv))
432 { 687 {
688 // trust that perl will do the right thing w.r.t. JSON syntax.
433 need (enc, NV_DIG + 32); 689 need (enc, NV_DIG + 32);
434 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); 690 Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur);
435 enc->cur += strlen (enc->cur); 691 enc->cur += strlen (enc->cur);
436 } 692 }
437 else if (SvIOKp (sv)) 693 else if (SvIOKp (sv))
438 { 694 {
695 // we assume we can always read an IV as a UV and vice versa
696 // we assume two's complement
697 // we assume no aliasing issues in the union
698 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
699 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
700 {
701 // optimise the "small number case"
702 // code will likely be branchless and use only a single multiplication
703 // works for numbers up to 59074
704 I32 i = SvIVX (sv);
705 U32 u;
706 char digit, nz = 0;
707
439 need (enc, 64); 708 need (enc, 6);
709
710 *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0;
711 u = i < 0 ? -i : i;
712
713 // convert to 4.28 fixed-point representation
714 u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits
715
716 // now output digit by digit, each time masking out the integer part
717 // and multiplying by 5 while moving the decimal point one to the right,
718 // resulting in a net multiplication by 10.
719 // we always write the digit to memory but conditionally increment
720 // the pointer, to enable the use of conditional move instructions.
721 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
722 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
723 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
724 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
725 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
726 }
727 else
728 {
729 // large integer, use the (rather slow) snprintf way.
730 need (enc, IVUV_MAXCHARS);
440 enc->cur += 731 enc->cur +=
441 SvIsUV(sv) 732 SvIsUV(sv)
442 ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) 733 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
443 : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); 734 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
735 }
444 } 736 }
445 else if (SvROK (sv)) 737 else if (SvROK (sv))
446 { 738 encode_rv (enc, SvRV (sv));
447 SV *rv = SvRV (sv); 739 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
448
449 if (enc->indent >= enc->max_depth)
450 croak ("data structure too deep (hit recursion limit)");
451
452 switch (SvTYPE (rv))
453 {
454 case SVt_PVAV: encode_av (enc, (AV *)rv); break;
455 case SVt_PVHV: encode_hv (enc, (HV *)rv); break;
456
457 default:
458 croak ("encountered %s, but JSON can only represent references to arrays or hashes",
459 SvPV_nolen (sv));
460 }
461 }
462 else if (!SvOK (sv))
463 encode_str (enc, "null", 4, 0); 740 encode_str (enc, "null", 4, 0);
464 else 741 else
465 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 742 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
466 SvPV_nolen (sv), SvFLAGS (sv)); 743 SvPV_nolen (sv), SvFLAGS (sv));
467} 744}
468 745
469static SV * 746static SV *
470encode_json (SV *scalar, UV flags) 747encode_json (SV *scalar, JSON *json)
471{ 748{
749 enc_t enc;
750
472 if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) 751 if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar))
473 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); 752 croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)");
474 753
475 enc_t enc; 754 enc.json = *json;
476 enc.flags = flags;
477 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 755 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
478 enc.cur = SvPVX (enc.sv); 756 enc.cur = SvPVX (enc.sv);
479 enc.end = SvEND (enc.sv); 757 enc.end = SvEND (enc.sv);
480 enc.indent = 0; 758 enc.indent = 0;
481 enc.max_depth = 0x7fffffffUL; 759 enc.limit = enc.json.flags & F_ASCII ? 0x000080UL
760 : enc.json.flags & F_LATIN1 ? 0x000100UL
761 : 0x110000UL;
482 762
483 SvPOK_only (enc.sv); 763 SvPOK_only (enc.sv);
484 encode_sv (&enc, scalar); 764 encode_sv (&enc, scalar);
765 encode_nl (&enc);
485 766
767 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
768 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
769
486 if (!(flags & (F_ASCII | F_UTF8))) 770 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
487 SvUTF8_on (enc.sv); 771 SvUTF8_on (enc.sv);
488 772
489 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
490
491 if (enc.flags & F_SHRINK) 773 if (enc.json.flags & F_SHRINK)
492 shrink (enc.sv); 774 shrink (enc.sv);
493 775
494 return enc.sv; 776 return enc.sv;
495} 777}
496 778
… …
501typedef struct 783typedef struct
502{ 784{
503 char *cur; // current parser pointer 785 char *cur; // current parser pointer
504 char *end; // end of input string 786 char *end; // end of input string
505 const char *err; // parse error, if != 0 787 const char *err; // parse error, if != 0
506 UV flags; // F_* 788 JSON json;
789 U32 depth; // recursion depth
790 U32 maxdepth; // recursion depth limit
507} dec_t; 791} dec_t;
508 792
509static void 793INLINE void
794decode_comment (dec_t *dec)
795{
796 // only '#'-style comments allowed a.t.m.
