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 |
1.42 |
// guarentees, 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 |
1.75 |
#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2) |
23 |
|
|
|
24 |
root |
1.44 |
#define F_ASCII 0x00000001UL |
25 |
|
|
#define F_LATIN1 0x00000002UL |
26 |
|
|
#define F_UTF8 0x00000004UL |
27 |
|
|
#define F_INDENT 0x00000008UL |
28 |
|
|
#define F_CANONICAL 0x00000010UL |
29 |
|
|
#define F_SPACE_BEFORE 0x00000020UL |
30 |
|
|
#define F_SPACE_AFTER 0x00000040UL |
31 |
|
|
#define F_ALLOW_NONREF 0x00000100UL |
32 |
|
|
#define F_SHRINK 0x00000200UL |
33 |
|
|
#define F_ALLOW_BLESSED 0x00000400UL |
34 |
root |
1.49 |
#define F_CONV_BLESSED 0x00000800UL |
35 |
root |
1.63 |
#define F_RELAXED 0x00001000UL |
36 |
|
|
|
37 |
root |
1.44 |
#define F_MAXDEPTH 0xf8000000UL |
38 |
|
|
#define S_MAXDEPTH 27 |
39 |
root |
1.45 |
#define F_MAXSIZE 0x01f00000UL |
40 |
|
|
#define S_MAXSIZE 20 |
41 |
root |
1.49 |
#define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
42 |
root |
1.18 |
|
43 |
|
|
#define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
44 |
root |
1.45 |
#define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE )) |
45 |
root |
1.18 |
|
46 |
root |
1.2 |
#define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
47 |
root |
1.24 |
#define F_DEFAULT (9UL << S_MAXDEPTH) |
48 |
root |
1.1 |
|
49 |
|
|
#define INIT_SIZE 32 // initial scalar size to be allocated |
50 |
root |
1.12 |
#define INDENT_STEP 3 // spaces per indentation level |
51 |
|
|
|
52 |
root |
1.38 |
#define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
53 |
root |
1.1 |
|
54 |
|
|
#define SB do { |
55 |
|
|
#define SE } while (0) |
56 |
|
|
|
57 |
root |
1.35 |
#if __GNUC__ >= 3 |
58 |
root |
1.71 |
# define expect(expr,value) __builtin_expect ((expr), (value)) |
59 |
|
|
# define INLINE static inline |
60 |
root |
1.35 |
#else |
61 |
|
|
# define expect(expr,value) (expr) |
62 |
root |
1.71 |
# define INLINE static |
63 |
root |
1.35 |
#endif |
64 |
|
|
|
65 |
|
|
#define expect_false(expr) expect ((expr) != 0, 0) |
66 |
|
|
#define expect_true(expr) expect ((expr) != 0, 1) |
67 |
|
|
|
68 |
root |
1.72 |
#define IN_RANGE_INC(type,val,beg,end) \ |
69 |
|
|
((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
70 |
|
|
<= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
71 |
|
|
|
72 |
root |
1.57 |
#ifdef USE_ITHREADS |
73 |
|
|
# define JSON_SLOW 1 |
74 |
root |
1.59 |
# define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
75 |
root |
1.57 |
#else |
76 |
root |
1.58 |
# define JSON_SLOW 0 |
77 |
root |
1.59 |
# define JSON_STASH json_stash |
78 |
root |
1.57 |
#endif |
79 |
|
|
|
80 |
root |
1.44 |
static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
81 |
root |
1.43 |
static SV *json_true, *json_false; |
82 |
root |
1.1 |
|
83 |
root |
1.48 |
typedef struct { |
84 |
root |
1.47 |
U32 flags; |
85 |
root |
1.52 |
SV *cb_object; |
86 |
|
|
HV *cb_sk_object; |
87 |
root |
1.48 |
} JSON; |
88 |
root |
1.47 |
|
89 |
root |
1.12 |
///////////////////////////////////////////////////////////////////////////// |
90 |
|
|
// utility functions |
91 |
root |
1.1 |
|
92 |
root |
1.71 |
INLINE void |
93 |
root |
1.7 |
shrink (SV *sv) |
94 |
|
|
{ |
95 |
|
|
sv_utf8_downgrade (sv, 1); |
96 |
root |
1.12 |
if (SvLEN (sv) > SvCUR (sv) + 1) |
97 |
|
|
{ |
98 |
root |
1.7 |
#ifdef SvPV_shrink_to_cur |
99 |
root |
1.12 |
SvPV_shrink_to_cur (sv); |
100 |
|
|
#elif defined (SvPV_renew) |
101 |
|
|
SvPV_renew (sv, SvCUR (sv) + 1); |
102 |
root |
1.7 |
#endif |
103 |
root |
1.12 |
} |
104 |
root |
1.7 |
} |
105 |
|
|
|
106 |
root |
1.13 |
// decode an utf-8 character and return it, or (UV)-1 in |
107 |
|
|
// case of an error. |
108 |
|
|
// we special-case "safe" characters from U+80 .. U+7FF, |
109 |
|
|
// but use the very good perl function to parse anything else. |
110 |
|
|
// note that we never call this function for a ascii codepoints |
111 |
root |
1.71 |
INLINE UV |
112 |
root |
1.13 |
decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
113 |
|
|
{ |
114 |
root |
1.72 |
if (expect_true (len >= 2 |
115 |
|
|
&& IN_RANGE_INC (char, s[0], 0xc2, 0xdf) |
116 |
|
|
&& IN_RANGE_INC (char, s[1], 0x80, 0xbf))) |
117 |
root |
1.13 |
{ |
118 |
|
|
*clen = 2; |
119 |
|
|
return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
120 |
|
|
} |
121 |
|
|
else |
122 |
root |
1.72 |
return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
123 |
|
|
} |
124 |
|
|
|
125 |
|
|
// likewise for encoding, also never called for ascii codepoints |
126 |
|
|
// this function takes advantage of this fact, although current gccs |
127 |
|
|
// seem to optimise the check for >= 0x80 away anyways |
128 |
|
|
INLINE unsigned char * |
129 |
|
|
encode_utf8 (unsigned char *s, UV ch) |
130 |
|
|
{ |
131 |
root |
1.74 |
if (expect_false (ch < 0x000080)) |
132 |
|
|
*s++ = ch; |
133 |
|
|
else if (expect_true (ch < 0x000800)) |
134 |
|
|
*s++ = 0xc0 | ( ch >> 6), |
135 |
|
|
*s++ = 0x80 | ( ch & 0x3f); |
136 |
|
|
else if ( ch < 0x010000) |
137 |
|
|
*s++ = 0xe0 | ( ch >> 12), |
138 |
|
|
*s++ = 0x80 | ((ch >> 6) & 0x3f), |
139 |
|
|
*s++ = 0x80 | ( ch & 0x3f); |
140 |
|
|
else if ( ch < 0x110000) |
141 |
|
|
*s++ = 0xf0 | ( ch >> 18), |
142 |
|
|
*s++ = 0x80 | ((ch >> 12) & 0x3f), |
143 |
|
|
*s++ = 0x80 | ((ch >> 6) & 0x3f), |
144 |
|
|
*s++ = 0x80 | ( ch & 0x3f); |
145 |
root |
1.72 |
|
146 |
|
|
return s; |
147 |
root |
1.13 |
} |
148 |
|
|
|
149 |
root |
1.1 |
///////////////////////////////////////////////////////////////////////////// |
150 |
root |
1.12 |
// encoder |
151 |
|
|
|
152 |
|
|
// structure used for encoding JSON |
153 |
|
|
typedef struct |
154 |
|
|
{ |
155 |
|
|
char *cur; // SvPVX (sv) + current output position |
156 |
|
|
char *end; // SvEND (sv) |
157 |
|
|
SV *sv; // result scalar |
158 |
root |
1.48 |
JSON json; |
159 |
root |
1.18 |
U32 indent; // indentation level |
160 |
|
|
U32 maxdepth; // max. indentation/recursion level |
161 |
root |
1.71 |
UV limit; // escape character values >= this value when encoding |
162 |
root |
1.12 |
} enc_t; |
163 |
root |
1.1 |
|
164 |
root |
1.71 |
INLINE void |
165 |
root |
1.1 |
need (enc_t *enc, STRLEN len) |
166 |
|
|
{ |
167 |
root |
1.35 |
if (expect_false (enc->cur + len >= enc->end)) |
168 |
root |
1.1 |
{ |
