1 |
root |
1.1 |
#include "EXTERN.h" |
2 |
|
|
#include "perl.h" |
3 |
|
|
#include "XSUB.h" |
4 |
|
|
|
5 |
root |
1.56 |
#include <assert.h> |
6 |
|
|
#include <string.h> |
7 |
|
|
#include <stdlib.h> |
8 |
|
|
#include <stdio.h> |
9 |
root |
1.75 |
#include <limits.h> |
10 |
root |
1.56 |
#include <float.h> |
11 |
root |
1.26 |
|
12 |
|
|
#if defined(__BORLANDC__) || defined(_MSC_VER) |
13 |
|
|
# define snprintf _snprintf // C compilers have this in stdio.h |
14 |
|
|
#endif |
15 |
root |
1.1 |
|
16 |
root |
1.41 |
// some old perls do not have this, try to make it work, no |
17 |
root |
1.99 |
// guarantees, though. if it breaks, you get to keep the pieces. |
18 |
root |
1.41 |
#ifndef UTF8_MAXBYTES |
19 |
|
|
# define UTF8_MAXBYTES 13 |
20 |
|
|
#endif |
21 |
|
|
|
22 |
root |
1.102 |
// three extra for rounding, sign, and end of string |
23 |
|
|
#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3) |
24 |
root |
1.75 |
|
25 |
root |
1.44 |
#define F_ASCII 0x00000001UL |
26 |
|
|
#define F_LATIN1 0x00000002UL |
27 |
|
|
#define F_UTF8 0x00000004UL |
28 |
|
|
#define F_INDENT 0x00000008UL |
29 |
|
|
#define F_CANONICAL 0x00000010UL |
30 |
|
|
#define F_SPACE_BEFORE 0x00000020UL |
31 |
|
|
#define F_SPACE_AFTER 0x00000040UL |
32 |
|
|
#define F_ALLOW_NONREF 0x00000100UL |
33 |
|
|
#define F_SHRINK 0x00000200UL |
34 |
|
|
#define F_ALLOW_BLESSED 0x00000400UL |
35 |
root |
1.49 |
#define F_CONV_BLESSED 0x00000800UL |
36 |
root |
1.63 |
#define F_RELAXED 0x00001000UL |
37 |
root |
1.84 |
#define F_ALLOW_UNKNOWN 0x00002000UL |
38 |
root |
1.49 |
#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
39 |
root |
1.18 |
|
40 |
root |
1.2 |
#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
41 |
root |
1.1 |
|
42 |
|
|
#define INIT_SIZE 32 // initial scalar size to be allocated |
43 |
root |
1.12 |
#define INDENT_STEP 3 // spaces per indentation level |
44 |
|
|
|
45 |
root |
1.38 |
#define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
46 |
root |
1.1 |
|
47 |
|
|
#define SB do { |
48 |
|
|
#define SE } while (0) |
49 |
|
|
|
50 |
root |
1.35 |
#if __GNUC__ >= 3 |
51 |
root |
1.71 |
# define expect(expr,value) __builtin_expect ((expr), (value)) |
52 |
|
|
# define INLINE static inline |
53 |
root |
1.35 |
#else |
54 |
|
|
# define expect(expr,value) (expr) |
55 |
root |
1.71 |
# define INLINE static |
56 |
root |
1.35 |
#endif |
57 |
|
|
|
58 |
|
|
#define expect_false(expr) expect ((expr) != 0, 0) |
59 |
|
|
#define expect_true(expr) expect ((expr) != 0, 1) |
60 |
|
|
|
61 |
root |
1.72 |
#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 |
root |
1.85 |
#define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)" |
66 |
|
|
|
67 |
root |
1.57 |
#ifdef USE_ITHREADS |
68 |
|
|
# define JSON_SLOW 1 |
69 |
root |
1.59 |
# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
70 |
root |
1.57 |
#else |
71 |
root |
1.58 |
# define JSON_SLOW 0 |
72 |
root |
1.59 |
# define JSON_STASH json_stash |
73 |
root |
1.57 |
#endif |
74 |
|
|
|
75 |
root |
1.44 |
static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
76 |
root |
1.43 |
static SV *json_true, *json_false; |
77 |
root |
1.1 |
|
78 |
root |
1.77 |
enum { |
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 |
root |
1.88 |
#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) |
86 |
root |
1.77 |
|
87 |
root |
1.48 |
typedef struct { |
88 |
root |
1.47 |
U32 flags; |
89 |
root |
1.85 |
U32 max_depth; |
90 |
|
|
STRLEN max_size; |
91 |
|
|
|
92 |
root |
1.52 |
SV *cb_object; |
93 |
|
|
HV *cb_sk_object; |
94 |
root |
1.77 |
|
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 |
root |
1.88 |
int incr_nest; // {[]}-nesting level |
99 |
root |
1.85 |
unsigned char incr_mode; |
100 |
root |
1.48 |
} JSON; |
101 |
root |
1.47 |
|
102 |
root |
1.85 |
INLINE void |
103 |
|
|
json_init (JSON *json) |
104 |
|
|
{ |
105 |
|
|
Zero (json, 1, JSON); |
106 |
|
|
json->max_depth = 512; |
107 |
|
|
} |
108 |
|
|
|
109 |
root |
1.12 |
///////////////////////////////////////////////////////////////////////////// |
110 |
|
|
// utility functions |
111 |
root |
1.1 |
|
112 |
root |
1.83 |
INLINE SV * |
113 |
|
|
get_bool (const char *name) |
114 |
|
|
{ |
115 |
|
|
SV *sv = get_sv (name, 1); |
116 |
|
|
|
117 |
|
|
SvREADONLY_on (sv); |
118 |
|
|
SvREADONLY_on (SvRV (sv)); |
119 |
|
|
|
120 |
|
|
return sv; |
121 |
|
|
} |
122 |
|
|
|
123 |
root |
1.71 |
INLINE void |
124 |
root |
1.7 |
shrink (SV *sv) |
125 |
|
|
{ |
126 |
|
|
sv_utf8_downgrade (sv, 1); |
127 |
root |
1.87 |
|
128 |
root |
1.12 |
if (SvLEN (sv) > SvCUR (sv) + 1) |
129 |
|
|
{ |
130 |
root |
1.7 |
#ifdef SvPV_shrink_to_cur |
131 |
root |
1.12 |
SvPV_shrink_to_cur (sv); |
132 |
|
|
#elif defined (SvPV_renew) |
133 |
|
|
SvPV_renew (sv, SvCUR (sv) + 1); |
134 |
root |
1.7 |
#endif |
135 |
root |
1.12 |
} |
136 |
root |
1.7 |
} |
137 |
|
|
|
138 |
root |
1.13 |
// decode an utf-8 character and return it, or (UV)-1 in |
139 |
|
|
// case of an error. |
140 |
|
|
// we special-case "safe" characters from U+80 .. U+7FF, |
141 |
|
|
// but use the very good perl function to parse anything else. |
142 |
|
|
// note that we never call this function for a ascii codepoints |
143 |
root |
1.71 |
INLINE UV |
144 |
root |
1.13 |
decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
145 |
|
|
{ |
146 |
root |
1.72 |
if (expect_true (len >= 2 |
147 |
|
|
&& IN_RANGE_INC (char, s[0], 0xc2, 0xdf) |
148 |
|
|
&& IN_RANGE_INC (char, s[1], 0x80, 0xbf))) |
149 |
root |
1.13 |
{ |
150 |
|
|
*clen = 2; |
151 |
|
|
return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
152 |
|
|
} |
153 |
|
|
else |
154 |
root |
1.72 |
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 |
160 |
|
|
INLINE unsigned char * |
161 |
|
|
encode_utf8 (unsigned char *s, UV ch) |
162 |
|
|
{ |
163 |
root |
1.74 |
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 |
root |
1.72 |
|
178 |
|
|
return s; |
179 |
root |
1.13 |
} |
180 |
|
|
|
181 |
root |
1.96 |
// convert offset pointer to character index, sv must be string |
182 |
|
|
static STRLEN |
183 |
|
|
ptr_to_index (SV *sv, char *offset) |
184 |
|
|
{ |
185 |
|
|
return SvUTF8 (sv) |
186 |
|
|
? utf8_distance (offset, SvPVX (sv)) |
187 |
|
|
: offset - SvPVX (sv); |
188 |
|
|
} |
189 |
|
|
|
190 |
root |
1.1 |
///////////////////////////////////////////////////////////////////////////// |
191 |
root |
1.12 |
// encoder |
192 |
|
|
|
193 |
|
|
// structure used for encoding JSON |
194 |
|
|
typedef struct |
195 |
|
|
{ |
196 |
|
|
char *cur; // SvPVX (sv) + current output position |
197 |
|
|
char *end; // SvEND (sv) |
198 |
|
|
SV *sv; // result scalar |
199 |
root |
1.48 |
JSON json; |
200 |
root |
1.18 |
U32 indent; // indentation level |
201 |
root |
1.71 |
UV limit; // escape character values >= this value when encoding |
202 |
root |
1.12 |
} enc_t; |
203 |
root |
1.1 |
|
204 |
root |
1.71 |
INLINE void |
205 |
root |
1.1 |