797
798 while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d)
799 ++dec->cur;
800}
801
802INLINE void
510decode_ws (dec_t *dec) 803decode_ws (dec_t *dec)
511{ 804{
512 for (;;) 805 for (;;)
513 { 806 {
514 char ch = *dec->cur; 807 char ch = *dec->cur;
515 808
516 if (ch > 0x20 809 if (ch > 0x20)
810 {
811 if (expect_false (ch == '#'))
812 {
813 if (dec->json.flags & F_RELAXED)
814 decode_comment (dec);
815 else
816 break;
817 }
818 else
819 break;
820 }
517 || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) 821 else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)
518 break; 822 break; // parse error, but let higher level handle it, gives better error messages
519 823
520 ++dec->cur; 824 ++dec->cur;
521 } 825 }
522} 826}
523 827
524#define ERR(reason) SB dec->err = reason; goto fail; SE 828#define ERR(reason) SB dec->err = reason; goto fail; SE
829
525#define EXPECT_CH(ch) SB \ 830#define EXPECT_CH(ch) SB \
526 if (*dec->cur != ch) \ 831 if (*dec->cur != ch) \
527 ERR (# ch " expected"); \ 832 ERR (# ch " expected"); \
528 ++dec->cur; \ 833 ++dec->cur; \
529 SE 834 SE
530 835
836#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED)
837#define DEC_DEC_DEPTH --dec->depth
838
531static SV *decode_sv (dec_t *dec); 839static SV *decode_sv (dec_t *dec);
532 840
533static signed char decode_hexdigit[256]; 841static signed char decode_hexdigit[256];
534 842
535static UV 843static UV
536decode_4hex (dec_t *dec) 844decode_4hex (dec_t *dec)
537{ 845{
538 signed char d1, d2, d3, d4; 846 signed char d1, d2, d3, d4;
539 unsigned char *cur = (unsigned char *)dec->cur; 847 unsigned char *cur = (unsigned char *)dec->cur;
540 848
541 d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); 849 d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected");
542 d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); 850 d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected");
543 d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); 851 d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected");
544 d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); 852 d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected");
545 853
546 dec->cur += 4; 854 dec->cur += 4;
547 855
548 return ((UV)d1) << 12 856 return ((UV)d1) << 12
549 | ((UV)d2) << 8 857 | ((UV)d2) << 8
… …
557static SV * 865static SV *
558decode_str (dec_t *dec) 866decode_str (dec_t *dec)
559{ 867{
560 SV *sv = 0; 868 SV *sv = 0;
561 int utf8 = 0; 869 int utf8 = 0;
870 char *dec_cur = dec->cur;
562 871
563 do 872 do
564 { 873 {
565 char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; 874 char buf [SHORT_STRING_LEN + UTF8_MAXBYTES];
566 char *cur = buf; 875 char *cur = buf;
567 876
568 do 877 do
569 { 878 {
570 unsigned char ch = *(unsigned char *)dec->cur++; 879 unsigned char ch = *(unsigned char *)dec_cur++;
571 880
572 if (ch == '"') 881 if (expect_false (ch == '"'))
573 { 882 {
574 --dec->cur; 883 --dec_cur;
575 break; 884 break;
576 } 885 }
577 else if (ch == '\\') 886 else if (expect_false (ch == '\\'))
578 { 887 {
579 switch (*dec->cur) 888 switch (*dec_cur)
580 { 889 {
581 case '\\': 890 case '\\':
582 case '/': 891 case '/':
583 case '"': *cur++ = *dec->cur++; break; 892 case '"': *cur++ = *dec_cur++; break;
584 893
585 case 'b': ++dec->cur; *cur++ = '\010'; break; 894 case 'b': ++dec_cur; *cur++ = '\010'; break;
586 case 't': ++dec->cur; *cur++ = '\011'; break; 895 case 't': ++dec_cur; *cur++ = '\011'; break;
587 case 'n': ++dec->cur; *cur++ = '\012'; break; 896 case 'n': ++dec_cur; *cur++ = '\012'; break;
588 case 'f': ++dec->cur; *cur++ = '\014'; break; 897 case 'f': ++dec_cur; *cur++ = '\014'; break;
589 case 'r': ++dec->cur; *cur++ = '\015'; break; 898 case 'r': ++dec_cur; *cur++ = '\015'; break;
590 899
591 case 'u': 900 case 'u':
592 { 901 {
593 UV lo, hi; 902 UV lo, hi;
594 ++dec->cur; 903 ++dec_cur;
595 904
905 dec->cur = dec_cur;
596 hi = decode_4hex (dec); 906 hi = decode_4hex (dec);
907 dec_cur = dec->cur;
597 if (hi == (UV)-1) 908 if (hi == (UV)-1)
598 goto fail; 909 goto fail;
599 910
600 // possibly a surrogate pair 911 // possibly a surrogate pair
601 if (hi >= 0xd800) 912 if (hi >= 0xd800)
602 if (hi < 0xdc00) 913 if (hi < 0xdc00)
603 { 914 {
604 if (dec->cur [0] != '\\' || dec->cur [1] != 'u') 915 if (dec_cur [0] != '\\' || dec_cur [1] != 'u')
605 ERR ("missing low surrogate character in surrogate pair"); 916 ERR ("missing low surrogate character in surrogate pair");
606 917
607 dec->cur += 2; 918 dec_cur += 2;
608 919
920 dec->cur = dec_cur;