169 |
|
|
STRLEN cur = enc->cur - SvPVX (enc->sv); |
170 |
|
|
SvGROW (enc->sv, cur + len + 1); |
171 |
|
|
enc->cur = SvPVX (enc->sv) + cur; |
172 |
root |
1.27 |
enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
173 |
root |
1.1 |
} |
174 |
|
|
} |
175 |
|
|
|
176 |
root |
1.71 |
INLINE void |
177 |
root |
1.1 |
encode_ch (enc_t *enc, char ch) |
178 |
|
|
{ |
179 |
|
|
need (enc, 1); |
180 |
|
|
*enc->cur++ = ch; |
181 |
|
|
} |
182 |
|
|
|
183 |
|
|
static void |
184 |
|
|
encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) |
185 |
|
|
{ |
186 |
|
|
char *end = str + len; |
187 |
|
|
|
188 |
root |
1.4 |
need (enc, len); |
189 |
|
|
|
190 |
root |
1.1 |
while (str < end) |
191 |
|
|
{ |
192 |
|
|
unsigned char ch = *(unsigned char *)str; |
193 |
root |
1.4 |
|
194 |
root |
1.35 |
if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
195 |
root |
1.4 |
{ |
196 |
root |
1.35 |
if (expect_false (ch == '"')) // but with slow exceptions |
197 |
root |
1.6 |
{ |
198 |
|
|
need (enc, len += 1); |
199 |
|
|
*enc->cur++ = '\\'; |
200 |
|
|
*enc->cur++ = '"'; |
201 |
|
|
} |
202 |
root |
1.35 |
else if (expect_false (ch == '\\')) |
203 |
root |
1.6 |
{ |
204 |
|
|
need (enc, len += 1); |
205 |
|
|
*enc->cur++ = '\\'; |
206 |
|
|
*enc->cur++ = '\\'; |
207 |
|
|
} |
208 |
|
|
else |
209 |
|
|
*enc->cur++ = ch; |
210 |
|
|
|
211 |
root |
1.4 |
++str; |
212 |
root |
1.1 |
} |
213 |
|
|
else |
214 |
|
|
{ |
215 |
root |
1.6 |
switch (ch) |
216 |
root |
1.1 |
{ |
217 |
root |
1.6 |
case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; |
218 |
|
|
case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; |
219 |
|
|
case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; |
220 |
|
|
case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; |
221 |
|
|
case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; |
222 |
root |
1.1 |
|
223 |
root |
1.6 |
default: |
224 |
root |
1.1 |
{ |
225 |
root |
1.6 |
STRLEN clen; |
226 |
|
|
UV uch; |
227 |
|
|
|
228 |
|
|
if (is_utf8) |
229 |
|
|
{ |
230 |
root |
1.13 |
uch = decode_utf8 (str, end - str, &clen); |
231 |
root |
1.6 |
if (clen == (STRLEN)-1) |
232 |
root |
1.9 |
croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
233 |
root |
1.6 |
} |
234 |
|
|
else |
235 |
|
|
{ |
236 |
|
|
uch = ch; |
237 |
|
|
clen = 1; |
238 |
|
|
} |
239 |
|
|
|
240 |
root |
1.72 |
if (uch < 0x80/*0x20*/ || uch >= enc->limit) |
241 |
root |
1.6 |
{ |
242 |
root |
1.74 |
if (uch >= 0x10000UL) |
243 |
root |
1.6 |
{ |
244 |
root |
1.74 |
if (uch >= 0x110000UL) |
245 |
root |
1.71 |
croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
246 |
|
|
|
247 |
root |
1.6 |
need (enc, len += 11); |
248 |
|
|
sprintf (enc->cur, "\\u%04x\\u%04x", |
249 |
root |
1.10 |
(int)((uch - 0x10000) / 0x400 + 0xD800), |
250 |
|
|
(int)((uch - 0x10000) % 0x400 + 0xDC00)); |
251 |
root |
1.6 |
enc->cur += 12; |
252 |
|
|
} |
253 |
|
|
else |
254 |
|
|
{ |
255 |
|
|
static char hexdigit [16] = "0123456789abcdef"; |
256 |
|
|
need (enc, len += 5); |
257 |
|
|
*enc->cur++ = '\\'; |
258 |
|
|
*enc->cur++ = 'u'; |
259 |
|
|
*enc->cur++ = hexdigit [ uch >> 12 ]; |
260 |
|
|
*enc->cur++ = hexdigit [(uch >> 8) & 15]; |
261 |
|
|
*enc->cur++ = hexdigit [(uch >> 4) & 15]; |
262 |
|
|
*enc->cur++ = hexdigit [(uch >> 0) & 15]; |
263 |
|
|
} |
264 |
root |
1.4 |
|
265 |
root |
1.6 |
str += clen; |
266 |
|
|
} |
267 |
root |
1.47 |
else if (enc->json.flags & F_LATIN1) |
268 |
root |
1.30 |
{ |
269 |
|
|
*enc->cur++ = uch; |
270 |
|
|
str += clen; |
271 |
|
|
} |
272 |
root |
1.6 |
else if (is_utf8) |
273 |
|
|
{ |
274 |
|
|
need (enc, len += clen); |
275 |
|
|
do |
276 |
|
|
{ |
277 |
|
|
*enc->cur++ = *str++; |
278 |
|
|
} |
279 |
|
|
while (--clen); |
280 |
|
|
} |
281 |
|
|
else |
282 |
|
|
{ |
283 |
root |
1.28 |
need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
284 |
root |
1.72 |
enc->cur = encode_utf8 (enc->cur, uch); |
285 |
root |
1.6 |
++str; |
286 |
|
|
} |
287 |
root |
1.5 |
} |
288 |
root |
1.4 |
} |
289 |
root |
1.1 |
} |
290 |
|
|
|
291 |
|
|
--len; |
292 |
|
|
} |
293 |
|
|
} |
294 |
|
|
|
295 |
root |
1.71 |
INLINE void |
296 |
root |
1.12 |
encode_indent (enc_t *enc) |
297 |
|
|
{ |
298 |
root |
1.47 |
if (enc->json.flags & F_INDENT) |
299 |
root |
1.12 |
{ |
300 |
|
|
int spaces = enc->indent * INDENT_STEP; |
301 |
|
|
|
302 |
|
|
need (enc, spaces); |
303 |
|
|
memset (enc->cur, ' ', spaces); |
304 |
|
|
enc->cur += spaces; |
305 |
|
|
} |
306 |
|
|
} |
307 |
|
|
|
308 |
root |
1.71 |
INLINE void |
309 |
root |
1.12 |
encode_space (enc_t *enc) |
310 |
|
|
{ |
311 |
|
|
need (enc, 1); |
312 |
|
|
encode_ch (enc, ' '); |
313 |
|
|
} |
314 |
|
|
|
315 |
root |
1.71 |
INLINE void |
316 |
root |
1.12 |
encode_nl (enc_t *enc) |
317 |
|
|
{ |
318 |
root |
1.47 |
if (enc->json.flags & F_INDENT) |
319 |
root |
1.12 |
{ |
320 |
|
|
need (enc, 1); |
321 |
|
|
encode_ch (enc, '\n'); |
322 |
|
|
} |
323 |
|
|
} |
324 |
|
|
|
325 |
root |
1.71 |
INLINE void |
326 |
root |
1.12 |
encode_comma (enc_t *enc) |
327 |
|
|
{ |
328 |
|
|
encode_ch (enc, ','); |
329 |
root |
1.1 |
|
330 |
root |
1.47 |
if (enc->json.flags & F_INDENT) |
331 |
root |
1.12 |
encode_nl (enc); |
332 |
root |
1.47 |
else if (enc->json.flags & F_SPACE_AFTER) |
333 |
root |
1.12 |
encode_space (enc); |
334 |
|
|
} |
335 |
root |
1.1 |
|
336 |
|
|
static void encode_sv (enc_t *enc, SV *sv); |
337 |
|
|
|
338 |
|
|
static void |
339 |
|
|
encode_av (enc_t *enc, AV *av) |
340 |
|
|
{ |
341 |
|
|
int i, len = av_len (av); |
342 |
|
|
|
343 |
root |
1.21 |
if (enc->indent >= enc->maxdepth) |
344 |
|
|
croak ("data structure too deep (hit recursion limit)"); |
345 |
|
|
|
346 |
root |
1.65 |
encode_ch (enc, '['); |
347 |
|
|
|
348 |
|
|
if (len >= 0) |
349 |
root |
1.1 |
{ |
350 |
root |
1.65 |
encode_nl (enc); ++enc->indent; |
351 |
root |
1.55 |
|
352 |
root |
1.65 |
for (i = 0; i <= len; ++i) |
353 |
|
|
{ |
354 |
|
|
SV **svp = av_fetch (av, i, 0); |
355 |
root |
1.55 |
|
356 |
root |
1.65 |
encode_indent (enc); |
357 |
root |
1.1 |
|
358 |
root |
1.65 |
if (svp) |
359 |
|
|
encode_sv (enc, *svp); |
360 |
|
|
else |
361 |
|
|
encode_str (enc, "null", 4, 0); |
362 |
root |
1.1 |
|
363 |
root |
1.65 |
if (i < len) |
364 |
|
|
encode_comma (enc); |
365 |
|
|
} |
366 |
root |
1.1 |
|
367 |
root |
1.65 |