need (enc_t *enc, STRLEN len) |
206 |
|
|
{ |
207 |
root |
1.35 |
if (expect_false (enc->cur + len >= enc->end)) |
208 |
root |
1.1 |
{ |
209 |
root |
1.93 |
STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
210 |
root |
1.95 |
SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
211 |
root |
1.1 |
enc->cur = SvPVX (enc->sv) + cur; |
212 |
root |
1.27 |
enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
213 |
root |
1.1 |
} |
214 |
|
|
} |
215 |
|
|
|
216 |
root |
1.71 |
INLINE void |
217 |
root |
1.1 |
encode_ch (enc_t *enc, char ch) |
218 |
|
|
{ |
219 |
|
|
need (enc, 1); |
220 |
|
|
*enc->cur++ = ch; |
221 |
|
|
} |
222 |
|
|
|
223 |
|
|
static void |
224 |
|
|
encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) |
225 |
|
|
{ |
226 |
|
|
char *end = str + len; |
227 |
|
|
|
228 |
root |
1.4 |
need (enc, len); |
229 |
|
|
|
230 |
root |
1.1 |
while (str < end) |
231 |
|
|
{ |
232 |
|
|
unsigned char ch = *(unsigned char *)str; |
233 |
root |
1.4 |
|
234 |
root |
1.35 |
if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
235 |
root |
1.4 |
{ |
236 |
root |
1.35 |
if (expect_false (ch == '"')) // but with slow exceptions |
237 |
root |
1.6 |
{ |
238 |
|
|
need (enc, len += 1); |
239 |
|
|
*enc->cur++ = '\\'; |
240 |
|
|
*enc->cur++ = '"'; |
241 |
|
|
} |
242 |
root |
1.35 |
else if (expect_false (ch == '\\')) |
243 |
root |
1.6 |
{ |
244 |
|
|
need (enc, len += 1); |
245 |
|
|
*enc->cur++ = '\\'; |
246 |
|
|
*enc->cur++ = '\\'; |
247 |
|
|
} |
248 |
|
|
else |
249 |
|
|
*enc->cur++ = ch; |
250 |
|
|
|
251 |
root |
1.4 |
++str; |
252 |
root |
1.1 |
} |
253 |
|
|
else |
254 |
|
|
{ |
255 |
root |
1.6 |
switch (ch) |
256 |
root |
1.1 |
{ |
257 |
root |
1.6 |
case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; |
258 |
|
|
case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; |
259 |
|
|
case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; |
260 |
|
|
case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; |
261 |
|
|
case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; |
262 |
root |
1.1 |
|
263 |
root |
1.6 |
default: |
264 |
root |
1.1 |
{ |
265 |
root |
1.6 |
STRLEN clen; |
266 |
|
|
UV uch; |
267 |
|
|
|
268 |
|
|
if (is_utf8) |
269 |
|
|
{ |
270 |
root |
1.13 |
uch = decode_utf8 (str, end - str, &clen); |
271 |
root |
1.6 |
if (clen == (STRLEN)-1) |
272 |
root |
1.9 |
croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
273 |
root |
1.6 |
} |
274 |
|
|
else |
275 |
|
|
{ |
276 |
|
|
uch = ch; |
277 |
|
|
clen = 1; |
278 |
|
|
} |
279 |
|
|
|
280 |
root |
1.72 |
if (uch < 0x80/*0x20*/ || uch >= enc->limit) |
281 |
root |
1.6 |
{ |
282 |
root |
1.74 |
if (uch >= 0x10000UL) |
283 |
root |
1.6 |
{ |
284 |
root |
1.74 |
if (uch >= 0x110000UL) |
285 |
root |
1.71 |
croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
286 |
|
|
|
287 |
root |
1.6 |
need (enc, len += 11); |
288 |
|
|
sprintf (enc->cur, "\\u%04x\\u%04x", |
289 |
root |
1.10 |
(int)((uch - 0x10000) / 0x400 + 0xD800), |
290 |
|
|
(int)((uch - 0x10000) % 0x400 + 0xDC00)); |
291 |
root |
1.6 |
enc->cur += 12; |
292 |
|
|
} |
293 |
|
|
else |
294 |
|
|
{ |
295 |
|
|
need (enc, len += 5); |
296 |
|
|
*enc->cur++ = '\\'; |
297 |
|
|
*enc->cur++ = 'u'; |
298 |
root |
1.91 |
*enc->cur++ = PL_hexdigit [ uch >> 12 ]; |
299 |
|
|
*enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; |
300 |
|
|
*enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; |
301 |
|
|
*enc->cur++ = PL_hexdigit [(uch >> 0) & 15]; |
302 |
root |
1.6 |
} |
303 |
root |
1.4 |
|
304 |
root |
1.6 |
str += clen; |
305 |
|
|
} |
306 |
root |
1.47 |
else if (enc->json.flags & F_LATIN1) |
307 |
root |
1.30 |
{ |
308 |
|
|
*enc->cur++ = uch; |
309 |
|
|
str += clen; |
310 |
|
|
} |
311 |
root |
1.6 |
else if (is_utf8) |
312 |
|
|
{ |
313 |
|
|
need (enc, len += clen); |
314 |
|
|
do |
315 |
|
|
{ |
316 |
|
|
*enc->cur++ = *str++; |
317 |
|
|
} |
318 |
|
|
while (--clen); |
319 |
|
|
} |
320 |
|
|
else |
321 |
|
|
{ |
322 |
root |
1.28 |
need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
323 |
root |
1.72 |
enc->cur = encode_utf8 (enc->cur, uch); |
324 |
root |
1.6 |
++str; |
325 |
|
|
} |
326 |
root |
1.5 |
} |
327 |
root |
1.4 |
} |
328 |
root |
1.1 |
} |
329 |
|
|
|
330 |
|
|
--len; |
331 |
|
|
} |
332 |
|
|
} |
333 |
|
|
|
334 |
root |
1.71 |
INLINE void |
335 |
root |
1.12 |
encode_indent (enc_t *enc) |
336 |
|
|
{ |
337 |
root |
1.47 |
if (enc->json.flags & F_INDENT) |
338 |
root |
1.12 |
{ |
339 |
|
|
int spaces = enc->indent * INDENT_STEP; |
340 |
|
|
|
341 |
|
|
need (enc, spaces); |
342 |
|
|
memset (enc->cur, ' ', spaces); |
343 |
|
|
enc->cur += spaces; |
344 |
|
|
} |
345 |
|
|
} |
346 |
|
|
|
347 |
root |
1.71 |
INLINE void |
348 |
root |
1.12 |
encode_space (enc_t *enc) |
349 |
|
|
{ |
350 |
|
|
need (enc, 1); |
351 |
|
|
encode_ch (enc, ' '); |
352 |
|
|
} |
353 |
|
|
|
354 |
root |
1.71 |
INLINE void |
355 |
root |
1.12 |
encode_nl (enc_t *enc) |
356 |
|
|
{ |
357 |
root |
1.47 |
if (enc->json.flags & F_INDENT) |
358 |
root |
1.12 |
{ |
359 |
|
|
need (enc, 1); |
360 |
|
|
encode_ch (enc, '\n'); |
361 |
|
|
} |
362 |
|
|
} |
363 |
|
|
|
364 |
root |
1.71 |
INLINE void |
365 |
root |
1.12 |
encode_comma (enc_t *enc) |
366 |
|
|
{ |
367 |
|
|
encode_ch (enc, ','); |
368 |
root |
1.1 |
|
369 |
root |
1.47 |
if (enc->json.flags & F_INDENT) |
370 |
root |
1.12 |
encode_nl (enc); |
371 |
root |
1.47 |
else if (enc->json.flags & F_SPACE_AFTER) |
372 |
root |
1.12 |
encode_space (enc); |
373 |
|
|
} |
374 |
root |
1.1 |
|
375 |
|
|
static void encode_sv (enc_t *enc, SV *sv); |
376 |
|
|
|
377 |
|
|
static void |
378 |
|
|
encode_av (enc_t *enc, AV *av) |
379 |
|
|
{ |
380 |
|
|
int i, len = av_len (av); |
381 |
|
|
|
382 |
root |
1.85 |
if (enc->indent >= enc->json.max_depth) |
383 |
|
|
croak (ERR_NESTING_EXCEEDED); |
384 |
root |
1.21 |
|
385 |
root |
1.65 |
encode_ch (enc, '['); |
386 |
|
|
|
387 |
|
|
if (len >= 0) |
388 |
root |
1.1 |
{ |
389 |
root |
1.65 |
encode_nl (enc); ++enc->indent; |
390 |
root |
1.55 |
|
391 |
root |
1.65 |
for (i = 0; i <= len; ++i) |
392 |
|
|
{ |
393 |
|
|
SV **svp = av_fetch (av, i, 0); |
394 |
root |
1.55 |
|
395 |
root |
1.65 |
encode_indent (enc); |
396 |
root |
1.1 |
|
397 |
root |
1.65 |
if (svp) |
398 |
|
|
encode_sv (enc, *svp); |
399 |
|
|
else |
400 |
|
|
encode_str (enc, "null", 4, 0); |
401 |
root |
1.1 |
|
402 |
root |
1.65 |
if (i < len) |
403 |
|
|
encode_comma (enc); |
404 |
|
|
} |
405 |
root |
1.1 |
|
406 |
root |
1.65 |
encode_nl (enc); --enc->indent; encode_indent (enc); |
407 |
|
|
} |
408 |
|
|
|
409 |
|
|
encode_ch (enc, ']'); |
410 |
root |
1.1 |
} |
411 |
|
|
|
412 |
|
|
static void |
413 |
root |
1.61 |
encode_hk (enc_t *enc, HE *he) |
414 |
root |
1.1 |
{ |
415 |
|
|
encode_ch (enc, '"'); |
416 |
|
|
|
417 |
|
|
if (HeKLEN (he) == HEf_SVKEY) |
418 |
|
|
{ |
419 |
|
|
SV *sv = HeSVKEY (he); |
420 |
|
|
STRLEN len; |
421 |