609 lo = decode_4hex (dec); 921 lo = decode_4hex (dec);
922 dec_cur = dec->cur;
610 if (lo == (UV)-1) 923 if (lo == (UV)-1)
611 goto fail; 924 goto fail;
612 925
613 if (lo < 0xdc00 || lo >= 0xe000) 926 if (lo < 0xdc00 || lo >= 0xe000)
614 ERR ("surrogate pair expected"); 927 ERR ("surrogate pair expected");
… …
620 933
621 if (hi >= 0x80) 934 if (hi >= 0x80)
622 { 935 {
623 utf8 = 1; 936 utf8 = 1;
624 937
625 cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); 938 cur = encode_utf8 (cur, hi);
626 } 939 }
627 else 940 else
628 *cur++ = hi; 941 *cur++ = hi;
629 } 942 }
630 break; 943 break;
631 944
632 default: 945 default:
633 --dec->cur; 946 --dec_cur;
634 ERR ("illegal backslash escape sequence in string"); 947 ERR ("illegal backslash escape sequence in string");
635 } 948 }
636 } 949 }
637 else if (ch >= 0x20 && ch <= 0x7f) 950 else if (expect_true (ch >= 0x20 && ch < 0x80))
638 *cur++ = ch; 951 *cur++ = ch;
639 else if (ch >= 0x80) 952 else if (ch >= 0x80)
640 { 953 {
641 --dec->cur;
642
643 STRLEN clen; 954 STRLEN clen;
955
956 --dec_cur;
957
644 UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); 958 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
645 if (clen == (STRLEN)-1) 959 if (clen == (STRLEN)-1)
646 ERR ("malformed UTF-8 character in JSON string"); 960 ERR ("malformed UTF-8 character in JSON string");
647 961
648 do 962 do
649 {
650 *cur++ = *dec->cur++; 963 *cur++ = *dec_cur++;
651 }
652 while (--clen); 964 while (--clen);
653 965
654 utf8 = 1; 966 utf8 = 1;
655 } 967 }
656 else if (!ch)
657 ERR ("unexpected end of string while parsing json string");
658 else 968 else
969 {
970 --dec_cur;
971
972 if (!ch)
973 ERR ("unexpected end of string while parsing JSON string");
974 else
659 ERR ("invalid character encountered"); 975 ERR ("invalid character encountered while parsing JSON string");
660 976 }
661 } 977 }
662 while (cur < buf + SHORT_STRING_LEN); 978 while (cur < buf + SHORT_STRING_LEN);
663 979
980 {
664 STRLEN len = cur - buf; 981 STRLEN len = cur - buf;
665 982
666 if (sv) 983 if (sv)
667 { 984 {
668 SvGROW (sv, SvCUR (sv) + len + 1); 985 STRLEN cur = SvCUR (sv);
986
987 if (SvLEN (sv) <= cur + len)
988 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
989
669 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 990 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
670 SvCUR_set (sv, SvCUR (sv) + len); 991 SvCUR_set (sv, SvCUR (sv) + len);
671 } 992 }
672 else 993 else
673 sv = newSVpvn (buf, len); 994 sv = newSVpvn (buf, len);
674 } 995 }
996 }
675 while (*dec->cur != '"'); 997 while (*dec_cur != '"');
676 998
677 ++dec->cur; 999 ++dec_cur;
678 1000
679 if (sv) 1001 if (sv)
680 { 1002 {
681 SvPOK_only (sv); 1003 SvPOK_only (sv);
682 *SvEND (sv) = 0; 1004 *SvEND (sv) = 0;
… …
685 SvUTF8_on (sv); 1007 SvUTF8_on (sv);
686 } 1008 }
687 else 1009 else
688 sv = newSVpvn ("", 0); 1010 sv = newSVpvn ("", 0);
689 1011
1012 dec->cur = dec_cur;
690 return sv; 1013 return sv;
691 1014
692fail: 1015fail:
1016 dec->cur = dec_cur;
693 return 0; 1017 return 0;
694} 1018}
695 1019
696static SV * 1020static SV *
697decode_num (dec_t *dec) 1021decode_num (dec_t *dec)
… …
755 is_nv = 1; 1079 is_nv = 1;
756 } 1080 }
757 1081
758 if (!is_nv) 1082 if (!is_nv)
759 { 1083 {
760 UV uv; 1084 int len = dec->cur - start;
761 int numtype = grok_number (start, dec->cur - start, &uv); 1085
762 if (numtype & IS_NUMBER_IN_UV) 1086 // special case the rather common 1..5-digit-int case
763 if (numtype & IS_NUMBER_NEG) 1087 if (*start == '-')
1088 switch (len)
764 { 1089 {
765 if (uv < (UV)IV_MIN) 1090 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
766 return newSViv (-(IV)uv); 1091 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1092 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1093 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1094 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
767 } 1095 }
1096 else
1097 switch (len)
1098 {
1099 case 1: return newSViv ( start [0] - '0' * 1);
1100 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1101 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1102 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1103 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1104 }
1105
1106 {
1107 UV uv;
1108 int numtype = grok_number (start, len, &uv);
1109 if (numtype & IS_NUMBER_IN_UV)
1110 if (numtype & IS_NUMBER_NEG)
1111 {
1112 if (uv < (UV)IV_MIN)
1113 return newSViv (-(IV)uv);
1114 }
768 else 1115 else
769 return newSVuv (uv); 1116 return newSVuv (uv);