encode_nl (enc); --enc->indent; encode_indent (enc); |
368 |
|
|
} |
369 |
|
|
|
370 |
|
|
encode_ch (enc, ']'); |
371 |
root |
1.1 |
} |
372 |
|
|
|
373 |
|
|
static void |
374 |
root |
1.61 |
encode_hk (enc_t *enc, HE *he) |
375 |
root |
1.1 |
{ |
376 |
|
|
encode_ch (enc, '"'); |
377 |
|
|
|
378 |
|
|
if (HeKLEN (he) == HEf_SVKEY) |
379 |
|
|
{ |
380 |
|
|
SV *sv = HeSVKEY (he); |
381 |
|
|
STRLEN len; |
382 |
root |
1.4 |
char *str; |
383 |
|
|
|
384 |
|
|
SvGETMAGIC (sv); |
385 |
|
|
str = SvPV (sv, len); |
386 |
root |
1.1 |
|
387 |
|
|
encode_str (enc, str, len, SvUTF8 (sv)); |
388 |
|
|
} |
389 |
|
|
else |
390 |
|
|
encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
391 |
|
|
|
392 |
|
|
encode_ch (enc, '"'); |
393 |
|
|
|
394 |
root |
1.47 |
if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); |
395 |
root |
1.1 |
encode_ch (enc, ':'); |
396 |
root |
1.47 |
if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); |
397 |
root |
1.1 |
} |
398 |
|
|
|
399 |
|
|
// compare hash entries, used when all keys are bytestrings |
400 |
|
|
static int |
401 |
|
|
he_cmp_fast (const void *a_, const void *b_) |
402 |
|
|
{ |
403 |
|
|
int cmp; |
404 |
|
|
|
405 |
|
|
HE *a = *(HE **)a_; |
406 |
|
|
HE *b = *(HE **)b_; |
407 |
|
|
|
408 |
|
|
STRLEN la = HeKLEN (a); |
409 |
|
|
STRLEN lb = HeKLEN (b); |
410 |
|
|
|
411 |
root |
1.61 |
if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la))) |
412 |
|
|
cmp = lb - la; |
413 |
root |
1.1 |
|
414 |
|
|
return cmp; |
415 |
|
|
} |
416 |
|
|
|
417 |
|
|
// compare hash entries, used when some keys are sv's or utf-x |
418 |
|
|
static int |
419 |
|
|
he_cmp_slow (const void *a, const void *b) |
420 |
|
|
{ |
421 |
root |
1.61 |
return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a)); |
422 |
root |
1.1 |
} |
423 |
|
|
|
424 |
|
|
static void |
425 |
|
|
encode_hv (enc_t *enc, HV *hv) |
426 |
|
|
{ |
427 |
root |
1.61 |
HE *he; |
428 |
|
|
int count; |
429 |
root |
1.1 |
|
430 |
root |
1.21 |
if (enc->indent >= enc->maxdepth) |
431 |
|
|
croak ("data structure too deep (hit recursion limit)"); |
432 |
|
|
|
433 |
root |
1.65 |
encode_ch (enc, '{'); |
434 |
root |
1.1 |
|
435 |
root |
1.61 |
// for canonical output we have to sort by keys first |
436 |
|
|
// actually, this is mostly due to the stupid so-called |
437 |
|
|
// security workaround added somewhere in 5.8.x. |
438 |
|
|
// that randomises hash orderings |
439 |
|
|
if (enc->json.flags & F_CANONICAL) |
440 |
root |
1.1 |
{ |
441 |
root |
1.61 |
int count = hv_iterinit (hv); |
442 |
|
|
|
443 |
|
|
if (SvMAGICAL (hv)) |
444 |
|
|
{ |
445 |
|
|
// need to count by iterating. could improve by dynamically building the vector below |
446 |
|
|
// but I don't care for the speed of this special case. |
447 |
|
|
// note also that we will run into undefined behaviour when the two iterations |
448 |
|
|
// do not result in the same count, something I might care for in some later release. |
449 |
|
|
|
450 |
|
|
count = 0; |
451 |
|
|
while (hv_iternext (hv)) |
452 |
|
|
++count; |
453 |
|
|
|
454 |
|
|
hv_iterinit (hv); |
455 |
|
|
} |
456 |
|
|
|
457 |
|
|
if (count) |
458 |
root |
1.1 |
{ |
459 |
root |
1.61 |
int i, fast = 1; |
460 |
root |
1.34 |
#if defined(__BORLANDC__) || defined(_MSC_VER) |
461 |
root |
1.33 |
HE **hes = _alloca (count * sizeof (HE)); |
462 |
|
|
#else |
463 |
|
|
HE *hes [count]; // if your compiler dies here, you need to enable C99 mode |
464 |
|
|
#endif |
465 |
root |
1.1 |
|
466 |
|
|
i = 0; |
467 |
|
|
while ((he = hv_iternext (hv))) |
468 |
|
|
{ |
469 |
|
|
hes [i++] = he; |
470 |
|
|
if (HeKLEN (he) < 0 || HeKUTF8 (he)) |
471 |
|
|
fast = 0; |
472 |
|
|
} |
473 |
|
|
|
474 |
|
|
assert (i == count); |
475 |
|
|
|
476 |
|
|
if (fast) |
477 |
|
|
qsort (hes, count, sizeof (HE *), he_cmp_fast); |
478 |
|
|
else |
479 |
|
|
{ |
480 |
root |
1.8 |
// hack to forcefully disable "use bytes" |
481 |
|
|
COP cop = *PL_curcop; |
482 |
root |
1.1 |
cop.op_private = 0; |
483 |
root |
1.8 |
|
484 |
|
|
ENTER; |
485 |
|
|
SAVETMPS; |
486 |
|
|
|
487 |
|
|
SAVEVPTR (PL_curcop); |
488 |
root |
1.1 |
PL_curcop = &cop; |
489 |
|
|
|
490 |
|
|
qsort (hes, count, sizeof (HE *), he_cmp_slow); |
491 |
root |
1.8 |
|
492 |
root |
1.1 |
FREETMPS; |
493 |
root |
1.8 |
LEAVE; |
494 |
root |
1.1 |
} |
495 |
|
|
|
496 |
root |
1.65 |
encode_nl (enc); ++enc->indent; |
497 |
|
|
|
498 |
root |
1.61 |
while (count--) |
499 |
root |
1.1 |
{ |
500 |
root |
1.12 |
encode_indent (enc); |
501 |
root |
1.61 |
he = hes [count]; |
502 |
|
|
encode_hk (enc, he); |
503 |
root |
1.62 |
encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
504 |
root |
1.1 |
|
505 |
root |
1.61 |
if (count) |
506 |
root |
1.12 |
encode_comma (enc); |
507 |
root |
1.1 |
} |
508 |
root |
1.65 |
|
509 |
|
|
encode_nl (enc); --enc->indent; encode_indent (enc); |
510 |
root |
1.1 |
} |
511 |
root |
1.61 |
} |
512 |
|
|
else |
513 |
|
|
{ |
514 |
|
|
if (hv_iterinit (hv) || SvMAGICAL (hv)) |
515 |
|
|
if ((he = hv_iternext (hv))) |
516 |
root |
1.65 |
{ |
517 |
|
|
encode_nl (enc); ++enc->indent; |
518 |
|
|
|
519 |
|
|
for (;;) |
520 |
|
|
{ |
521 |
|
|
encode_indent (enc); |
522 |
|
|
encode_hk (enc, he); |
523 |
|
|
encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
524 |
|
|
|
525 |
|
|
if (!(he = hv_iternext (hv))) |
526 |
|
|
break; |
527 |
root |
1.1 |
|
528 |
root |
1.65 |
encode_comma (enc); |
529 |
|
|
} |
530 |
root |
1.1 |
|
531 |
root |
1.65 |
encode_nl (enc); --enc->indent; encode_indent (enc); |
532 |
|
|
} |
533 |
root |
1.61 |
} |
534 |
root |
1.1 |
|
535 |
root |
1.65 |
encode_ch (enc, '}'); |
536 |
root |
1.1 |
} |
537 |
|
|
|
538 |
root |
1.21 |
// encode objects, arrays and special \0=false and \1=true values. |
539 |
|
|
static void |
540 |
|
|
encode_rv (enc_t *enc, SV *sv) |
541 |
|
|
{ |
542 |
root |
1.25 |
svtype svt; |
543 |
|
|
|
544 |
root |
1.21 |
SvGETMAGIC (sv); |
545 |
root |
1.25 |
svt = SvTYPE (sv); |
546 |
root |
1.21 |
|
547 |
root |
1.44 |
if (expect_false (SvOBJECT (sv))) |
548 |
|
|
{ |
549 |
root |
1.57 |
HV *stash = !JSON_SLOW || json_boolean_stash |
550 |
|
|
? json_boolean_stash |
551 |
|
|
: gv_stashpv ("JSON::XS::Boolean", 1); |
552 |
|
|
|
553 |
|
|
if (SvSTASH (sv) == stash) |
554 |
root |
1.44 |
{ |
555 |
root |
1.51 |
if (SvIV (sv)) |
556 |
|
|
encode_str (enc, "true", 4, 0); |