root |
1.4 |
char *str; |
422 |
|
|
|
423 |
|
|
SvGETMAGIC (sv); |
424 |
|
|
str = SvPV (sv, len); |
425 |
root |
1.1 |
|
426 |
|
|
encode_str (enc, str, len, SvUTF8 (sv)); |
427 |
|
|
} |
428 |
|
|
else |
429 |
|
|
encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
430 |
|
|
|
431 |
|
|
encode_ch (enc, '"'); |
432 |
|
|
|
433 |
root |
1.47 |
if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); |
434 |
root |
1.1 |
encode_ch (enc, ':'); |
435 |
root |
1.47 |
if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); |
436 |
root |
1.1 |
} |
437 |
|
|
|
438 |
|
|
// compare hash entries, used when all keys are bytestrings |
439 |
|
|
static int |
440 |
|
|
he_cmp_fast (const void *a_, const void *b_) |
441 |
|
|
{ |
442 |
|
|
int cmp; |
443 |
|
|
|
444 |
|
|
HE *a = *(HE **)a_; |
445 |
|
|
HE *b = *(HE **)b_; |
446 |
|
|
|
447 |
|
|
STRLEN la = HeKLEN (a); |
448 |
|
|
STRLEN lb = HeKLEN (b); |
449 |
|
|
|
450 |
root |
1.61 |
if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la))) |
451 |
|
|
cmp = lb - la; |
452 |
root |
1.1 |
|
453 |
|
|
return cmp; |
454 |
|
|
} |
455 |
|
|
|
456 |
|
|
// compare hash entries, used when some keys are sv's or utf-x |
457 |
|
|
static int |
458 |
|
|
he_cmp_slow (const void *a, const void *b) |
459 |
|
|
{ |
460 |
root |
1.61 |
return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a)); |
461 |
root |
1.1 |
} |
462 |
|
|
|
463 |
|
|
static void |
464 |
|
|
encode_hv (enc_t *enc, HV *hv) |
465 |
|
|
{ |
466 |
root |
1.61 |
HE *he; |
467 |
root |
1.1 |
|
468 |
root |
1.85 |
if (enc->indent >= enc->json.max_depth) |
469 |
|
|
croak (ERR_NESTING_EXCEEDED); |
470 |
root |
1.21 |
|
471 |
root |
1.65 |
encode_ch (enc, '{'); |
472 |
root |
1.1 |
|
473 |
root |
1.61 |
// for canonical output we have to sort by keys first |
474 |
|
|
// actually, this is mostly due to the stupid so-called |
475 |
root |
1.92 |
// security workaround added somewhere in 5.8.x |
476 |
root |
1.61 |
// that randomises hash orderings |
477 |
root |
1.98 |
if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv)) |
478 |
root |
1.1 |
{ |
479 |
root |
1.61 |
int count = hv_iterinit (hv); |
480 |
|
|
|
481 |
|
|
if (SvMAGICAL (hv)) |
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 |
root |
1.1 |
{ |
497 |
root |
1.61 |
int i, fast = 1; |
498 |
root |
1.34 |
#if defined(__BORLANDC__) || defined(_MSC_VER) |
499 |
root |
1.33 |
HE **hes = _alloca (count * sizeof (HE)); |
500 |
|
|
#else |
501 |
|
|
HE *hes [count]; // if your compiler dies here, you need to enable C99 mode |
502 |
|
|
#endif |
503 |
root |
1.1 |
|
504 |
|
|
i = 0; |
505 |
|
|
while ((he = hv_iternext (hv))) |
506 |
|
|
{ |
507 |
|
|
hes [i++] = he; |
508 |
|
|
if (HeKLEN (he) < 0 || HeKUTF8 (he)) |
509 |
|
|
fast = 0; |
510 |
|
|
} |
511 |
|
|
|
512 |
|
|
assert (i == count); |
513 |
|
|
|
514 |
|
|
if (fast) |
515 |
|
|
qsort (hes, count, sizeof (HE *), he_cmp_fast); |
516 |
|
|
else |
517 |
|
|
{ |
518 |
root |
1.8 |
// hack to forcefully disable "use bytes" |
519 |
|
|
COP cop = *PL_curcop; |
520 |
root |
1.1 |
cop.op_private = 0; |
521 |
root |
1.8 |
|
522 |
|
|
ENTER; |
523 |
|
|
SAVETMPS; |
524 |
|
|
|
525 |
|
|
SAVEVPTR (PL_curcop); |
526 |
root |
1.1 |
PL_curcop = &cop; |
527 |
|
|
|
528 |
|
|
qsort (hes, count, sizeof (HE *), he_cmp_slow); |
529 |
root |
1.8 |
|
530 |
root |
1.1 |
FREETMPS; |
531 |
root |
1.8 |
LEAVE; |
532 |
root |
1.1 |
} |
533 |
|
|
|
534 |
root |
1.65 |
encode_nl (enc); ++enc->indent; |
535 |
|
|
|
536 |
root |
1.61 |
while (count--) |
537 |
root |
1.1 |
{ |
538 |
root |
1.12 |
encode_indent (enc); |
539 |
root |
1.61 |
he = hes [count]; |
540 |
|
|
encode_hk (enc, he); |
541 |
root |
1.62 |
encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
542 |
root |
1.1 |
|
543 |
root |
1.61 |
if (count) |
544 |
root |
1.12 |
encode_comma (enc); |
545 |
root |
1.1 |
} |
546 |
root |
1.65 |
|
547 |
|
|
encode_nl (enc); --enc->indent; encode_indent (enc); |
548 |
root |
1.1 |
} |
549 |
root |
1.61 |
} |
550 |
|
|
else |
551 |
|
|
{ |
552 |
|
|
if (hv_iterinit (hv) || SvMAGICAL (hv)) |
553 |
|
|
if ((he = hv_iternext (hv))) |
554 |
root |
1.65 |
{ |
555 |
|
|
encode_nl (enc); ++enc->indent; |
556 |
|
|
|
557 |
|
|
for (;;) |
558 |
|
|
{ |
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 |
root |
1.1 |
|
566 |
root |
1.65 |
encode_comma (enc); |
567 |
|
|
} |
568 |
root |
1.1 |
|
569 |
root |
1.65 |
encode_nl (enc); --enc->indent; encode_indent (enc); |
570 |
|
|
} |
571 |
root |
1.61 |
} |
572 |
root |
1.1 |
|
573 |
root |
1.65 |
encode_ch (enc, '}'); |
574 |
root |
1.1 |
} |
575 |
|
|
|
576 |
root |
1.21 |
// encode objects, arrays and special \0=false and \1=true values. |
577 |
|
|
static void |
578 |
|
|
encode_rv (enc_t *enc, SV *sv) |
579 |
|
|
{ |
580 |
root |
1.25 |
svtype svt; |
581 |
|
|
|
582 |
root |
1.21 |
SvGETMAGIC (sv); |
583 |
root |
1.25 |
svt = SvTYPE (sv); |
584 |
root |
1.21 |
|
585 |
root |
1.44 |
if (expect_false (SvOBJECT (sv))) |
586 |
|
|
{ |
587 |
root |
1.57 |
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 |
root |
1.44 |
{ |
593 |
root |
1.51 |
if (SvIV (sv)) |
594 |
|
|
encode_str (enc, "true", 4, 0); |
595 |
|
|
else |
596 |
root |
1.44 |
encode_str (enc, "false", 5, 0); |
597 |
|
|
} |
598 |
|
|
else |
599 |
|
|
{ |
600 |
|
|
#if 0 |
601 |
|
|
if (0 && sv_derived_from (rv, "JSON::Literal")) |
602 |
|
|
{ |
603 |
|
|
// not yet |
604 |
|
|
} |
605 |
|
|
#endif |
606 |
root |
1.47 |
if (enc->json.flags & F_CONV_BLESSED) |
607 |
root |
1.44 |
{ |
608 |
|
|
// we re-bless the reference to get overload and other niceties right |
609 |
root |
1.53 |
GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0); |
610 |
root |
1.44 |
|
611 |
|
|
if (to_json) |
612 |
|
|
{ |
613 |
root |
1.53 |
dSP; |
614 |
|
|
|
615 |
|
|
ENTER; SAVETMPS; PUSHMARK (SP); |
616 |
root |
1.44 |
XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); |
617 |
|
|
|
618 |
root |
1.53 |
// calling with G_SCALAR ensures that we always get a 1 return value |
619 |
root |
1.44 |
PUTBACK; |
620 |
root |
1.54 |
call_sv ((SV *)GvCV (to_json), G_SCALAR); |
621 |
root |
1.44 |
SPAGAIN; |
622 |
|
|
|
623 |
root |
1.53 |
// 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 |
|
|
|
630 |
|
|
encode_sv (enc, sv); |
631 |
root |
1.44 |
|
632 |
root |
1.49 |
FREETMPS; LEAVE; |
633 |
root |
1.44 |
} |
634 |
root |
1.47 |
else if (enc->json.flags & F_ALLOW_BLESSED) |
635 |
root |
1.44 |
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 |
root |
1.47 |
else if (enc->json.flags & F_ALLOW_BLESSED) |
641 |
root |
1.44 |
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)))); |
645 |
|
|
} |
646 |
|
|
} |
647 |
|
|
else if (svt == SVt_PVHV) |
648 |
root |
1.21 |
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 |
root |
1.50 |
STRLEN len = 0; |
654 |
|
|
char *pv = svt ? SvPV (sv, len) : 0; |
655 |
|
|
|
656 |
root |
1.51 |
if (len == 1 && *pv == '1') |
657 |
|