770 } 1117 }
771 1118
1119 len -= *start == '-' ? 1 : 0;
1120
1121 // does not fit into IV or UV, try NV
1122 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len)
1123 #if defined (LDBL_DIG)
1124 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1125 #endif
1126 )
1127 // fits into NV without loss of precision
1128 return newSVnv (Atof (start));
1129
1130 // everything else fails, convert it to a string
1131 return newSVpvn (start, dec->cur - start);
1132 }
1133
1134 // loss of precision here
772 return newSVnv (Atof (start)); 1135 return newSVnv (Atof (start));
773 1136
774fail: 1137fail:
775 return 0; 1138 return 0;
776} 1139}
… …
778static SV * 1141static SV *
779decode_av (dec_t *dec) 1142decode_av (dec_t *dec)
780{ 1143{
781 AV *av = newAV (); 1144 AV *av = newAV ();
782 1145
1146 DEC_INC_DEPTH;
783 decode_ws (dec); 1147 decode_ws (dec);
1148
784 if (*dec->cur == ']') 1149 if (*dec->cur == ']')
785 ++dec->cur; 1150 ++dec->cur;
786 else 1151 else
787 for (;;) 1152 for (;;)
788 { 1153 {
… …
804 1169
805 if (*dec->cur != ',') 1170 if (*dec->cur != ',')
806 ERR (", or ] expected while parsing array"); 1171 ERR (", or ] expected while parsing array");
807 1172
808 ++dec->cur; 1173 ++dec->cur;
1174
1175 decode_ws (dec);
1176
1177 if (*dec->cur == ']' && dec->json.flags & F_RELAXED)
1178 {
1179 ++dec->cur;
1180 break;
1181 }
809 } 1182 }
810 1183
1184 DEC_DEC_DEPTH;
811 return newRV_noinc ((SV *)av); 1185 return newRV_noinc ((SV *)av);
812 1186
813fail: 1187fail:
814 SvREFCNT_dec (av); 1188 SvREFCNT_dec (av);
1189 DEC_DEC_DEPTH;
815 return 0; 1190 return 0;
816} 1191}
817 1192
818static SV * 1193static SV *
819decode_hv (dec_t *dec) 1194decode_hv (dec_t *dec)
820{ 1195{
1196 SV *sv;
821 HV *hv = newHV (); 1197 HV *hv = newHV ();
822 1198
1199 DEC_INC_DEPTH;
823 decode_ws (dec); 1200 decode_ws (dec);
1201
824 if (*dec->cur == '}') 1202 if (*dec->cur == '}')
825 ++dec->cur; 1203 ++dec->cur;
826 else 1204 else
827 for (;;) 1205 for (;;)
828 { 1206 {
1207 EXPECT_CH ('"');
1208
1209 // heuristic: assume that
1210 // a) decode_str + hv_store_ent are abysmally slow.
1211 // b) most hash keys are short, simple ascii text.
1212 // => try to "fast-match" such strings to avoid
1213 // the overhead of decode_str + hv_store_ent.
1214 {
829 SV *key, *value; 1215 SV *value;
1216 char *p = dec->cur;
1217 char *e = p + 24; // only try up to 24 bytes
830 1218
831 decode_ws (dec); EXPECT_CH ('"'); 1219 for (;;)
832
833 key = decode_str (dec);
834 if (!key)
835 goto fail;
836
837 decode_ws (dec); EXPECT_CH (':');
838
839 value = decode_sv (dec);
840 if (!value)
841 { 1220 {
1221 // the >= 0x80 is false on most architectures
1222 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1223 {
1224 // slow path, back up and use decode_str
1225 SV *key = decode_str (dec);
1226 if (!key)
1227 goto fail;
1228
1229 decode_ws (dec); EXPECT_CH (':');
1230
1231 decode_ws (dec);
1232 value = decode_sv (dec);
1233 if (!value)
1234 {
1235 SvREFCNT_dec (key);
1236 goto fail;
1237 }
1238
1239 hv_store_ent (hv, key, value, 0);
842 SvREFCNT_dec (key); 1240 SvREFCNT_dec (key);
1241
1242 break;
1243 }
1244 else if (*p == '"')
1245 {
1246 // fast path, got a simple key
1247 char *key = dec->cur;
1248 int len = p - key;
1249 dec->cur = p + 1;
1250
1251 decode_ws (dec); EXPECT_CH (':');
1252
1253 decode_ws (dec);
1254 value = decode_sv (dec);
1255 if (!value)
843 goto fail; 1256 goto fail;
1257
1258 hv_store (hv, key, len, value, 0);
1259
1260 break;
1261 }
1262
1263 ++p;
844 } 1264 }
845 1265 }
846 //TODO: optimise
847 hv_store_ent (hv, key, value, 0);
848 1266
849 decode_ws (dec); 1267 decode_ws (dec);
850 1268
851 if (*dec->cur == '}') 1269 if (*dec->cur == '}')
852 { 1270 {
… …
856 1274
857 if (*dec->cur != ',') 1275 if (*dec->cur != ',')
858 ERR (", or } expected while parsing object/hash"); 1276 ERR (", or } expected while parsing object/hash");
859 1277
860 ++dec->cur; 1278 ++dec->cur;
1279
1280 decode_ws (dec);
1281
1282 if (*dec->cur == '}' && dec->json.flags & F_RELAXED)
1283 {
1284 ++dec->cur;
1285 break;
1286 }
861 } 1287 }
862 1288
1289 DEC_DEC_DEPTH;
863 return newRV_noinc ((SV *)hv); 1290 sv = newRV_noinc ((SV *)hv);
1291
1292 // check filter callbacks
1293 if (dec->json.flags & F_HOOK)
1294 {
1295 if (dec->json.cb_sk_object && HvKEYS (hv) == 1)
1296 {
1297 HE *cb, *he;
1298
1299 hv_iterinit (hv);
1300 he = hv_iternext (hv);
1301 hv_iterinit (hv);
1302
1303 // the next line creates a mortal sv each time its called.
1304 // might want to optimise this for common cases.