557 |
|
|
else |
558 |
root |
1.44 |
encode_str (enc, "false", 5, 0); |
559 |
|
|
} |
560 |
|
|
else |
561 |
|
|
{ |
562 |
|
|
#if 0 |
563 |
|
|
if (0 && sv_derived_from (rv, "JSON::Literal")) |
564 |
|
|
{ |
565 |
|
|
// not yet |
566 |
|
|
} |
567 |
|
|
#endif |
568 |
root |
1.47 |
if (enc->json.flags & F_CONV_BLESSED) |
569 |
root |
1.44 |
{ |
570 |
|
|
// we re-bless the reference to get overload and other niceties right |
571 |
root |
1.53 |
GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0); |
572 |
root |
1.44 |
|
573 |
|
|
if (to_json) |
574 |
|
|
{ |
575 |
root |
1.53 |
dSP; |
576 |
|
|
|
577 |
|
|
ENTER; SAVETMPS; PUSHMARK (SP); |
578 |
root |
1.44 |
XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); |
579 |
|
|
|
580 |
root |
1.53 |
// calling with G_SCALAR ensures that we always get a 1 return value |
581 |
root |
1.44 |
PUTBACK; |
582 |
root |
1.54 |
call_sv ((SV *)GvCV (to_json), G_SCALAR); |
583 |
root |
1.44 |
SPAGAIN; |
584 |
|
|
|
585 |
root |
1.53 |
// catch this surprisingly common error |
586 |
|
|
if (SvROK (TOPs) && SvRV (TOPs) == sv) |
587 |
|
|
croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
588 |
|
|
|
589 |
|
|
sv = POPs; |
590 |
|
|
PUTBACK; |
591 |
|
|
|
592 |
|
|
encode_sv (enc, sv); |
593 |
root |
1.44 |
|
594 |
root |
1.49 |
FREETMPS; LEAVE; |
595 |
root |
1.44 |
} |
596 |
root |
1.47 |
else if (enc->json.flags & F_ALLOW_BLESSED) |
597 |
root |
1.44 |
encode_str (enc, "null", 4, 0); |
598 |
|
|
else |
599 |
|
|
croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it", |
600 |
|
|
SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
601 |
|
|
} |
602 |
root |
1.47 |
else if (enc->json.flags & F_ALLOW_BLESSED) |
603 |
root |
1.44 |
encode_str (enc, "null", 4, 0); |
604 |
|
|
else |
605 |
|
|
croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", |
606 |
|
|
SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
607 |
|
|
} |
608 |
|
|
} |
609 |
|
|
else if (svt == SVt_PVHV) |
610 |
root |
1.21 |
encode_hv (enc, (HV *)sv); |
611 |
|
|
else if (svt == SVt_PVAV) |
612 |
|
|
encode_av (enc, (AV *)sv); |
613 |
|
|
else if (svt < SVt_PVAV) |
614 |
|
|
{ |
615 |
root |
1.50 |
STRLEN len = 0; |
616 |
|
|
char *pv = svt ? SvPV (sv, len) : 0; |
617 |
|
|
|
618 |
root |
1.51 |
if (len == 1 && *pv == '1') |
619 |
|
|
encode_str (enc, "true", 4, 0); |
620 |
|
|
else if (len == 1 && *pv == '0') |
621 |
root |
1.21 |
encode_str (enc, "false", 5, 0); |
622 |
|
|
else |
623 |
|
|
croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
624 |
|
|
SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
625 |
|
|
} |
626 |
|
|
else |
627 |
|
|
croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
628 |
|
|
SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
629 |
|
|
} |
630 |
|
|
|
631 |
root |
1.1 |
static void |
632 |
|
|
encode_sv (enc_t *enc, SV *sv) |
633 |
|
|
{ |
634 |
root |
1.4 |
SvGETMAGIC (sv); |
635 |
|
|
|
636 |
root |
1.1 |
if (SvPOKp (sv)) |
637 |
|
|
{ |
638 |
|
|
STRLEN len; |
639 |
|
|
char *str = SvPV (sv, len); |
640 |
|
|
encode_ch (enc, '"'); |
641 |
|
|
encode_str (enc, str, len, SvUTF8 (sv)); |
642 |
|
|
encode_ch (enc, '"'); |
643 |
|
|
} |
644 |
|
|
else if (SvNOKp (sv)) |
645 |
|
|
{ |
646 |
root |
1.39 |
// trust that perl will do the right thing w.r.t. JSON syntax. |
647 |
root |
1.1 |
need (enc, NV_DIG + 32); |
648 |
|
|
Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
649 |
|
|
enc->cur += strlen (enc->cur); |
650 |
|
|
} |
651 |
|
|
else if (SvIOKp (sv)) |
652 |
|
|
{ |
653 |
root |
1.74 |
// we assume we can always read an IV as a UV and vice versa |
654 |
|
|
// we assume two's complement |
655 |
|
|
// we assume no aliasing issues in the union |
656 |
root |
1.75 |
if (SvIsUV (sv) ? SvUVX (sv) <= 59000 |
657 |
|
|
: SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000) |
658 |
root |
1.35 |
{ |
659 |
|
|
// optimise the "small number case" |
660 |
|
|
// code will likely be branchless and use only a single multiplication |
661 |
root |
1.74 |
// works for numbers up to 59074 |
662 |
|
|
I32 i = SvIVX (sv); |
663 |
root |
1.35 |
U32 u; |
664 |
root |
1.39 |
char digit, nz = 0; |
665 |
root |
1.35 |
|
666 |
|
|
need (enc, 6); |
667 |
|
|
|
668 |
|
|
*enc->cur = '-'; enc->cur += i < 0 ? 1 : 0; |
669 |
|
|
u = i < 0 ? -i : i; |
670 |
|
|
|
671 |
|
|
// convert to 4.28 fixed-point representation |
672 |
|
|
u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits |
673 |
|
|
|
674 |
root |
1.39 |
// now output digit by digit, each time masking out the integer part |
675 |
|
|
// and multiplying by 5 while moving the decimal point one to the right, |
676 |
|
|
// resulting in a net multiplication by 10. |
677 |
|
|
// we always write the digit to memory but conditionally increment |
678 |
root |
1.75 |
// the pointer, to enable the use of conditional move instructions. |
679 |
|
|
digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5; |
680 |
|
|
digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5; |
681 |
|
|
digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5; |
682 |
|
|
digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5; |
683 |
root |
1.39 |
digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
684 |
root |
1.35 |
} |
685 |
root |
1.75 |
else |
686 |
|
|
{ |
687 |
|
|
// large integer, use the (rather slow) snprintf way. |
688 |
|
|
need (enc, IVUV_MAXCHARS); |
689 |
|
|
enc->cur += |
690 |
|
|
SvIsUV(sv) |
691 |
|
|
? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) |
692 |
|
|
: snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); |
693 |
|
|
} |
694 |
root |
1.1 |
} |
695 |
|
|
else if (SvROK (sv)) |
696 |
root |
1.21 |
encode_rv (enc, SvRV (sv)); |
697 |
root |
1.1 |
else if (!SvOK (sv)) |
698 |
|
|
encode_str (enc, "null", 4, 0); |
699 |
|
|
else |
700 |
root |
1.9 |
croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
701 |
|
|
SvPV_nolen (sv), SvFLAGS (sv)); |
702 |
root |
1.1 |
} |
703 |
|
|
|
704 |
|
|
static SV * |
705 |
root |
1.48 |
encode_json (SV *scalar, JSON *json) |
706 |
root |
1.1 |
{ |
707 |
root |
1.25 |
enc_t enc; |
708 |
|
|
|
709 |
root |
1.47 |
if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
710 |
root |
1.9 |
croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
711 |
root |
1.3 |
|
712 |
root |
1.47 |
enc.json = *json; |
713 |