|
encode_str (enc, "true", 4, 0); |
658 |
|
|
else if (len == 1 && *pv == '0') |
659 |
root |
1.21 |
encode_str (enc, "false", 5, 0); |
660 |
root |
1.84 |
else if (enc->json.flags & F_ALLOW_UNKNOWN) |
661 |
|
|
encode_str (enc, "null", 4, 0); |
662 |
root |
1.21 |
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 |
root |
1.84 |
else if (enc->json.flags & F_ALLOW_UNKNOWN) |
667 |
|
|
encode_str (enc, "null", 4, 0); |
668 |
root |
1.21 |
else |
669 |
|
|
croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
670 |
|
|
SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
671 |
|
|
} |
672 |
|
|
|
673 |
root |
1.1 |
static void |
674 |
|
|
encode_sv (enc_t *enc, SV *sv) |
675 |
|
|
{ |
676 |
root |
1.4 |
SvGETMAGIC (sv); |
677 |
|
|
|
678 |
root |
1.1 |
if (SvPOKp (sv)) |
679 |
|
|
{ |
680 |
|
|
STRLEN len; |
681 |
|
|
char *str = SvPV (sv, len); |
682 |
|
|
encode_ch (enc, '"'); |
683 |
|
|
encode_str (enc, str, len, SvUTF8 (sv)); |
684 |
|
|
encode_ch (enc, '"'); |
685 |
|
|
} |
686 |
|
|
else if (SvNOKp (sv)) |
687 |
|
|
{ |
688 |
root |
1.39 |
// trust that perl will do the right thing w.r.t. JSON syntax. |
689 |
root |
1.1 |
need (enc, NV_DIG + 32); |
690 |
|
|
Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
691 |
|
|
enc->cur += strlen (enc->cur); |
692 |
|
|
} |
693 |
|
|
else if (SvIOKp (sv)) |
694 |
|
|
{ |
695 |
root |
1.74 |
// 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 |
root |
1.75 |
if (SvIsUV (sv) ? SvUVX (sv) <= 59000 |
699 |
|
|
: SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000) |
700 |
root |
1.35 |
{ |
701 |
|
|
// optimise the "small number case" |
702 |
|
|
// code will likely be branchless and use only a single multiplication |
703 |
root |
1.74 |
// works for numbers up to 59074 |
704 |
|
|
I32 i = SvIVX (sv); |
705 |
root |
1.35 |
U32 u; |
706 |
root |
1.39 |
char digit, nz = 0; |
707 |
root |
1.35 |
|
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 |
root |
1.39 |
// 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 |
root |
1.75 |
// 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 |
root |
1.39 |
digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
726 |
root |
1.35 |
} |
727 |
root |
1.75 |
else |
728 |
|
|
{ |
729 |
|
|
// large integer, use the (rather slow) snprintf way. |
730 |
|
|
need (enc, IVUV_MAXCHARS); |
731 |
|
|
enc->cur += |
732 |
|
|
SvIsUV(sv) |
733 |
|
|
? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) |
734 |
|
|
: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); |
735 |
|
|
} |
736 |
root |
1.1 |
} |
737 |
|
|
else if (SvROK (sv)) |
738 |
root |
1.21 |
encode_rv (enc, SvRV (sv)); |
739 |
root |
1.84 |
else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) |
740 |
root |
1.1 |
encode_str (enc, "null", 4, 0); |
741 |
|
|
else |
742 |
root |
1.9 |
croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
743 |
|
|
SvPV_nolen (sv), SvFLAGS (sv)); |
744 |
root |
1.1 |
} |
745 |
|
|
|
746 |
|
|
static SV * |
747 |
root |
1.48 |
encode_json (SV *scalar, JSON *json) |
748 |
root |
1.1 |
{ |
749 |
root |
1.25 |
enc_t enc; |
750 |
|
|
|
751 |
root |
1.47 |
if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
752 |
root |
1.9 |
croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
753 |
root |
1.3 |
|
754 |
root |
1.47 |
enc.json = *json; |
755 |
root |
1.12 |
enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
756 |
|
|
enc.cur = SvPVX (enc.sv); |
757 |
|
|
enc.end = SvEND (enc.sv); |
758 |
|
|
enc.indent = 0; |
759 |
root |
1.71 |
enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
760 |
|
|
: enc.json.flags & F_LATIN1 ? 0x000100UL |
761 |
root |
1.74 |
: 0x110000UL; |
762 |
root |
1.1 |
|
763 |
|
|
SvPOK_only (enc.sv); |
764 |
|
|
encode_sv (&enc, scalar); |
765 |
root |
1.100 |
encode_nl (&enc); |
766 |
root |
1.1 |
|
767 |
|
|
SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
768 |
root |
1.27 |
*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
769 |
root |
1.6 |
|
770 |
root |
1.47 |
if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
771 |
root |
1.30 |
SvUTF8_on (enc.sv); |
772 |
|
|
|
773 |
root |
1.47 |
if (enc.json.flags & F_SHRINK) |
774 |
root |
1.7 |
shrink (enc.sv); |
775 |
|
|
|
776 |
root |
1.1 |
return enc.sv; |
777 |
|
|
} |
778 |
|
|
|
779 |
|
|
///////////////////////////////////////////////////////////////////////////// |
780 |
root |
1.12 |
// decoder |
781 |
root |
1.1 |
|
782 |
root |
1.12 |
// structure used for decoding JSON |
783 |
|
|
typedef struct |
784 |
|
|
{ |
785 |
|
|
char *cur; // current parser pointer |
786 |
|
|
char *end; // end of input string |
787 |
|
|
const char *err; // parse error, if != 0 |
788 |
root |
1.48 |
JSON json; |
789 |
root |
1.18 |
U32 depth; // recursion depth |
790 |
|
|
U32 maxdepth; // recursion depth limit |
791 |
root |
1.12 |
} dec_t; |
792 |
|
|
|
793 |
root |
1.71 |
INLINE void |
794 |
root |
1.64 |
decode_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 |
|
|
|
802 |
root |
1.71 |
INLINE void |
803 |
root |
1.12 |
decode_ws (dec_t *dec) |
804 |
|
|
{ |
805 |
|
|
for (;;) |
806 |
|
|
{ |
807 |
|
|
char ch = *dec->cur; |
808 |
|
|
|
809 |
root |
1.64 |
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 |
|
|
} |
821 |
|
|
else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09) |
822 |
|
|
break; // parse error, but let higher level handle it, gives better error messages |
823 |
root |
1.63 |
|
824 |
root |
1.12 |
++dec->cur; |
825 |
root |
1.1 |
} |
826 |
root |
1.12 |
} |
827 |
root |
1.1 |
|
828 |
|
|
#define ERR(reason) SB dec->err = reason; goto fail; SE |
829 |
root |
1.18 |
|
830 |
root |
1.1 |
#define EXPECT_CH(ch) SB \ |
831 |
|
|
if (*dec->cur != ch) \ |
832 |
|
|
ERR (# ch " expected"); \ |
833 |
|
|
++dec->cur; \ |
834 |
|
|
SE |
835 |
|
|
|
836 |
root |
1.85 |
#define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED) |
837 |
root |
1.18 |
#define DEC_DEC_DEPTH --dec->depth |
838 |
|
|
|
839 |
root |
1.1 |
static SV *decode_sv (dec_t *dec); |
840 |
|
|
|
841 |
|
|
static signed char decode_hexdigit[256]; |
842 |
|
|
|
843 |
|
|
static UV |
844 |
|
|
decode_4hex (dec_t *dec) |
845 |
|
|
{ |
846 |
|
|
signed char d1, d2, d3, d4; |
847 |
root |
1.12 |
unsigned char *cur = (unsigned char *)dec->cur; |
848 |
root |
1.1 |
|
849 |
root |
1.40 |
d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected"); |
850 |
|
|
d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected"); |
851 |
|
|
d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected"); |
852 |
|
|
d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected"); |
853 |
root |
1.1 |
|
854 |
|
|
dec->cur += 4; |
855 |
|
|
|
856 |
|
|
return ((UV)d1) << 12 |
857 |
|
|
| ((UV)d2) << 8 |
858 |
|
|
| ((UV)d3) << 4 |
859 |
|
|
| ((UV)d4); |
860 |
|
|
|
861 |
|
|
fail: |
862 |
|
|
return (UV)-1; |
863 |
|
|
} |
864 |
|
|
|
865 |
|
|
static SV * |
866 |
|
|
decode_str (dec_t *dec) |
867 |
|
|
{ |