1305 cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0);
1306
1307 if (cb)
1308 {
1309 dSP;
1310 int count;
1311
1312 ENTER; SAVETMPS; PUSHMARK (SP);
1313 XPUSHs (HeVAL (he));
1314
1315 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1316
1317 if (count == 1)
1318 {
1319 sv = newSVsv (POPs);
1320 FREETMPS; LEAVE;
1321 return sv;
1322 }
1323
1324 FREETMPS; LEAVE;
1325 }
1326 }
1327
1328 if (dec->json.cb_object)
1329 {
1330 dSP;
1331 int count;
1332
1333 ENTER; SAVETMPS; PUSHMARK (SP);
1334 XPUSHs (sv_2mortal (sv));
1335
1336 PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN;
1337
1338 if (count == 1)
1339 {
1340 sv = newSVsv (POPs);
1341 FREETMPS; LEAVE;
1342 return sv;
1343 }
1344
1345 SvREFCNT_inc (sv);
1346 FREETMPS; LEAVE;
1347 }
1348 }
1349
1350 return sv;
864 1351
865fail: 1352fail:
866 SvREFCNT_dec (hv); 1353 SvREFCNT_dec (hv);
1354 DEC_DEC_DEPTH;
867 return 0; 1355 return 0;
868} 1356}
869 1357
870static SV * 1358static SV *
871decode_sv (dec_t *dec) 1359decode_sv (dec_t *dec)
872{ 1360{
873 decode_ws (dec); 1361 // the beauty of JSON: you need exactly one character lookahead
1362 // to parse everything.
874 switch (*dec->cur) 1363 switch (*dec->cur)
875 { 1364 {
876 case '"': ++dec->cur; return decode_str (dec); 1365 case '"': ++dec->cur; return decode_str (dec);
877 case '[': ++dec->cur; return decode_av (dec); 1366 case '[': ++dec->cur; return decode_av (dec);
878 case '{': ++dec->cur; return decode_hv (dec); 1367 case '{': ++dec->cur; return decode_hv (dec);
879 1368
880 case '-': 1369 case '-':
881 case '0': case '1': case '2': case '3': case '4': 1370 case '0': case '1': case '2': case '3': case '4':
882 case '5': case '6': case '7': case '8': case '9': 1371 case '5': case '6': case '7': case '8': case '9':
883 return decode_num (dec); 1372 return decode_num (dec);
884 1373
885 case 't': 1374 case 't':
886 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1375 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
887 { 1376 {
888 dec->cur += 4; 1377 dec->cur += 4;
1378#if JSON_SLOW
1379 json_true = get_bool ("JSON::XS::true");
1380#endif
889 return newSViv (1); 1381 return newSVsv (json_true);
890 } 1382 }
891 else 1383 else
892 ERR ("'true' expected"); 1384 ERR ("'true' expected");
893 1385
894 break; 1386 break;
895 1387
896 case 'f': 1388 case 'f':
897 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1389 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
898 { 1390 {
899 dec->cur += 5; 1391 dec->cur += 5;
1392#if JSON_SLOW
1393 json_false = get_bool ("JSON::XS::false");
1394#endif
900 return newSViv (0); 1395 return newSVsv (json_false);
901 } 1396 }
902 else 1397 else
903 ERR ("'false' expected"); 1398 ERR ("'false' expected");
904 1399
905 break; 1400 break;
… …
914 ERR ("'null' expected"); 1409 ERR ("'null' expected");
915 1410
916 break; 1411 break;
917 1412
918 default: 1413 default:
919 ERR ("malformed json string, neither array, object, number, string or atom"); 1414 ERR ("malformed JSON string, neither array, object, number, string or atom");
920 break; 1415 break;
921 } 1416 }
922 1417
923fail: 1418fail:
924 return 0; 1419 return 0;
925} 1420}
926 1421
927static SV * 1422static SV *
928decode_json (SV *string, UV flags) 1423decode_json (SV *string, JSON *json, char **offset_return)
929{ 1424{
1425 dec_t dec;
930 SV *sv; 1426 SV *sv;
931 1427
1428 /* work around bugs in 5.10 where manipulating magic values
1429 * will perl ignore the magic in subsequent accesses
1430 */
1431 /*SvGETMAGIC (string);*/
1432 if (SvMAGICAL (string))
1433 string = sv_2mortal (newSVsv (string));
1434
1435 SvUPGRADE (string, SVt_PV);
1436
1437 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1438 * assertion with -DDEBUGGING, although SvCUR is documented to
1439 * return the xpv_cur field which certainly exists after upgrading.
1440 * according to nicholas clark, calling SvPOK fixes this.
1441 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1442 * and hope for the best.
1443 * Damnit, SvPV_nolen still trips over yet another assertion. This
1444 * assertion business is seriously broken, try yet another workaround
1445 * for the broken -DDEBUGGING.