root |
1.12 |
enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
714 |
|
|
enc.cur = SvPVX (enc.sv); |
715 |
|
|
enc.end = SvEND (enc.sv); |
716 |
|
|
enc.indent = 0; |
717 |
root |
1.47 |
enc.maxdepth = DEC_DEPTH (enc.json.flags); |
718 |
root |
1.71 |
enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
719 |
|
|
: enc.json.flags & F_LATIN1 ? 0x000100UL |
720 |
root |
1.74 |
: 0x110000UL; |
721 |
root |
1.1 |
|
722 |
|
|
SvPOK_only (enc.sv); |
723 |
|
|
encode_sv (&enc, scalar); |
724 |
|
|
|
725 |
|
|
SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
726 |
root |
1.27 |
*SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
727 |
root |
1.6 |
|
728 |
root |
1.47 |
if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
729 |
root |
1.30 |
SvUTF8_on (enc.sv); |
730 |
|
|
|
731 |
root |
1.47 |
if (enc.json.flags & F_SHRINK) |
732 |
root |
1.7 |
shrink (enc.sv); |
733 |
|
|
|
734 |
root |
1.1 |
return enc.sv; |
735 |
|
|
} |
736 |
|
|
|
737 |
|
|
///////////////////////////////////////////////////////////////////////////// |
738 |
root |
1.12 |
// decoder |
739 |
root |
1.1 |
|
740 |
root |
1.12 |
// structure used for decoding JSON |
741 |
|
|
typedef struct |
742 |
|
|
{ |
743 |
|
|
char *cur; // current parser pointer |
744 |
|
|
char *end; // end of input string |
745 |
|
|
const char *err; // parse error, if != 0 |
746 |
root |
1.48 |
JSON json; |
747 |
root |
1.18 |
U32 depth; // recursion depth |
748 |
|
|
U32 maxdepth; // recursion depth limit |
749 |
root |
1.12 |
} dec_t; |
750 |
|
|
|
751 |
root |
1.71 |
INLINE void |
752 |
root |
1.64 |
decode_comment (dec_t *dec) |
753 |
|
|
{ |
754 |
|
|
// only '#'-style comments allowed a.t.m. |
755 |
|
|
|
756 |
|
|
while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) |
757 |
|
|
++dec->cur; |
758 |
|
|
} |
759 |
|
|
|
760 |
root |
1.71 |
INLINE void |
761 |
root |
1.12 |
decode_ws (dec_t *dec) |
762 |
|
|
{ |
763 |
|
|
for (;;) |
764 |
|
|
{ |
765 |
|
|
char ch = *dec->cur; |
766 |
|
|
|
767 |
root |
1.64 |
if (ch > 0x20) |
768 |
|
|
{ |
769 |
|
|
if (expect_false (ch == '#')) |
770 |
|
|
{ |
771 |
|
|
if (dec->json.flags & F_RELAXED) |
772 |
|
|
decode_comment (dec); |
773 |
|
|
else |
774 |
|
|
break; |
775 |
|
|
} |
776 |
|
|
else |
777 |
|
|
break; |
778 |
|
|
} |
779 |
|
|
else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09) |
780 |
|
|
break; // parse error, but let higher level handle it, gives better error messages |
781 |
root |
1.63 |
|
782 |
root |
1.12 |
++dec->cur; |
783 |
root |
1.1 |
} |
784 |
root |
1.12 |
} |
785 |
root |
1.1 |
|
786 |
|
|
#define ERR(reason) SB dec->err = reason; goto fail; SE |
787 |
root |
1.18 |
|
788 |
root |
1.1 |
#define EXPECT_CH(ch) SB \ |
789 |
|
|
if (*dec->cur != ch) \ |
790 |
|
|
ERR (# ch " expected"); \ |
791 |
|
|
++dec->cur; \ |
792 |
|
|
SE |
793 |
|
|
|
794 |
root |
1.18 |
#define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") |
795 |
|
|
#define DEC_DEC_DEPTH --dec->depth |
796 |
|
|
|
797 |
root |
1.1 |
static SV *decode_sv (dec_t *dec); |
798 |
|
|
|
799 |
|
|
static signed char decode_hexdigit[256]; |
800 |
|
|
|
801 |
|
|
static UV |
802 |
|
|
decode_4hex (dec_t *dec) |
803 |
|
|
{ |
804 |
|
|
signed char d1, d2, d3, d4; |
805 |
root |
1.12 |
unsigned char *cur = (unsigned char *)dec->cur; |
806 |
root |
1.1 |
|
807 |
root |
1.40 |
d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected"); |
808 |
|
|
d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected"); |
809 |
|
|
d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected"); |
810 |
|
|
d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected"); |
811 |
root |
1.1 |
|
812 |
|
|
dec->cur += 4; |
813 |
|
|
|
814 |
|
|
return ((UV)d1) << 12 |
815 |
|
|
| ((UV)d2) << 8 |
816 |
|
|
| ((UV)d3) << 4 |
817 |
|
|
| ((UV)d4); |
818 |
|
|
|
819 |
|
|
fail: |
820 |
|
|
return (UV)-1; |
821 |
|
|
} |
822 |
|
|
|
823 |
|
|
static SV * |
824 |
|
|
decode_str (dec_t *dec) |
825 |
|
|
{ |
826 |
root |
1.12 |
SV *sv = 0; |
827 |
root |
1.1 |
int utf8 = 0; |
828 |
root |
1.38 |
char *dec_cur = dec->cur; |
829 |
root |
1.1 |
|
830 |
root |
1.12 |
do |
831 |
root |
1.1 |
{ |
832 |
root |
1.28 |
char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; |
833 |
root |
1.12 |
char *cur = buf; |
834 |
root |
1.1 |
|
835 |
root |
1.12 |
do |
836 |
root |
1.1 |
{ |
837 |
root |
1.38 |
unsigned char ch = *(unsigned char *)dec_cur++; |
838 |
root |
1.12 |
|
839 |
root |
1.35 |
if (expect_false (ch == '"')) |
840 |
root |
1.12 |
{ |
841 |
root |
1.38 |
--dec_cur; |
842 |
root |
1.12 |
break; |
843 |
|
|
} |
844 |
root |
1.35 |
else if (expect_false (ch == '\\')) |
845 |
root |
1.1 |
{ |
846 |
root |
1.38 |
switch (*dec_cur) |
847 |
root |
1.1 |
{ |
848 |
root |
1.12 |
case '\\': |
849 |
|
|
case '/': |
850 |
root |
1.38 |
case '"': *cur++ = *dec_cur++; break; |
851 |
root |
1.12 |
|
852 |
root |
1.38 |
case 'b': ++dec_cur; *cur++ = '\010'; break; |
853 |
|
|
case 't': ++dec_cur; *cur++ = '\011'; break; |
854 |
|
|
case 'n': ++dec_cur; *cur++ = '\012'; break; |
855 |
|
|
case 'f': ++dec_cur; *cur++ = '\014'; break; |
856 |
|
|
case 'r': ++dec_cur; *cur++ = '\015'; break; |
857 |
root |
1.1 |
|
858 |
root |
1.12 |
case 'u': |
859 |
root |
1.1 |
{ |
860 |
root |
1.12 |
UV lo, hi; |
861 |
root |
1.38 |
++dec_cur; |
862 |
root |
1.1 |
|
863 |
root |
1.38 |
dec->cur = dec_cur; |
864 |
root |
1.12 |
hi = decode_4hex (dec); |
865 |
root |
1.38 |
dec_cur = dec->cur; |
866 |
root |
1.12 |
if (hi == (UV)-1) |
867 |
|
|
goto fail; |
868 |
root |
1.1 |
|
869 |
root |
1.12 |
// possibly a surrogate pair |
870 |
|
|
if (hi >= 0xd800) |
871 |
|
|
if (hi < 0xdc00) |
872 |
|
|
{ |
873 |
root |
1.38 |
if (dec_cur [0] != '\\' || dec_cur [1] != 'u') |
874 |
root |
1.12 |
ERR ("missing low surrogate character in surrogate pair"); |
875 |
|
|
|
876 |
root |
1.38 |
dec_cur += 2; |
877 |
root |
1.12 |
|
878 |
root |
1.38 |
dec->cur = dec_cur; |
879 |
root |
1.12 |
lo = decode_4hex (dec); |
880 |
root |
1.38 |
dec_cur = dec->cur; |
881 |
root |
1.12 |
if (lo == (UV)-1) |
882 |
|
|
goto fail; |
883 |
|
|
|
884 |
|
|
if (lo < 0xdc00 || lo >= 0xe000) |
885 |
|
|
ERR ("surrogate pair expected"); |
886 |
|
|
|
887 |
|
|
hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; |
888 |
|
|
} |
889 |
|
|
else if (hi < 0xe000) |
890 |
|
|