868 |
root |
1.12 |
SV *sv = 0; |
869 |
root |
1.1 |
int utf8 = 0; |
870 |
root |
1.38 |
char *dec_cur = dec->cur; |
871 |
root |
1.1 |
|
872 |
root |
1.12 |
do |
873 |
root |
1.1 |
{ |
874 |
root |
1.28 |
char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; |
875 |
root |
1.12 |
char *cur = buf; |
876 |
root |
1.1 |
|
877 |
root |
1.12 |
do |
878 |
root |
1.1 |
{ |
879 |
root |
1.38 |
unsigned char ch = *(unsigned char *)dec_cur++; |
880 |
root |
1.12 |
|
881 |
root |
1.35 |
if (expect_false (ch == '"')) |
882 |
root |
1.12 |
{ |
883 |
root |
1.38 |
--dec_cur; |
884 |
root |
1.12 |
break; |
885 |
|
|
} |
886 |
root |
1.35 |
else if (expect_false (ch == '\\')) |
887 |
root |
1.1 |
{ |
888 |
root |
1.38 |
switch (*dec_cur) |
889 |
root |
1.1 |
{ |
890 |
root |
1.12 |
case '\\': |
891 |
|
|
case '/': |
892 |
root |
1.38 |
case '"': *cur++ = *dec_cur++; break; |
893 |
root |
1.12 |
|
894 |
root |
1.38 |
case 'b': ++dec_cur; *cur++ = '\010'; break; |
895 |
|
|
case 't': ++dec_cur; *cur++ = '\011'; break; |
896 |
|
|
case 'n': ++dec_cur; *cur++ = '\012'; break; |
897 |
|
|
case 'f': ++dec_cur; *cur++ = '\014'; break; |
898 |
|
|
case 'r': ++dec_cur; *cur++ = '\015'; break; |
899 |
root |
1.1 |
|
900 |
root |
1.12 |
case 'u': |
901 |
root |
1.1 |
{ |
902 |
root |
1.12 |
UV lo, hi; |
903 |
root |
1.38 |
++dec_cur; |
904 |
root |
1.1 |
|
905 |
root |
1.38 |
dec->cur = dec_cur; |
906 |
root |
1.12 |
hi = decode_4hex (dec); |
907 |
root |
1.38 |
dec_cur = dec->cur; |
908 |
root |
1.12 |
if (hi == (UV)-1) |
909 |
|
|
goto fail; |
910 |
root |
1.1 |
|
911 |
root |
1.12 |
// possibly a surrogate pair |
912 |
|
|
if (hi >= 0xd800) |
913 |
|
|
if (hi < 0xdc00) |
914 |
|
|
{ |
915 |
root |
1.38 |
if (dec_cur [0] != '\\' || dec_cur [1] != 'u') |
916 |
root |
1.12 |
ERR ("missing low surrogate character in surrogate pair"); |
917 |
|
|
|
918 |
root |
1.38 |
dec_cur += 2; |
919 |
root |
1.12 |
|
920 |
root |
1.38 |
dec->cur = dec_cur; |
921 |
root |
1.12 |
lo = decode_4hex (dec); |
922 |
root |
1.38 |
dec_cur = dec->cur; |
923 |
root |
1.12 |
if (lo == (UV)-1) |
924 |
|
|
goto fail; |
925 |
|
|
|
926 |
|
|
if (lo < 0xdc00 || lo >= 0xe000) |
927 |
|
|
ERR ("surrogate pair expected"); |
928 |
|
|
|
929 |
|
|
hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; |
930 |
|
|
} |
931 |
|
|
else if (hi < 0xe000) |
932 |
|
|
ERR ("missing high surrogate character in surrogate pair"); |
933 |
root |
1.1 |
|
934 |
root |
1.12 |
if (hi >= 0x80) |
935 |
|
|
{ |
936 |
|
|
utf8 = 1; |
937 |
root |
1.1 |
|
938 |
root |
1.72 |
cur = encode_utf8 (cur, hi); |
939 |
root |
1.12 |
} |
940 |
|
|
else |
941 |
|
|
*cur++ = hi; |
942 |
root |
1.1 |
} |
943 |
root |
1.12 |
break; |
944 |
|
|
|
945 |
|
|
default: |
946 |
root |
1.38 |
--dec_cur; |
947 |
root |
1.12 |
ERR ("illegal backslash escape sequence in string"); |
948 |
|
|
} |
949 |
|
|
} |
950 |
root |
1.74 |
else if (expect_true (ch >= 0x20 && ch < 0x80)) |
951 |
root |
1.12 |
*cur++ = ch; |
952 |
|
|
else if (ch >= 0x80) |
953 |
|
|
{ |
954 |
root |
1.25 |
STRLEN clen; |
955 |
|
|
|
956 |
root |
1.38 |
--dec_cur; |
957 |
root |
1.1 |
|
958 |
root |
1.101 |
decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
959 |
root |
1.12 |
if (clen == (STRLEN)-1) |
960 |
|
|
ERR ("malformed UTF-8 character in JSON string"); |
961 |
root |
1.1 |
|
962 |
root |
1.12 |
do |
963 |
root |
1.38 |
*cur++ = *dec_cur++; |
964 |
root |
1.12 |
while (--clen); |
965 |
root |
1.5 |
|
966 |
root |
1.12 |
utf8 = 1; |
967 |
root |
1.1 |
} |
968 |
root |
1.12 |
else |
969 |
root |
1.23 |
{ |
970 |
root |
1.38 |
--dec_cur; |
971 |
root |
1.12 |
|
972 |
root |
1.23 |
if (!ch) |
973 |
root |
1.24 |
ERR ("unexpected end of string while parsing JSON string"); |
974 |
root |
1.23 |
else |
975 |
root |
1.24 |
ERR ("invalid character encountered while parsing JSON string"); |
976 |
root |
1.23 |
} |
977 |
root |
1.1 |
} |
978 |
root |
1.12 |
while (cur < buf + SHORT_STRING_LEN); |
979 |
root |
1.1 |
|
980 |
root |
1.25 |
{ |
981 |
|
|
STRLEN len = cur - buf; |
982 |
root |
1.5 |
|
983 |
root |
1.25 |
if (sv) |
984 |
|
|
{ |
985 |
root |
1.95 |
STRLEN cur = SvCUR (sv); |
986 |
|
|
|
987 |
|
|
if (SvLEN (sv) <= cur + len) |
988 |
|
|
SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
989 |
|
|
|
990 |
root |
1.25 |
memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
991 |
|
|
SvCUR_set (sv, SvCUR (sv) + len); |
992 |
|
|
} |
993 |
|
|
else |
994 |
|
|
sv = newSVpvn (buf, len); |
995 |
|
|
} |
996 |
root |
1.1 |
} |
997 |
root |
1.38 |
while (*dec_cur != '"'); |
998 |
root |
1.1 |
|
999 |
root |
1.38 |
++dec_cur; |
1000 |
root |
1.1 |
|
1001 |
root |
1.12 |
if (sv) |
1002 |
|
|
{ |
1003 |
|
|
SvPOK_only (sv); |
1004 |
|
|
*SvEND (sv) = 0; |
1005 |
root |
1.4 |
|
1006 |
root |
1.12 |
if (utf8) |
1007 |
|
|
SvUTF8_on (sv); |
1008 |
|
|
} |
1009 |
|
|
else |
1010 |
|
|
sv = newSVpvn ("", 0); |
1011 |
root |
1.6 |
|
1012 |
root |
1.38 |
dec->cur = dec_cur; |
1013 |
root |
1.1 |
return sv; |
1014 |
|
|
|
1015 |
|
|
fail: |
1016 |
root |
1.38 |
dec->cur = dec_cur; |
1017 |
root |
1.1 |
return 0; |
1018 |
|
|
} |
1019 |
|
|
|
1020 |
|
|
static SV * |
1021 |
|
|
decode_num (dec_t *dec) |
1022 |
|
|
{ |
1023 |
|
|
int is_nv = 0; |
1024 |
|
|
char *start = dec->cur; |
1025 |
|
|
|
1026 |
|
|
// [minus] |
1027 |
|
|
if (*dec->cur == '-') |
1028 |
|
|
++dec->cur; |
1029 |
|
|
|
1030 |
|
|
if (*dec->cur == '0') |
1031 |
|
|
{ |
1032 |
|
|
++dec->cur; |
1033 |
|
|
if (*dec->cur >= '0' && *dec->cur <= '9') |
1034 |
|
|
ERR ("malformed number (leading zero must not be followed by another digit)"); |
1035 |
|
|
} |
1036 |
root |
1.5 |
else if (*dec->cur < '0' || *dec->cur > '9') |
1037 |
|
|
ERR ("malformed number (no digits after initial minus)"); |
1038 |
|
|
else |
1039 |
|
|
do |
1040 |
|
|
{ |
1041 |
|
|
++dec->cur; |
1042 |
|
|
} |
1043 |
|
|
while (*dec->cur >= '0' && *dec->cur <= '9'); |
1044 |
root |
1.1 |
|
1045 |
|
|
// [frac] |
1046 |
|
|
if (*dec->cur == '.') |
1047 |
|
|
{ |
1048 |
root |
1.5 |
++dec->cur; |
1049 |
|
|
|
1050 |
|
|
if (*dec->cur < '0' || *dec->cur > '9') |
1051 |
|
|
ERR ("malformed number (no digits after decimal point)"); |
1052 |
root |
1.1 |
|
1053 |
|
|
do |
1054 |
|
|
{ |
1055 |
|
|
++dec->cur; |
1056 |
|
|
} |
1057 |
|
|
while (*dec->cur >= '0' && *dec->cur <= '9'); |
1058 |
root |
1.5 |
|
1059 |
|
|
is_nv = 1; |
1060 |
root |
1.1 |
} |
1061 |
|
|
|
1062 |
|
|
// [exp] |
1063 |
|
|
if (*dec->cur == 'e' || *dec->cur == 'E') |
1064 |
|
|
{ |
1065 |
root |
1.5 |
++dec->cur; |
1066 |
root |
1.1 |
|
1067 |
|
|
if (*dec->cur == '-' || *dec->cur == '+') |
1068 |
|
|
++dec->cur; |
1069 |
|
|
|
1070 |
root |
1.5 |
if (*dec->cur < '0' || *dec->cur > '9') |
1071 |
|
|
ERR ("malformed number (no digits after exp sign)"); |
1072 |
|
|
|
1073 |
|
|
do |
1074 |
|
|
{ |
1075 |
|
|
++dec->cur; |
1076 |
|
|
} |
1077 |
|
|
while (*dec->cur >= '0' && *dec->cur <= '9'); |