1446 */
1447 {
1448#ifdef DEBUGGING
1449 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1450#else
1451 STRLEN offset = SvCUR (string);
1452#endif
1453
1454 if (offset > json->max_size && json->max_size)
1455 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1456 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1457 }
1458
932 if (flags & F_UTF8) 1459 if (json->flags & F_UTF8)
933 sv_utf8_downgrade (string, 0); 1460 sv_utf8_downgrade (string, 0);
934 else 1461 else
935 sv_utf8_upgrade (string); 1462 sv_utf8_upgrade (string);
936 1463
937 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1464 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
938 1465
939 dec_t dec; 1466 dec.json = *json;
940 dec.flags = flags;
941 dec.cur = SvPVX (string); 1467 dec.cur = SvPVX (string);
942 dec.end = SvEND (string); 1468 dec.end = SvEND (string);
943 dec.err = 0; 1469 dec.err = 0;
1470 dec.depth = 0;
944 1471
1472 if (dec.json.cb_object || dec.json.cb_sk_object)
1473 dec.json.flags |= F_HOOK;
1474
1475 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1476
1477 decode_ws (&dec);
945 sv = decode_sv (&dec); 1478 sv = decode_sv (&dec);
946 1479
1480 if (offset_return)
1481 *offset_return = dec.cur;
1482
1483 if (!(offset_return || !sv))
1484 {
1485 // check for trailing garbage
1486 decode_ws (&dec);
1487
1488 if (*dec.cur)
1489 {
1490 dec.err = "garbage after JSON object";
1491 SvREFCNT_dec (sv);
1492 sv = 0;
1493 }
1494 }
1495
947 if (!sv) 1496 if (!sv)
948 { 1497 {
949 IV offset = dec.flags & F_UTF8
950 ? dec.cur - SvPVX (string)
951 : utf8_distance (dec.cur, SvPVX (string));
952 SV *uni = sv_newmortal (); 1498 SV *uni = sv_newmortal ();
953 1499
954 // horrible hack to silence warning inside pv_uni_display 1500 // horrible hack to silence warning inside pv_uni_display
955 COP cop = *PL_curcop; 1501 COP cop = *PL_curcop;
956 cop.cop_warnings = pWARN_NONE; 1502 cop.cop_warnings = pWARN_NONE;
… …
958 SAVEVPTR (PL_curcop); 1504 SAVEVPTR (PL_curcop);
959 PL_curcop = &cop; 1505 PL_curcop = &cop;
960 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1506 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
961 LEAVE; 1507 LEAVE;
962 1508
963 croak ("%s, at character offset %d (%s)", 1509 croak ("%s, at character offset %d (before \"%s\")",
964 dec.err, 1510 dec.err,
965 (int)offset, 1511 ptr_to_index (string, dec.cur),
966 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1512 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
967 } 1513 }
968 1514
969 sv = sv_2mortal (sv); 1515 sv = sv_2mortal (sv);
970 1516
971 if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) 1517 if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv))
972 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); 1518 croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)");
973 1519
974 return sv; 1520 return sv;
1521}
1522
1523/////////////////////////////////////////////////////////////////////////////
1524// incremental parser
1525
1526static void
1527incr_parse (JSON *self)
1528{
1529 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1530
1531 for (;;)
1532 {
1533 //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
1534 switch (self->incr_mode)
1535 {
1536 // only used for intiial whitespace skipping
1537 case INCR_M_WS:
1538 for (;;)
1539 {
1540 if (*p > 0x20)
1541 {
1542 self->incr_mode = INCR_M_JSON;
1543 goto incr_m_json;
1544 }
1545 else if (!*p)
1546 goto interrupt;
1547
1548 ++p;
1549 }
1550
1551 // skip a single char inside a string (for \\-processing)
1552 case INCR_M_BS:
1553 if (!*p)
1554 goto interrupt;
1555
1556 ++p;
1557 self->incr_mode = INCR_M_STR;
1558 goto incr_m_str;
1559
1560 // inside a string
1561 case INCR_M_STR:
1562 incr_m_str:
1563 for (;;)
1564 {
1565 if (*p == '"')
1566 {
1567 ++p;
1568 self->incr_mode = INCR_M_JSON;
1569
1570 if (!self->incr_nest)
1571 goto interrupt;
1572
1573 goto incr_m_json;
1574 }
1575 else if (*p == '\\')
1576 {
1577 ++p; // "virtually" consumes character after \
1578
1579 if (!*p) // if at end of string we have to switch modes
1580 {
1581 self->incr_mode = INCR_M_BS;
1582 goto interrupt;
1583 }
1584 }
1585 else if (!*p)
1586 goto interrupt;
1587
1588 ++p;
1589 }
1590
1591 // after initial ws, outside string
1592 case INCR_M_JSON:
1593 incr_m_json:
1594 for (;;)
1595 {
1596 switch (*p++)
1597 {
1598 case 0:
1599 --p;
1600 goto interrupt;
1601
1602 case 0x09:
1603 case 0x0a:
1604 case 0x0d:
1605 case 0x20:
1606 if (!self->incr_nest)
1607 {
1608 --p; // do not eat the whitespace, let the next round do it
1609 goto interrupt;
1610 }
1611 break;
1612
1613 case '"':
1614 self->incr_mode = INCR_M_STR;
1615 goto incr_m_str;
1616
1617 case '[':
1618 case '{':
1619 if (++self->incr_nest > self->max_depth)
1620 croak (ERR_NESTING_EXCEEDED);
1621 break;
1622
1623 case ']':
1624 case '}':
1625 if (--self->incr_nest <= 0)
1626 goto interrupt;
1627 }
1628 }
1629 }
1630
1631 modechange:
1632 ;
1633 }
1634
1635interrupt:
1636 self->incr_pos = p - SvPVX (self->incr_text);
1637 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
975} 1638}
976 1639
977///////////////////////////////////////////////////////////////////////////// 1640/////////////////////////////////////////////////////////////////////////////
978// XS interface functions 1641// XS interface functions
979 1642
… …
981 1644
982BOOT: 1645BOOT:
983{ 1646{
984 int i; 1647 int i;
985 1648
986 memset (decode_hexdigit, 0xff, 256);
987 for (i = 10; i--; ) 1649 for (i = 0; i < 256; ++i)
988 decode_hexdigit ['0' + i] = i; 1650 decode_hexdigit [i] =
1651 i >= '0' && i <= '9' ? i - '0'
1652 : i >= 'a' && i <= 'f' ? i - 'a' + 10
1653 : i >= 'A' && i <= 'F' ? i - 'A' + 10
1654 : -1;
989 1655
990 for (i = 7; i--; )
991 {
992 decode_hexdigit ['a' + i] = 10 + i;
993 decode_hexdigit ['A' + i] = 10 + i;
994 }
995
996 json_stash = gv_stashpv ("JSON::XS", 1); 1656 json_stash = gv_stashpv ("JSON::XS" , 1);
1657 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1658
1659 json_true = get_bool ("JSON::XS::true");
1660 json_false = get_bool ("JSON::XS::false");
1661
1662 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
997} 1663}
998 1664
999PROTOTYPES: DISABLE 1665PROTOTYPES: DISABLE
1000 1666
1001SV *new (char *dummy) 1667void CLONE (...)