ERR ("missing high surrogate character in surrogate pair"); |
891 |
root |
1.1 |
|
892 |
root |
1.12 |
if (hi >= 0x80) |
893 |
|
|
{ |
894 |
|
|
utf8 = 1; |
895 |
root |
1.1 |
|
896 |
root |
1.72 |
cur = encode_utf8 (cur, hi); |
897 |
root |
1.12 |
} |
898 |
|
|
else |
899 |
|
|
*cur++ = hi; |
900 |
root |
1.1 |
} |
901 |
root |
1.12 |
break; |
902 |
|
|
|
903 |
|
|
default: |
904 |
root |
1.38 |
--dec_cur; |
905 |
root |
1.12 |
ERR ("illegal backslash escape sequence in string"); |
906 |
|
|
} |
907 |
|
|
} |
908 |
root |
1.74 |
else if (expect_true (ch >= 0x20 && ch < 0x80)) |
909 |
root |
1.12 |
*cur++ = ch; |
910 |
|
|
else if (ch >= 0x80) |
911 |
|
|
{ |
912 |
root |
1.25 |
STRLEN clen; |
913 |
|
|
UV uch; |
914 |
|
|
|
915 |
root |
1.38 |
--dec_cur; |
916 |
root |
1.1 |
|
917 |
root |
1.38 |
uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
918 |
root |
1.12 |
if (clen == (STRLEN)-1) |
919 |
|
|
ERR ("malformed UTF-8 character in JSON string"); |
920 |
root |
1.1 |
|
921 |
root |
1.12 |
do |
922 |
root |
1.38 |
*cur++ = *dec_cur++; |
923 |
root |
1.12 |
while (--clen); |
924 |
root |
1.5 |
|
925 |
root |
1.12 |
utf8 = 1; |
926 |
root |
1.1 |
} |
927 |
root |
1.12 |
else |
928 |
root |
1.23 |
{ |
929 |
root |
1.38 |
--dec_cur; |
930 |
root |
1.12 |
|
931 |
root |
1.23 |
if (!ch) |
932 |
root |
1.24 |
ERR ("unexpected end of string while parsing JSON string"); |
933 |
root |
1.23 |
else |
934 |
root |
1.24 |
ERR ("invalid character encountered while parsing JSON string"); |
935 |
root |
1.23 |
} |
936 |
root |
1.1 |
} |
937 |
root |
1.12 |
while (cur < buf + SHORT_STRING_LEN); |
938 |
root |
1.1 |
|
939 |
root |
1.25 |
{ |
940 |
|
|
STRLEN len = cur - buf; |
941 |
root |
1.5 |
|
942 |
root |
1.25 |
if (sv) |
943 |
|
|
{ |
944 |
|
|
SvGROW (sv, SvCUR (sv) + len + 1); |
945 |
|
|
memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
946 |
|
|
SvCUR_set (sv, SvCUR (sv) + len); |
947 |
|
|
} |
948 |
|
|
else |
949 |
|
|
sv = newSVpvn (buf, len); |
950 |
|
|
} |
951 |
root |
1.1 |
} |
952 |
root |
1.38 |
while (*dec_cur != '"'); |
953 |
root |
1.1 |
|
954 |
root |
1.38 |
++dec_cur; |
955 |
root |
1.1 |
|
956 |
root |
1.12 |
if (sv) |
957 |
|
|
{ |
958 |
|
|
SvPOK_only (sv); |
959 |
|
|
*SvEND (sv) = 0; |
960 |
root |
1.4 |
|
961 |
root |
1.12 |
if (utf8) |
962 |
|
|
SvUTF8_on (sv); |
963 |
|
|
} |
964 |
|
|
else |
965 |
|
|
sv = newSVpvn ("", 0); |
966 |
root |
1.6 |
|
967 |
root |
1.38 |
dec->cur = dec_cur; |
968 |
root |
1.1 |
return sv; |
969 |
|
|
|
970 |
|
|
fail: |
971 |
root |
1.38 |
dec->cur = dec_cur; |
972 |
root |
1.1 |
return 0; |
973 |
|
|
} |
974 |
|
|
|
975 |
|
|
static SV * |
976 |
|
|
decode_num (dec_t *dec) |
977 |
|
|
{ |
978 |
|
|
int is_nv = 0; |
979 |
|
|
char *start = dec->cur; |
980 |
|
|
|
981 |
|
|
// [minus] |
982 |
|
|
if (*dec->cur == '-') |
983 |
|
|
++dec->cur; |
984 |
|
|
|
985 |
|
|
if (*dec->cur == '0') |
986 |
|
|
{ |
987 |
|
|
++dec->cur; |
988 |
|
|
if (*dec->cur >= '0' && *dec->cur <= '9') |
989 |
|
|
ERR ("malformed number (leading zero must not be followed by another digit)"); |
990 |
|
|
} |
991 |
root |
1.5 |
else if (*dec->cur < '0' || *dec->cur > '9') |
992 |
|
|
ERR ("malformed number (no digits after initial minus)"); |
993 |
|
|
else |
994 |
|
|
do |
995 |
|
|
{ |
996 |
|
|
++dec->cur; |
997 |
|
|
} |
998 |
|
|
while (*dec->cur >= '0' && *dec->cur <= '9'); |
999 |
root |
1.1 |
|
1000 |
|
|
// [frac] |
1001 |
|
|
if (*dec->cur == '.') |
1002 |
|
|
{ |
1003 |
root |
1.5 |
++dec->cur; |
1004 |
|
|
|
1005 |
|
|
if (*dec->cur < '0' || *dec->cur > '9') |
1006 |
|
|
ERR ("malformed number (no digits after decimal point)"); |
1007 |
root |
1.1 |
|
1008 |
|
|
do |
1009 |
|
|
{ |
1010 |
|
|
++dec->cur; |
1011 |
|
|
} |
1012 |
|
|
while (*dec->cur >= '0' && *dec->cur <= '9'); |
1013 |
root |
1.5 |
|
1014 |
|
|
is_nv = 1; |
1015 |
root |
1.1 |
} |
1016 |
|
|
|
1017 |
|
|
// [exp] |
1018 |
|
|
if (*dec->cur == 'e' || *dec->cur == 'E') |
1019 |
|
|
{ |
1020 |
root |
1.5 |
++dec->cur; |
1021 |
root |
1.1 |
|
1022 |
|
|
if (*dec->cur == '-' || *dec->cur == '+') |
1023 |
|
|
++dec->cur; |
1024 |
|
|
|
1025 |
root |
1.5 |
if (*dec->cur < '0' || *dec->cur > '9') |
1026 |
|
|
ERR ("malformed number (no digits after exp sign)"); |
1027 |
|
|
|
1028 |
|
|
do |
1029 |
|
|
{ |
1030 |
|
|
++dec->cur; |
1031 |
|
|
} |
1032 |
|
|
while (*dec->cur >= '0' && *dec->cur <= '9'); |
1033 |
|
|
|
1034 |
|
|
is_nv = 1; |
1035 |
root |
1.1 |
} |
1036 |
|
|
|
1037 |
|
|
if (!is_nv) |
1038 |
|
|
{ |
1039 |
root |
1.56 |
int len = dec->cur - start; |
1040 |
|
|
|
1041 |
root |
1.70 |
// special case the rather common 1..5-digit-int case |
1042 |
root |
1.35 |
if (*start == '-') |
1043 |
root |
1.56 |
switch (len) |
1044 |
root |
1.1 |
{ |
1045 |
root |
1.70 |
case 2: return newSViv (-( start [1] - '0' * 1)); |
1046 |
|
|
case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
1047 |
|
|
case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
1048 |
|
|
case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
1049 |
|
|
case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
1050 |
root |
1.1 |
} |
1051 |
root |
1.35 |
else |
1052 |
root |
1.56 |
switch (len) |
1053 |
root |
1.35 |
{ |
1054 |
root |
1.70 |
case 1: return newSViv ( start [0] - '0' * 1); |
1055 |
|
|
case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1056 |
|
|
case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
1057 |
|
|
case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
1058 |
|
|
case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
1059 |
root |
1.35 |
} |
1060 |
|
|
|
1061 |
|
|
{ |
1062 |
|
|
UV uv; |
1063 |
root |
1.56 |
int numtype = grok_number (start, len, &uv); |
1064 |
root |
1.35 |
if (numtype & IS_NUMBER_IN_UV) |
1065 |
|
|
if (numtype & IS_NUMBER_NEG) |
1066 |
|
|
{ |
1067 |
|
|
if (uv < (UV)IV_MIN) |
1068 |
|
|
return newSViv (-(IV)uv); |
1069 |
|
|
} |
1070 |
|
|
else |
1071 |
|
|
return newSVuv (uv); |
1072 |
root |
1.56 |
} |
1073 |
|
|
|
1074 |
|
|
len -= *start == '-' ? 1 : 0; |
1075 |
root |
1.40 |
|
1076 |
root |
1.56 |
// does not fit into IV or UV, try NV |
1077 |
|
|
if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) |
1078 |
|
|
#if defined (LDBL_DIG) |
1079 |
|
|
|| (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len) |
1080 |
|
|
#endif |
1081 |
|
|
) |
1082 |