1078 |
|
|
|
1079 |
|
|
is_nv = 1; |
1080 |
root |
1.1 |
} |
1081 |
|
|
|
1082 |
|
|
if (!is_nv) |
1083 |
|
|
{ |
1084 |
root |
1.56 |
int len = dec->cur - start; |
1085 |
|
|
|
1086 |
root |
1.70 |
// special case the rather common 1..5-digit-int case |
1087 |
root |
1.35 |
if (*start == '-') |
1088 |
root |
1.56 |
switch (len) |
1089 |
root |
1.1 |
{ |
1090 |
root |
1.97 |
case 2: return newSViv (-(IV)( start [1] - '0' * 1)); |
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)); |
1095 |
root |
1.1 |
} |
1096 |
root |
1.35 |
else |
1097 |
root |
1.56 |
switch (len) |
1098 |
root |
1.35 |
{ |
1099 |
root |
1.97 |
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 |
root |
1.35 |
} |
1105 |
|
|
|
1106 |
|
|
{ |
1107 |
|
|
UV uv; |
1108 |
root |
1.56 |
int numtype = grok_number (start, len, &uv); |
1109 |
root |
1.35 |
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 |
|
|
} |
1115 |
|
|
else |
1116 |
|
|
return newSVuv (uv); |
1117 |
root |
1.56 |
} |
1118 |
|
|
|
1119 |
|
|
len -= *start == '-' ? 1 : 0; |
1120 |
root |
1.40 |
|
1121 |
root |
1.56 |
// 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 |
root |
1.1 |
} |
1133 |
|
|
|
1134 |
root |
1.56 |
// loss of precision here |
1135 |
root |
1.1 |
return newSVnv (Atof (start)); |
1136 |
|
|
|
1137 |
|
|
fail: |
1138 |
|
|
return 0; |
1139 |
|
|
} |
1140 |
|
|
|
1141 |
|
|
static SV * |
1142 |
|
|
decode_av (dec_t *dec) |
1143 |
|
|
{ |
1144 |
|
|
AV *av = newAV (); |
1145 |
|
|
|
1146 |
root |
1.18 |
DEC_INC_DEPTH; |
1147 |
root |
1.12 |
decode_ws (dec); |
1148 |
root |
1.18 |
|
1149 |
root |
1.5 |
if (*dec->cur == ']') |
1150 |
|
|
++dec->cur; |
1151 |
|
|
else |
1152 |
|
|
for (;;) |
1153 |
|
|
{ |
1154 |
|
|
SV *value; |
1155 |
root |
1.1 |
|
1156 |
root |
1.5 |
value = decode_sv (dec); |
1157 |
|
|
if (!value) |
1158 |
|
|
goto fail; |
1159 |
root |
1.1 |
|
1160 |
root |
1.5 |
av_push (av, value); |
1161 |
root |
1.1 |
|
1162 |
root |
1.12 |
decode_ws (dec); |
1163 |
root |
1.1 |
|
1164 |
root |
1.5 |
if (*dec->cur == ']') |
1165 |
|
|
{ |
1166 |
|
|
++dec->cur; |
1167 |
|
|
break; |
1168 |
|
|
} |
1169 |
|
|
|
1170 |
|
|
if (*dec->cur != ',') |
1171 |
|
|
ERR (", or ] expected while parsing array"); |
1172 |
root |
1.1 |
|
1173 |
root |
1.5 |
++dec->cur; |
1174 |
root |
1.63 |
|
1175 |
|
|
decode_ws (dec); |
1176 |
|
|
|
1177 |
|
|
if (*dec->cur == ']' && dec->json.flags & F_RELAXED) |
1178 |
|
|
{ |
1179 |
|
|
++dec->cur; |
1180 |
|
|
break; |
1181 |
|
|
} |
1182 |
root |
1.5 |
} |
1183 |
root |
1.1 |
|
1184 |
root |
1.18 |
DEC_DEC_DEPTH; |
1185 |
root |
1.1 |
return newRV_noinc ((SV *)av); |
1186 |
|
|
|
1187 |
|
|
fail: |
1188 |
|
|
SvREFCNT_dec (av); |
1189 |
root |
1.18 |
DEC_DEC_DEPTH; |
1190 |
root |
1.1 |
return 0; |
1191 |
|
|
} |
1192 |
|
|
|
1193 |
|
|
static SV * |
1194 |
|
|
decode_hv (dec_t *dec) |
1195 |
|
|
{ |
1196 |
root |
1.49 |
SV *sv; |
1197 |
root |
1.1 |
HV *hv = newHV (); |
1198 |
|
|
|
1199 |
root |
1.18 |
DEC_INC_DEPTH; |
1200 |
root |
1.12 |
decode_ws (dec); |
1201 |
root |
1.18 |
|
1202 |
root |
1.5 |
if (*dec->cur == '}') |
1203 |
|
|
++dec->cur; |
1204 |
|
|
else |
1205 |
|
|
for (;;) |
1206 |
|
|
{ |
1207 |
root |
1.63 |
EXPECT_CH ('"'); |
1208 |
root |
1.1 |
|
1209 |
root |
1.46 |
// heuristic: assume that |
1210 |
root |
1.47 |
// 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 |
root |
1.46 |
{ |
1215 |
|
|
SV *value; |
1216 |
|
|
char *p = dec->cur; |
1217 |
|
|
char *e = p + 24; // only try up to 24 bytes |
1218 |
|
|
|
1219 |
|
|
for (;;) |
1220 |
|
|
{ |
1221 |
root |
1.77 |
// the >= 0x80 is false on most architectures |
1222 |
root |
1.46 |
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 |
root |
1.63 |
decode_ws (dec); |
1232 |
root |
1.46 |
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); |
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 |
root |
1.63 |
decode_ws (dec); |
1254 |
root |
1.46 |
value = decode_sv (dec); |
1255 |
|
|
if (!value) |
1256 |
|
|
goto fail; |
1257 |
root |
1.1 |
|
1258 |
root |
1.46 |
hv_store (hv, key, len, value, 0); |
1259 |
root |
1.1 |
|
1260 |
root |
1.46 |
break; |
1261 |
|
|
} |
1262 |
root |
1.1 |
|
1263 |
root |
1.46 |
++p; |
1264 |
|
|
} |
1265 |
|
|
} |
1266 |
root |
1.1 |
|
1267 |
root |
1.12 |
decode_ws (dec); |
1268 |
root |
1.1 |
|
1269 |
root |
1.5 |
if (*dec->cur == '}') |
1270 |
|
|
{ |
1271 |
|
|
++dec->cur; |
1272 |
|
|
break; |
1273 |
|
|
} |
1274 |
root |
1.1 |
|
1275 |
root |
1.5 |
if (*dec->cur != ',') |
1276 |
|
|
ERR (", or } expected while parsing object/hash"); |
1277 |
root |
1.1 |
|
1278 |
root |
1.5 |
++dec->cur; |
1279 |
root |
1.63 |
|
1280 |
|
|
decode_ws (dec); |
1281 |
|
|
|
1282 |
|
|
if (*dec->cur == '}' && dec->json.flags & F_RELAXED) |
1283 |
|
|
{ |
1284 |
|
|
++dec->cur; |
1285 |
|
|
break; |
1286 |
|
|
} |
1287 |
root |
1.5 |
} |
1288 |
root |
1.1 |
|
1289 |
root |
1.18 |
DEC_DEC_DEPTH; |
1290 |
root |
1.49 |
sv = newRV_noinc ((SV *)hv); |
1291 |
|
|
|
1292 |
|
|
// check filter callbacks |
1293 |
|
|
if (dec->json.flags & F_HOOK) |
1294 |
|
|
{ |
1295 |
root |
1.51 |
if (dec->json.cb_sk_object && HvKEYS (hv) == 1) |
1296 |
root |
1.49 |
{ |
1297 |
root |
1.52 |
HE *cb, *he; |
1298 |
root |
1.49 |
|
1299 |
root |
1.52 |
hv_iterinit (hv); |
1300 |
|
|
he = hv_iternext (hv); |
1301 |
|
|
hv_iterinit (hv); |
1302 |
root |
1.49 |
|
1303 |
root |
1.52 |
// 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 |
root |
1.49 |
|
1307 |
root |
1.52 |
if (cb) |
1308 |
root |
1.51 |
{ |
1309 |
root |
1.53 |
dSP; |
1310 |
root |
1.52 |
int count; |
1311 |
|
|
|
1312 |
root |
1.53 |
ENTER; SAVETMPS; PUSHMARK (SP); |
1313 |
root |
1.52 |
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 |
root |
1.51 |
} |
1326 |
root |
1.49 |
} |
1327 |
|
|
|
1328 |
|
|
if (dec->json.cb_object) |
1329 |
|
|
{ |
1330 |
root |
1.53 |
dSP; |
1331 |
root |
1.49 |
int count; |
1332 |
|
|
|
1333 |
root |
1.53 |
ENTER; SAVETMPS; PUSHMARK (SP); |
1334 |
root |
1.49 |
XPUSHs (sv_2mortal (sv)); |
1335 |
|
|
|
1336 |
|
|
PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
1337 |
|
|
|
1338 |
|
|
if (count == 1) |
1339 |
root |
1.51 |
{ |
1340 |
|
|
sv = newSVsv (POPs); |
1341 |
root |
1.52 |
FREETMPS; LEAVE; |
1342 |
|
|
return sv; |
1343 |
root |
1.51 |
} |
1344 |
|
|
|
1345 |
|
|
SvREFCNT_inc (sv); |
1346 |
root |
1.52 |
FREETMPS; LEAVE; |
1347 |
root |
1.49 |
} |
1348 |
|
|
} |
1349 |
|
|
|
1350 |
root |
1.51 |
return sv; |
1351 |
root |
1.1 |
|
1352 |
|
|
fail: |
1353 |
|
|
SvREFCNT_dec (hv); |
1354 |
root |
1.18 |
DEC_DEC_DEPTH; |
1355 |
root |
1.1 |
return 0; |
1356 |
|
|
} |
1357 |
|
|
|
1358 |
|
|
static SV * |
1359 |
|
|
decode_sv (dec_t *dec) |
1360 |
|
|
{ |
1361 |
root |
1.40 |
// the beauty of JSON: you need exactly one character lookahead |
1362 |
root |
1.73 |
// to parse everything. |
1363 |
root |
1.1 |
switch (*dec->cur) |
1364 |
|
|
{ |
1365 |
|
|
case '"': ++dec->cur; return decode_str (dec); |