1002 CODE: 1668 CODE:
1003 RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); 1669 json_stash = 0;
1670 json_boolean_stash = 0;
1671
1672void new (char *klass)
1673 PPCODE:
1674{
1675 SV *pv = NEWSV (0, sizeof (JSON));
1676 SvPOK_only (pv);
1677 json_init ((JSON *)SvPVX (pv));
1678 XPUSHs (sv_2mortal (sv_bless (
1679 newRV_noinc (pv),
1680 strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1)
1681 )));
1682}
1683
1684void ascii (JSON *self, int enable = 1)
1685 ALIAS:
1686 ascii = F_ASCII
1687 latin1 = F_LATIN1
1688 utf8 = F_UTF8
1689 indent = F_INDENT
1690 canonical = F_CANONICAL
1691 space_before = F_SPACE_BEFORE
1692 space_after = F_SPACE_AFTER
1693 pretty = F_PRETTY
1694 allow_nonref = F_ALLOW_NONREF
1695 shrink = F_SHRINK
1696 allow_blessed = F_ALLOW_BLESSED
1697 convert_blessed = F_CONV_BLESSED
1698 relaxed = F_RELAXED
1699 allow_unknown = F_ALLOW_UNKNOWN
1700 PPCODE:
1701{
1702 if (enable)
1703 self->flags |= ix;
1704 else
1705 self->flags &= ~ix;
1706
1707 XPUSHs (ST (0));
1708}
1709
1710void get_ascii (JSON *self)
1711 ALIAS:
1712 get_ascii = F_ASCII
1713 get_latin1 = F_LATIN1
1714 get_utf8 = F_UTF8
1715 get_indent = F_INDENT
1716 get_canonical = F_CANONICAL
1717 get_space_before = F_SPACE_BEFORE
1718 get_space_after = F_SPACE_AFTER
1719 get_allow_nonref = F_ALLOW_NONREF
1720 get_shrink = F_SHRINK
1721 get_allow_blessed = F_ALLOW_BLESSED
1722 get_convert_blessed = F_CONV_BLESSED
1723 get_relaxed = F_RELAXED
1724 get_allow_unknown = F_ALLOW_UNKNOWN
1725 PPCODE:
1726 XPUSHs (boolSV (self->flags & ix));
1727
1728void max_depth (JSON *self, U32 max_depth = 0x80000000UL)
1729 PPCODE:
1730 self->max_depth = max_depth;
1731 XPUSHs (ST (0));
1732
1733U32 get_max_depth (JSON *self)
1734 CODE:
1735 RETVAL = self->max_depth;
1004 OUTPUT: 1736 OUTPUT:
1005 RETVAL 1737 RETVAL
1006 1738
1007SV *ascii (SV *self, int enable = 1) 1739void max_size (JSON *self, U32 max_size = 0)
1008 ALIAS: 1740 PPCODE:
1009 ascii = F_ASCII 1741 self->max_size = max_size;
1010 utf8 = F_UTF8 1742 XPUSHs (ST (0));
1011 indent = F_INDENT 1743
1012 canonical = F_CANONICAL 1744int get_max_size (JSON *self)
1013 space_before = F_SPACE_BEFORE
1014 space_after = F_SPACE_AFTER
1015 pretty = F_PRETTY
1016 allow_nonref = F_ALLOW_NONREF
1017 shrink = F_SHRINK
1018 CODE: 1745 CODE:
1019{ 1746 RETVAL = self->max_size;
1020 UV *uv = SvJSON (self);
1021 if (enable)
1022 *uv |= ix;
1023 else
1024 *uv &= ~ix;
1025
1026 RETVAL = newSVsv (self);
1027}
1028 OUTPUT: 1747 OUTPUT:
1029 RETVAL 1748 RETVAL
1030 1749
1031void encode (SV *self, SV *scalar) 1750void filter_json_object (JSON *self, SV *cb = &PL_sv_undef)
1032 PPCODE: 1751 PPCODE:
1033 XPUSHs (encode_json (scalar, *SvJSON (self))); 1752{
1753 SvREFCNT_dec (self->cb_object);
1754 self->cb_object = SvOK (cb) ? newSVsv (cb) : 0;
1034 1755
1035void decode (SV *self, SV *jsonstr) 1756 XPUSHs (ST (0));
1757}
1758
1759void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef)
1036 PPCODE: 1760 PPCODE:
1761{
1762 if (!self->cb_sk_object)
1763 self->cb_sk_object = newHV ();
1764
1765 if (SvOK (cb))
1766 hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0);
1767 else
1768 {
1769 hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0);
1770
1771 if (!HvKEYS (self->cb_sk_object))
1772 {
1773 SvREFCNT_dec (self->cb_sk_object);
1774 self->cb_sk_object = 0;
1775 }
1776 }
1777
1778 XPUSHs (ST (0));
1779}
1780
1781void encode (JSON *self, SV *scalar)
1782 PPCODE:
1783 XPUSHs (encode_json (scalar, self));
1784
1785void decode (JSON *self, SV *jsonstr)
1786 PPCODE:
1037 XPUSHs (decode_json (jsonstr, *SvJSON (self))); 1787 XPUSHs (decode_json (jsonstr, self, 0));
1788
1789void decode_prefix (JSON *self, SV *jsonstr)
1790 PPCODE:
1791{
1792 char *offset;
1793 EXTEND (SP, 2);
1794 PUSHs (decode_json (jsonstr, self, &offset));
1795 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1796}
1797
1798void incr_parse (JSON *self, SV *jsonstr = 0)
1799 PPCODE:
1800{
1801 if (!self->incr_text)
1802 self->incr_text = newSVpvn ("", 0);
1803
1804 // append data, if any
1805 if (jsonstr)
1806 {
1807 if (SvUTF8 (jsonstr))
1808 {
1809 if (!SvUTF8 (self->incr_text))
1810 {
1811 /* utf-8-ness differs, need to upgrade */
1812 sv_utf8_upgrade (self->incr_text);
1813
1814 if (self->incr_pos)
1815 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1816 - (U8 *)SvPVX (self->incr_text);
1817 }
1818 }
1819 else if (SvUTF8 (self->incr_text))
1820 sv_utf8_upgrade (jsonstr);
1821
1822 {
1823 STRLEN len;
1824 const char *str = SvPV (jsonstr, len);
1825 STRLEN cur = SvCUR (self->incr_text);
1826
1827 if (SvLEN (self->incr_text) <= cur + len)
1828 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1829
1830 Move (str, SvEND (self->incr_text), len, char);
1831 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1832 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1833 }
1834 }
1835
1836 if (GIMME_V != G_VOID)
1837 do
1838 {
1839 char *offset;
1840
1841 if (!INCR_DONE (self))
1842 {
1843 incr_parse (self);
1844
1845 if (self->incr_pos > self->max_size && self->max_size)
1846 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1847 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1848
1849 if (!INCR_DONE (self))
1850 break;
1851 }
1852
1853 XPUSHs (decode_json (self->incr_text, self, &offset));
1854
1855 self->incr_pos -= offset - SvPVX (self->incr_text);
1856 self->incr_nest = 0;
1857 self->incr_mode = 0;
1858
1859 sv_chop (self->incr_text, offset);
1860 }
1861 while (GIMME_V == G_ARRAY);
1862}
1863
1864SV *incr_text (JSON *self)
1865 ATTRS: lvalue
1866 CODE:
1867{
1868 if (self->incr_pos)
1869 croak ("incr_text can not be called when the incremental parser already started parsing");
1870
1871 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1872}
1873 OUTPUT:
1874 RETVAL
1875
1876void incr_skip (JSON *self)
1877 CODE:
1878{
1879 if (self->incr_pos)
1880 {
1881 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1882 self->incr_pos = 0;
1883 self->incr_nest = 0;
1884 self->incr_mode = 0;
1885 }
1886}
1887
1888void incr_reset (JSON *self)
1889 CODE:
1890{
1891 SvREFCNT_dec (self->incr_text);
1892 self->incr_text = 0;
1893 self->incr_pos = 0;
1894 self->incr_nest = 0;
1895 self->incr_mode = 0;
1896}
1897
1898void DESTROY (JSON *self)
1899 CODE:
1900 SvREFCNT_dec (self->cb_sk_object);
1901 SvREFCNT_dec (self->cb_object);
1902 SvREFCNT_dec (self->incr_text);
1038 1903
1039PROTOTYPES: ENABLE 1904PROTOTYPES: ENABLE
1040 1905
1041void to_json (SV *scalar) 1906void encode_json (SV *scalar)
1907 ALIAS:
1908 to_json_ = 0
1909 encode_json = F_UTF8
1042 PPCODE: 1910 PPCODE:
1911{
1912 JSON json;
1913 json_init (&json);
1914 json.flags |= ix;
1043 XPUSHs (encode_json (scalar, F_UTF8)); 1915 XPUSHs (encode_json (scalar, &json));
1916}
1044 1917
1045void from_json (SV *jsonstr) 1918void decode_json (SV *jsonstr)
1919 ALIAS:
1920 from_json_ = 0
1921 decode_json = F_UTF8
1046 PPCODE: 1922 PPCODE:
1923{
1924 JSON json;
1925 json_init (&json);
1926 json.flags |= ix;
1047 XPUSHs (decode_json (jsonstr, F_UTF8)); 1927 XPUSHs (decode_json (jsonstr, &json, 0));
1928}
1048 1929
1930

Diff Legend

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