|
|
// fits into NV without loss of precision |
1083 |
|
|
return newSVnv (Atof (start)); |
1084 |
|
|
|
1085 |
|
|
// everything else fails, convert it to a string |
1086 |
|
|
return newSVpvn (start, dec->cur - start); |
1087 |
root |
1.1 |
} |
1088 |
|
|
|
1089 |
root |
1.56 |
// loss of precision here |
1090 |
root |
1.1 |
return newSVnv (Atof (start)); |
1091 |
|
|
|
1092 |
|
|
fail: |
1093 |
|
|
return 0; |
1094 |
|
|
} |
1095 |
|
|
|
1096 |
|
|
static SV * |
1097 |
|
|
decode_av (dec_t *dec) |
1098 |
|
|
{ |
1099 |
|
|
AV *av = newAV (); |
1100 |
|
|
|
1101 |
root |
1.18 |
DEC_INC_DEPTH; |
1102 |
root |
1.12 |
decode_ws (dec); |
1103 |
root |
1.18 |
|
1104 |
root |
1.5 |
if (*dec->cur == ']') |
1105 |
|
|
++dec->cur; |
1106 |
|
|
else |
1107 |
|
|
for (;;) |
1108 |
|
|
{ |
1109 |
|
|
SV *value; |
1110 |
root |
1.1 |
|
1111 |
root |
1.5 |
value = decode_sv (dec); |
1112 |
|
|
if (!value) |
1113 |
|
|
goto fail; |
1114 |
root |
1.1 |
|
1115 |
root |
1.5 |
av_push (av, value); |
1116 |
root |
1.1 |
|
1117 |
root |
1.12 |
decode_ws (dec); |
1118 |
root |
1.1 |
|
1119 |
root |
1.5 |
if (*dec->cur == ']') |
1120 |
|
|
{ |
1121 |
|
|
++dec->cur; |
1122 |
|
|
break; |
1123 |
|
|
} |
1124 |
|
|
|
1125 |
|
|
if (*dec->cur != ',') |
1126 |
|
|
ERR (", or ] expected while parsing array"); |
1127 |
root |
1.1 |
|
1128 |
root |
1.5 |
++dec->cur; |
1129 |
root |
1.63 |
|
1130 |
|
|
decode_ws (dec); |
1131 |
|
|
|
1132 |
|
|
if (*dec->cur == ']' && dec->json.flags & F_RELAXED) |
1133 |
|
|
{ |
1134 |
|
|
++dec->cur; |
1135 |
|
|
break; |
1136 |
|
|
} |
1137 |
root |
1.5 |
} |
1138 |
root |
1.1 |
|
1139 |
root |
1.18 |
DEC_DEC_DEPTH; |
1140 |
root |
1.1 |
return newRV_noinc ((SV *)av); |
1141 |
|
|
|
1142 |
|
|
fail: |
1143 |
|
|
SvREFCNT_dec (av); |
1144 |
root |
1.18 |
DEC_DEC_DEPTH; |
1145 |
root |
1.1 |
return 0; |
1146 |
|
|
} |
1147 |
|
|
|
1148 |
|
|
static SV * |
1149 |
|
|
decode_hv (dec_t *dec) |
1150 |
|
|
{ |
1151 |
root |
1.49 |
SV *sv; |
1152 |
root |
1.1 |
HV *hv = newHV (); |
1153 |
|
|
|
1154 |
root |
1.18 |
DEC_INC_DEPTH; |
1155 |
root |
1.12 |
decode_ws (dec); |
1156 |
root |
1.18 |
|
1157 |
root |
1.5 |
if (*dec->cur == '}') |
1158 |
|
|
++dec->cur; |
1159 |
|
|
else |
1160 |
|
|
for (;;) |
1161 |
|
|
{ |
1162 |
root |
1.63 |
EXPECT_CH ('"'); |
1163 |
root |
1.1 |
|
1164 |
root |
1.46 |
// heuristic: assume that |
1165 |
root |
1.47 |
// a) decode_str + hv_store_ent are abysmally slow. |
1166 |
|
|
// b) most hash keys are short, simple ascii text. |
1167 |
|
|
// => try to "fast-match" such strings to avoid |
1168 |
|
|
// the overhead of decode_str + hv_store_ent. |
1169 |
root |
1.46 |
{ |
1170 |
|
|
SV *value; |
1171 |
|
|
char *p = dec->cur; |
1172 |
|
|
char *e = p + 24; // only try up to 24 bytes |
1173 |
|
|
|
1174 |
|
|
for (;;) |
1175 |
|
|
{ |
1176 |
root |
1.47 |
// the >= 0x80 is true on most architectures |
1177 |
root |
1.46 |
if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1178 |
|
|
{ |
1179 |
|
|
// slow path, back up and use decode_str |
1180 |
|
|
SV *key = decode_str (dec); |
1181 |
|
|
if (!key) |
1182 |
|
|
goto fail; |
1183 |
|
|
|
1184 |
|
|
decode_ws (dec); EXPECT_CH (':'); |
1185 |
|
|
|
1186 |
root |
1.63 |
decode_ws (dec); |
1187 |
root |
1.46 |
value = decode_sv (dec); |
1188 |
|
|
if (!value) |
1189 |
|
|
{ |
1190 |
|
|
SvREFCNT_dec (key); |
1191 |
|
|
goto fail; |
1192 |
|
|
} |
1193 |
|
|
|
1194 |
|
|
hv_store_ent (hv, key, value, 0); |
1195 |
|
|
SvREFCNT_dec (key); |
1196 |
|
|
|
1197 |
|
|
break; |
1198 |
|
|
} |
1199 |
|
|
else if (*p == '"') |
1200 |
|
|
{ |
1201 |
|
|
// fast path, got a simple key |
1202 |
|
|
char *key = dec->cur; |
1203 |
|
|
int len = p - key; |
1204 |
|
|
dec->cur = p + 1; |
1205 |
|
|
|
1206 |
|
|
decode_ws (dec); EXPECT_CH (':'); |
1207 |
|
|
|
1208 |
root |
1.63 |
decode_ws (dec); |
1209 |
root |
1.46 |
value = decode_sv (dec); |
1210 |
|
|
if (!value) |
1211 |
|
|
goto fail; |
1212 |
root |
1.1 |
|
1213 |
root |
1.46 |
hv_store (hv, key, len, value, 0); |
1214 |
root |
1.1 |
|
1215 |
root |
1.46 |
break; |
1216 |
|
|
} |
1217 |
root |
1.1 |
|
1218 |
root |
1.46 |
++p; |
1219 |
|
|
} |
1220 |
|
|
} |
1221 |
root |
1.1 |
|
1222 |
root |
1.12 |
decode_ws (dec); |
1223 |
root |
1.1 |
|
1224 |
root |
1.5 |
if (*dec->cur == '}') |
1225 |
|
|
{ |
1226 |
|
|
++dec->cur; |
1227 |
|
|
break; |
1228 |
|
|
} |
1229 |
root |
1.1 |
|
1230 |
root |
1.5 |
if (*dec->cur != ',') |
1231 |
|
|
ERR (", or } expected while parsing object/hash"); |
1232 |
root |
1.1 |
|
1233 |
root |
1.5 |
++dec->cur; |
1234 |
root |
1.63 |
|
1235 |
|
|
decode_ws (dec); |
1236 |
|
|
|
1237 |
|
|
if (*dec->cur == '}' && dec->json.flags & F_RELAXED) |
1238 |
|
|
{ |
1239 |
|
|
++dec->cur; |
1240 |
|
|
break; |
1241 |
|
|
} |
1242 |
root |
1.5 |
} |
1243 |
root |
1.1 |
|
1244 |
root |
1.18 |
DEC_DEC_DEPTH; |
1245 |
root |
1.49 |
sv = newRV_noinc ((SV *)hv); |
1246 |
|
|
|
1247 |
|
|
// check filter callbacks |
1248 |
|
|
if (dec->json.flags & F_HOOK) |
1249 |
|
|
{ |
1250 |
root |
1.51 |
if (dec->json.cb_sk_object && HvKEYS (hv) == 1) |
1251 |
root |
1.49 |
{ |
1252 |
root |
1.52 |
HE *cb, *he; |
1253 |
root |
1.49 |
|
1254 |
root |
1.52 |
hv_iterinit (hv); |
1255 |
|
|
he = hv_iternext (hv); |
1256 |
|
|
hv_iterinit (hv); |
1257 |
root |
1.49 |
|
1258 |
root |
1.52 |
// the next line creates a mortal sv each time its called. |
1259 |
|
|
// might want to optimise this for common cases. |
1260 |
|
|
cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); |
1261 |
root |
1.49 |
|
1262 |
root |
1.52 |
if (cb) |
1263 |
root |
1.51 |
{ |
1264 |
root |
1.53 |
dSP; |
1265 |
root |
1.52 |
int count; |
1266 |
|
|
|
1267 |
root |
1.53 |
ENTER; SAVETMPS; PUSHMARK (SP); |
1268 |
root |
1.52 |
XPUSHs (HeVAL (he)); |
1269 |
|
|
|
1270 |
|
|
PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1271 |
|
|
|
1272 |
|
|
if (count == 1) |
1273 |
|
|
{ |
1274 |
|
|
sv = newSVsv (POPs); |
1275 |
|
|
FREETMPS; LEAVE; |
1276 |
|
|
return sv; |
1277 |
|
|
} |
1278 |
|
|
|
1279 |
|
|
FREETMPS; LEAVE; |
1280 |
root |
1.51 |
} |
1281 |
root |
1.49 |
} |
1282 |
|
|
|
1283 |
|
|
if (dec->json.cb_object) |
1284 |
|
|
{ |
1285 |
root |
1.53 |
dSP; |
1286 |
root |
1.49 |
int count; |
1287 |
|
|
|
1288 |
root |
1.53 |
ENTER; SAVETMPS; PUSHMARK (SP); |
1289 |
root |
1.49 |
XPUSHs (sv_2mortal (sv)); |