1366 |
root |
1.73 |
case '[': ++dec->cur; return decode_av (dec); |
1367 |
|
|
case '{': ++dec->cur; return decode_hv (dec); |
1368 |
root |
1.1 |
|
1369 |
|
|
case '-': |
1370 |
|
|
case '0': case '1': case '2': case '3': case '4': |
1371 |
|
|
case '5': case '6': case '7': case '8': case '9': |
1372 |
|
|
return decode_num (dec); |
1373 |
|
|
|
1374 |
|
|
case 't': |
1375 |
|
|
if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1376 |
|
|
{ |
1377 |
|
|
dec->cur += 4; |
1378 |
root |
1.60 |
#if JSON_SLOW |
1379 |
root |
1.83 |
json_true = get_bool ("JSON::XS::true"); |
1380 |
root |
1.60 |
#endif |
1381 |
root |
1.83 |
return newSVsv (json_true); |
1382 |
root |
1.1 |
} |
1383 |
|
|
else |
1384 |
|
|
ERR ("'true' expected"); |
1385 |
|
|
|
1386 |
|
|
break; |
1387 |
|
|
|
1388 |
|
|
case 'f': |
1389 |
|
|
if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1390 |
|
|
{ |
1391 |
|
|
dec->cur += 5; |
1392 |
root |
1.60 |
#if JSON_SLOW |
1393 |
root |
1.83 |
json_false = get_bool ("JSON::XS::false"); |
1394 |
root |
1.60 |
#endif |
1395 |
root |
1.83 |
return newSVsv (json_false); |
1396 |
root |
1.1 |
} |
1397 |
|
|
else |
1398 |
|
|
ERR ("'false' expected"); |
1399 |
|
|
|
1400 |
|
|
break; |
1401 |
|
|
|
1402 |
|
|
case 'n': |
1403 |
|
|
if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
1404 |
|
|
{ |
1405 |
|
|
dec->cur += 4; |
1406 |
root |
1.5 |
return newSVsv (&PL_sv_undef); |
1407 |
root |
1.1 |
} |
1408 |
|
|
else |
1409 |
|
|
ERR ("'null' expected"); |
1410 |
|
|
|
1411 |
|
|
break; |
1412 |
|
|
|
1413 |
|
|
default: |
1414 |
root |
1.24 |
ERR ("malformed JSON string, neither array, object, number, string or atom"); |
1415 |
root |
1.1 |
break; |
1416 |
|
|
} |
1417 |
|
|
|
1418 |
|
|
fail: |
1419 |
|
|
return 0; |
1420 |
|
|
} |
1421 |
|
|
|
1422 |
|
|
static SV * |
1423 |
root |
1.96 |
decode_json (SV *string, JSON *json, char **offset_return) |
1424 |
root |
1.1 |
{ |
1425 |
root |
1.25 |
dec_t dec; |
1426 |
root |
1.1 |
SV *sv; |
1427 |
|
|
|
1428 |
root |
1.94 |
/* 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 |
root |
1.22 |
SvUPGRADE (string, SVt_PV); |
1436 |
|
|
|
1437 |
root |
1.87 |
/* 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 |
root |
1.89 |
* But it doesn't fix it, so try another workaround, call SvPV_nolen |
1442 |
|
|
* and hope for the best. |
1443 |
root |
1.90 |
* 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 |
root |
1.87 |
*/ |
1447 |
root |
1.96 |
{ |
1448 |
root |
1.89 |
#ifdef DEBUGGING |
1449 |
root |
1.96 |
STRLEN offset = SvOK (string) ? sv_len (string) : 0; |
1450 |
root |
1.90 |
#else |
1451 |
root |
1.96 |
STRLEN offset = SvCUR (string); |
1452 |
root |
1.89 |
#endif |
1453 |
root |
1.87 |
|
1454 |
root |
1.96 |
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 |
root |
1.45 |
|
1459 |
root |
1.47 |
if (json->flags & F_UTF8) |
1460 |
root |
1.5 |
sv_utf8_downgrade (string, 0); |
1461 |
|
|
else |
1462 |
root |
1.1 |
sv_utf8_upgrade (string); |
1463 |
|
|
|
1464 |
|
|
SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1465 |
|
|
|
1466 |
root |
1.85 |
dec.json = *json; |
1467 |
|
|
dec.cur = SvPVX (string); |
1468 |
|
|
dec.end = SvEND (string); |
1469 |
|
|
dec.err = 0; |
1470 |
|
|
dec.depth = 0; |
1471 |
root |
1.1 |
|
1472 |
root |
1.49 |
if (dec.json.cb_object || dec.json.cb_sk_object) |
1473 |
|
|
dec.json.flags |= F_HOOK; |
1474 |
|
|
|
1475 |
root |
1.31 |
*dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1476 |
root |
1.63 |
|
1477 |
|
|
decode_ws (&dec); |
1478 |
root |
1.1 |
sv = decode_sv (&dec); |
1479 |
|
|
|
1480 |
root |
1.96 |
if (offset_return) |
1481 |
|
|
*offset_return = dec.cur; |
1482 |
|
|
|
1483 |
root |
1.32 |
if (!(offset_return || !sv)) |
1484 |
root |
1.31 |
{ |
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 |
|
|
|
1496 |
root |
1.1 |
if (!sv) |
1497 |
|
|
{ |
1498 |
|
|
SV *uni = sv_newmortal (); |
1499 |
root |
1.8 |
|
1500 |
root |
1.5 |
// horrible hack to silence warning inside pv_uni_display |
1501 |
root |
1.8 |
COP cop = *PL_curcop; |
1502 |
root |
1.5 |
cop.cop_warnings = pWARN_NONE; |
1503 |
root |
1.8 |
ENTER; |
1504 |
root |
1.5 |
SAVEVPTR (PL_curcop); |
1505 |
|
|
PL_curcop = &cop; |
1506 |
root |
1.8 |
pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1507 |
|
|
LEAVE; |
1508 |
root |
1.1 |
|
1509 |
root |
1.96 |
croak ("%s, at character offset %d (before \"%s\")", |
1510 |
root |
1.1 |
dec.err, |
1511 |
root |
1.96 |
ptr_to_index (string, dec.cur), |
1512 |
root |
1.1 |
dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1513 |
|
|
} |
1514 |
|
|
|
1515 |
root |
1.3 |
sv = sv_2mortal (sv); |
1516 |
|
|
|
1517 |
root |
1.47 |
if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1518 |
root |
1.9 |
croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1519 |
root |
1.3 |
|
1520 |
|
|
return sv; |
1521 |
root |
1.1 |
} |
1522 |
|
|
|
1523 |
root |
1.12 |
///////////////////////////////////////////////////////////////////////////// |
1524 |
root |
1.77 |
// incremental parser |
1525 |
|
|
|
1526 |
|
|
static void |
1527 |
|
|
incr_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 |
root |
1.85 |
if (++self->incr_nest > self->max_depth) |
1620 |
|
|
croak (ERR_NESTING_EXCEEDED); |
1621 |
root |
1.77 |
break; |
1622 |
|
|
|
1623 |
|
|
case ']': |
1624 |
|
|
case '}': |
1625 |
root |
1.88 |
if (--self->incr_nest <= 0) |
1626 |
root |
1.77 |
goto interrupt; |
1627 |
|
|
} |
1628 |
|
|
} |
1629 |
|
|
} |
1630 |
|
|
|
1631 |
|
|
modechange: |
1632 |
|
|
; |
1633 |
|
|
} |
1634 |
|
|
|
1635 |
|
|
interrupt: |
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 |
1638 |
|
|
} |
1639 |
|
|
|
1640 |
|
|
///////////////////////////////////////////////////////////////////////////// |
1641 |
root |
1.12 |
// XS interface functions |
1642 |
|
|
|
1643 |
root |
1.1 |
MODULE = JSON::XS PACKAGE = JSON::XS |
1644 |
|
|
|
1645 |
|
|
BOOT: |
1646 |
|
|
{ |
1647 |
|
|
int i; |
1648 |
|
|
|
1649 |
root |
1.18 |
for (i = 0; i < 256; ++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; |
1655 |
root |
1.1 |
|
1656 |
root |
1.44 |
json_stash = gv_stashpv ("JSON::XS" , 1); |
1657 |
|
|
json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
1658 |
root |
1.43 |
|
1659 |
root |
1.83 |
json_true = get_bool ("JSON::XS::true"); |
1660 |
|
|
json_false = get_bool ("JSON::XS::false"); |
1661 |
root |
1.99 |
|
1662 |
|
|
CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ |
1663 |
root |
1.1 |
} |
1664 |
|
|
|
1665 |
root |
1.4 |
PROTOTYPES: DISABLE |
1666 |
|
|
|
1667 |
root |
1.57 |
void CLONE (...) |
1668 |
|
|
CODE: |
1669 |
|
|
json_stash = 0; |
1670 |
|
|
json_boolean_stash = 0; |
1671 |
|
|
|
1672 |
root |
1.48 |
void new (char *klass) |
1673 |
|
|
PPCODE: |
1674 |
|
|
{ |
1675 |
|
|
SV *pv = NEWSV (0, sizeof (JSON)); |
1676 |
|
|
SvPOK_only (pv); |
1677 |
root |
1.85 |
json_init ((JSON *)SvPVX (pv)); |
1678 |
root |
1.68 |
XPUSHs (sv_2mortal (sv_bless ( |
1679 |
|
|
newRV_noinc (pv), |
1680 |
|
|
strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
1681 |
|
|
))); |
1682 |
root |
1.48 |
} |
1683 |
root |
1.1 |
|
1684 |
root |
1.48 |
void ascii (JSON *self, int enable = 1) |
1685 |
root |
1.1 |
ALIAS: |
1686 |
root |
1.44 |
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 |
root |
1.63 |
relaxed = F_RELAXED |
1699 |
root |
1.84 |
allow_unknown = F_ALLOW_UNKNOWN |
1700 |
root |
1.48 |
PPCODE: |
1701 |
root |
1.1 |
{ |
1702 |
|
|
if (enable) |
1703 |
root |
1.48 |
self->flags |= ix; |
1704 |
root |
1.1 |
else |
1705 |
root |
1.48 |
self->flags &= ~ix; |
1706 |
root |
1.1 |
|
1707 |
root |
1.48 |
XPUSHs (ST (0)); |
1708 |
root |
1.1 |
} |
1709 |
|
|
|
1710 |
root |
1.66 |
void 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 |
root |
1.84 |
get_allow_unknown = F_ALLOW_UNKNOWN |
1725 |
root |
1.66 |
PPCODE: |
1726 |
|
|
XPUSHs (boolSV (self->flags & ix)); |
1727 |
|
|
|
1728 |
root |
1.85 |
void max_depth (JSON *self, U32 max_depth = 0x80000000UL) |
1729 |
root |
1.48 |
PPCODE: |
1730 |
root |
1.85 |
self->max_depth = max_depth; |
1731 |
root |
1.48 |
XPUSHs (ST (0)); |
1732 |
root |
1.18 |
|
1733 |
root |
1.67 |
U32 get_max_depth (JSON *self) |
1734 |
root |
1.66 |
CODE: |
1735 |
root |
1.85 |
RETVAL = self->max_depth; |
1736 |
root |
1.66 |
OUTPUT: |
1737 |
|
|
RETVAL |
1738 |
|
|
|
1739 |
root |
1.85 |
void max_size (JSON *self, U32 max_size = 0) |
1740 |
root |
1.48 |
PPCODE: |
1741 |
root |
1.85 |
self->max_size = max_size; |
1742 |
root |
1.48 |
XPUSHs (ST (0)); |
1743 |
root |
1.45 |
|
1744 |
root |
1.66 |
int get_max_size (JSON *self) |
1745 |
|
|
CODE: |
1746 |
root |
1.85 |
RETVAL = self->max_size; |
1747 |
root |
1.66 |
OUTPUT: |
1748 |
|
|
RETVAL |
1749 |
|
|
|
1750 |
root |
1.51 |
void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1751 |
root |
1.49 |
PPCODE: |
1752 |
|
|
{ |
1753 |
root |
1.52 |
SvREFCNT_dec (self->cb_object); |
1754 |
|
|
self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; |
1755 |
|
|
|
1756 |
|
|
XPUSHs (ST (0)); |
1757 |
|
|
} |
1758 |
|
|
|
1759 |
|
|
void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) |
1760 |
|
|
PPCODE: |
1761 |
|
|
{ |
1762 |
|
|
if (!self->cb_sk_object) |
1763 |
|
|
self->cb_sk_object = newHV (); |
1764 |
root |
1.51 |
|
1765 |
root |
1.52 |
if (SvOK (cb)) |
1766 |
|
|
hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
1767 |
root |
1.51 |
else |
1768 |
root |
1.52 |
{ |
1769 |
|
|
hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0); |
1770 |
root |
1.49 |
|
1771 |
root |
1.52 |
if (!HvKEYS (self->cb_sk_object)) |
1772 |
|
|
{ |
1773 |
|
|
SvREFCNT_dec (self->cb_sk_object); |
1774 |
|
|
self->cb_sk_object = 0; |
1775 |
|
|
} |
1776 |
root |
1.49 |
} |
1777 |
|
|
|
1778 |
|
|
XPUSHs (ST (0)); |
1779 |
|
|
} |
1780 |
|
|
|
1781 |
root |
1.48 |
void encode (JSON *self, SV *scalar) |
1782 |
root |
1.1 |
PPCODE: |
1783 |
root |
1.48 |
XPUSHs (encode_json (scalar, self)); |
1784 |
root |
1.1 |
|
1785 |
root |
1.48 |
void decode (JSON *self, SV *jsonstr) |
1786 |
root |
1.1 |
PPCODE: |
1787 |
root |
1.48 |
XPUSHs (decode_json (jsonstr, self, 0)); |
1788 |
root |
1.31 |
|
1789 |
root |
1.48 |
void decode_prefix (JSON *self, SV *jsonstr) |
1790 |
root |
1.31 |
PPCODE: |
1791 |
|
|
{ |
1792 |
root |
1.96 |
char *offset; |
1793 |
root |
1.31 |
EXTEND (SP, 2); |
1794 |
root |
1.48 |
PUSHs (decode_json (jsonstr, self, &offset)); |
1795 |
root |
1.96 |
PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); |
1796 |
root |
1.31 |
} |
1797 |
root |
1.2 |
|
1798 |
root |
1.77 |
void 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 |
root |
1.96 |
if (SvUTF8 (jsonstr)) |
1808 |
root |
1.77 |
{ |
1809 |
root |
1.96 |
if (!SvUTF8 (self->incr_text)) |
1810 |
|
|
{ |
1811 |
|
|
/* utf-8-ness differs, need to upgrade */ |
1812 |
|
|
sv_utf8_upgrade (self->incr_text); |
1813 |
root |
1.77 |
|
1814 |
root |
1.96 |
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 |
root |
1.77 |
} |
1819 |
root |
1.96 |
else if (SvUTF8 (self->incr_text)) |
1820 |
|
|
sv_utf8_upgrade (jsonstr); |
1821 |
root |
1.77 |
|
1822 |
|
|
{ |
1823 |
|
|
STRLEN len; |
1824 |
|
|
const char *str = SvPV (jsonstr, len); |
1825 |
root |
1.95 |
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 |
root |
1.77 |
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 |
root |
1.96 |
char *offset; |
1840 |
root |
1.77 |
|
1841 |
|
|
if (!INCR_DONE (self)) |
1842 |
root |
1.82 |
{ |
1843 |
|
|
incr_parse (self); |
1844 |
root |
1.85 |
|
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 |
root |
1.82 |
if (!INCR_DONE (self)) |
1850 |
|
|
break; |
1851 |
|
|
} |
1852 |
root |
1.77 |
|
1853 |
|
|
XPUSHs (decode_json (self->incr_text, self, &offset)); |
1854 |
|
|
|
1855 |
root |
1.96 |
self->incr_pos -= offset - SvPVX (self->incr_text); |
1856 |
root |
1.77 |
self->incr_nest = 0; |
1857 |
|
|
self->incr_mode = 0; |
1858 |
root |
1.96 |
|
1859 |
|
|
sv_chop (self->incr_text, offset); |
1860 |
root |
1.77 |
} |
1861 |
|
|
while (GIMME_V == G_ARRAY); |
1862 |
|
|
} |
1863 |
|
|
|
1864 |
|
|
SV *incr_text (JSON *self) |
1865 |
|
|
ATTRS: lvalue |
1866 |
|
|
CODE: |
1867 |
|
|
{ |
1868 |
|
|
if (self->incr_pos) |
1869 |
root |
1.82 |
croak ("incr_text can not be called when the incremental parser already started parsing"); |
1870 |
root |
1.77 |
|
1871 |
|
|
RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef; |
1872 |
|
|
} |
1873 |
|
|
OUTPUT: |
1874 |
|
|
RETVAL |
1875 |
|
|
|
1876 |
root |
1.79 |
void incr_skip (JSON *self) |
1877 |
|
|
CODE: |
1878 |
|
|
{ |
1879 |
root |
1.81 |
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 |
root |
1.79 |
} |
1887 |
|
|
|
1888 |
root |
1.86 |
void 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 |
|
|
|
1898 |
root |
1.52 |
void DESTROY (JSON *self) |
1899 |
|
|
CODE: |
1900 |
|
|
SvREFCNT_dec (self->cb_sk_object); |
1901 |
|
|
SvREFCNT_dec (self->cb_object); |
1902 |
root |
1.77 |
SvREFCNT_dec (self->incr_text); |
1903 |
root |
1.52 |
|
1904 |
root |
1.4 |
PROTOTYPES: ENABLE |
1905 |
|
|
|
1906 |
root |
1.69 |
void encode_json (SV *scalar) |
1907 |
root |
1.76 |
ALIAS: |
1908 |
|
|
to_json_ = 0 |
1909 |
|
|
encode_json = F_UTF8 |
1910 |
root |
1.2 |
PPCODE: |
1911 |
root |
1.47 |
{ |
1912 |
root |
1.85 |
JSON json; |
1913 |
|
|
json_init (&json); |
1914 |
|
|
json.flags |= ix; |
1915 |
root |
1.47 |
XPUSHs (encode_json (scalar, &json)); |
1916 |
|
|
} |
1917 |
root |
1.2 |
|
1918 |
root |
1.69 |
void decode_json (SV *jsonstr) |
1919 |
root |
1.76 |
ALIAS: |
1920 |
|
|
from_json_ = 0 |
1921 |
|
|
decode_json = F_UTF8 |
1922 |
root |
1.2 |
PPCODE: |
1923 |
root |
1.47 |
{ |
1924 |
root |
1.85 |
JSON json; |
1925 |
|
|
json_init (&json); |
1926 |
|
|
json.flags |= ix; |
1927 |
root |
1.47 |
XPUSHs (decode_json (jsonstr, &json, 0)); |
1928 |
|
|
} |
1929 |
root |
1.1 |
|
1930 |
root |
1.77 |
|