1290 |
|
|
|
1291 |
|
|
PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
1292 |
|
|
|
1293 |
|
|
if (count == 1) |
1294 |
root |
1.51 |
{ |
1295 |
|
|
sv = newSVsv (POPs); |
1296 |
root |
1.52 |
FREETMPS; LEAVE; |
1297 |
|
|
return sv; |
1298 |
root |
1.51 |
} |
1299 |
|
|
|
1300 |
|
|
SvREFCNT_inc (sv); |
1301 |
root |
1.52 |
FREETMPS; LEAVE; |
1302 |
root |
1.49 |
} |
1303 |
|
|
} |
1304 |
|
|
|
1305 |
root |
1.51 |
return sv; |
1306 |
root |
1.1 |
|
1307 |
|
|
fail: |
1308 |
|
|
SvREFCNT_dec (hv); |
1309 |
root |
1.18 |
DEC_DEC_DEPTH; |
1310 |
root |
1.1 |
return 0; |
1311 |
|
|
} |
1312 |
|
|
|
1313 |
|
|
static SV * |
1314 |
|
|
decode_sv (dec_t *dec) |
1315 |
|
|
{ |
1316 |
root |
1.40 |
// the beauty of JSON: you need exactly one character lookahead |
1317 |
root |
1.73 |
// to parse everything. |
1318 |
root |
1.1 |
switch (*dec->cur) |
1319 |
|
|
{ |
1320 |
|
|
case '"': ++dec->cur; return decode_str (dec); |
1321 |
root |
1.73 |
case '[': ++dec->cur; return decode_av (dec); |
1322 |
|
|
case '{': ++dec->cur; return decode_hv (dec); |
1323 |
root |
1.1 |
|
1324 |
|
|
case '-': |
1325 |
|
|
case '0': case '1': case '2': case '3': case '4': |
1326 |
|
|
case '5': case '6': case '7': case '8': case '9': |
1327 |
|
|
return decode_num (dec); |
1328 |
|
|
|
1329 |
|
|
case 't': |
1330 |
|
|
if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1331 |
|
|
{ |
1332 |
|
|
dec->cur += 4; |
1333 |
root |
1.60 |
#if JSON_SLOW |
1334 |
|
|
json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); |
1335 |
|
|
#endif |
1336 |
root |
1.43 |
return SvREFCNT_inc (json_true); |
1337 |
root |
1.1 |
} |
1338 |
|
|
else |
1339 |
|
|
ERR ("'true' expected"); |
1340 |
|
|
|
1341 |
|
|
break; |
1342 |
|
|
|
1343 |
|
|
case 'f': |
1344 |
|
|
if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1345 |
|
|
{ |
1346 |
|
|
dec->cur += 5; |
1347 |
root |
1.60 |
#if JSON_SLOW |
1348 |
|
|
json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1349 |
|
|
#endif |
1350 |
root |
1.43 |
return SvREFCNT_inc (json_false); |
1351 |
root |
1.1 |
} |
1352 |
|
|
else |
1353 |
|
|
ERR ("'false' expected"); |
1354 |
|
|
|
1355 |
|
|
break; |
1356 |
|
|
|
1357 |
|
|
case 'n': |
1358 |
|
|
if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
1359 |
|
|
{ |
1360 |
|
|
dec->cur += 4; |
1361 |
root |
1.5 |
return newSVsv (&PL_sv_undef); |
1362 |
root |
1.1 |
} |
1363 |
|
|
else |
1364 |
|
|
ERR ("'null' expected"); |
1365 |
|
|
|
1366 |
|
|
break; |
1367 |
|
|
|
1368 |
|
|
default: |
1369 |
root |
1.24 |
ERR ("malformed JSON string, neither array, object, number, string or atom"); |
1370 |
root |
1.1 |
break; |
1371 |
|
|
} |
1372 |
|
|
|
1373 |
|
|
fail: |
1374 |
|
|
return 0; |
1375 |
|
|
} |
1376 |
|
|
|
1377 |
|
|
static SV * |
1378 |
root |
1.48 |
decode_json (SV *string, JSON *json, UV *offset_return) |
1379 |
root |
1.1 |
{ |
1380 |
root |
1.25 |
dec_t dec; |
1381 |
root |
1.31 |
UV offset; |
1382 |
root |
1.1 |
SV *sv; |
1383 |
|
|
|
1384 |
root |
1.29 |
SvGETMAGIC (string); |
1385 |
root |
1.22 |
SvUPGRADE (string, SVt_PV); |
1386 |
|
|
|
1387 |
root |
1.47 |
if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) |
1388 |
root |
1.45 |
croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1389 |
root |
1.47 |
(unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); |
1390 |
root |
1.45 |
|
1391 |
root |
1.47 |
if (json->flags & F_UTF8) |
1392 |
root |
1.5 |
sv_utf8_downgrade (string, 0); |
1393 |
|
|
else |
1394 |
root |
1.1 |
sv_utf8_upgrade (string); |
1395 |
|
|
|
1396 |
|
|
SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1397 |
|
|
|
1398 |
root |
1.47 |
dec.json = *json; |
1399 |
root |
1.18 |
dec.cur = SvPVX (string); |
1400 |
|
|
dec.end = SvEND (string); |
1401 |
|
|
dec.err = 0; |
1402 |
|
|
dec.depth = 0; |
1403 |
root |
1.47 |
dec.maxdepth = DEC_DEPTH (dec.json.flags); |
1404 |
root |
1.1 |
|
1405 |
root |
1.49 |
if (dec.json.cb_object || dec.json.cb_sk_object) |
1406 |
|
|
dec.json.flags |= F_HOOK; |
1407 |
|
|
|
1408 |
root |
1.31 |
*dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1409 |
root |
1.63 |
|
1410 |
|
|
decode_ws (&dec); |
1411 |
root |
1.1 |
sv = decode_sv (&dec); |
1412 |
|
|
|
1413 |
root |
1.32 |
if (!(offset_return || !sv)) |
1414 |
root |
1.31 |
{ |
1415 |
|
|
// check for trailing garbage |
1416 |
|
|
decode_ws (&dec); |
1417 |
|
|
|
1418 |
|
|
if (*dec.cur) |
1419 |
|
|
{ |
1420 |
|
|
dec.err = "garbage after JSON object"; |
1421 |
|
|
SvREFCNT_dec (sv); |
1422 |
|
|
sv = 0; |
1423 |
|
|
} |
1424 |
|
|
} |
1425 |
|
|
|
1426 |
root |
1.32 |
if (offset_return || !sv) |
1427 |
|
|
{ |
1428 |
root |
1.47 |
offset = dec.json.flags & F_UTF8 |
1429 |
root |
1.32 |
? dec.cur - SvPVX (string) |
1430 |
|
|
: utf8_distance (dec.cur, SvPVX (string)); |
1431 |
|
|
|
1432 |
|
|
if (offset_return) |
1433 |
|
|
*offset_return = offset; |
1434 |
|
|
} |
1435 |
|
|
|
1436 |
root |
1.1 |
if (!sv) |
1437 |
|
|
{ |
1438 |
|
|
SV *uni = sv_newmortal (); |
1439 |
root |
1.8 |
|
1440 |
root |
1.5 |
// horrible hack to silence warning inside pv_uni_display |
1441 |
root |
1.8 |
COP cop = *PL_curcop; |
1442 |
root |
1.5 |
cop.cop_warnings = pWARN_NONE; |
1443 |
root |
1.8 |
ENTER; |
1444 |
root |
1.5 |
SAVEVPTR (PL_curcop); |
1445 |
|
|
PL_curcop = &cop; |
1446 |
root |
1.8 |
pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1447 |
|
|
LEAVE; |
1448 |
root |
1.1 |
|
1449 |
root |
1.23 |
croak ("%s, at character offset %d [\"%s\"]", |
1450 |
root |
1.1 |
dec.err, |
1451 |
|
|
(int)offset, |
1452 |
|
|
dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1453 |
|
|
} |
1454 |
|
|
|
1455 |
root |
1.3 |
sv = sv_2mortal (sv); |
1456 |
|
|
|
1457 |
root |
1.47 |
if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1458 |
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)"); |
1459 |
root |
1.3 |
|
1460 |
|
|
return sv; |
1461 |
root |
1.1 |
} |
1462 |
|
|
|
1463 |
root |
1.12 |
///////////////////////////////////////////////////////////////////////////// |
1464 |
|
|
// XS interface functions |
1465 |
|
|
|
1466 |
root |
1.1 |
MODULE = JSON::XS PACKAGE = JSON::XS |
1467 |
|
|
|
1468 |
|
|
BOOT: |
1469 |
|
|
{ |
1470 |
|
|
int i; |
1471 |
|
|
|
1472 |
root |
1.18 |
for (i = 0; i < 256; ++i) |
1473 |
|
|
decode_hexdigit [i] = |
1474 |
|
|
i >= '0' && i <= '9' ? i - '0' |
1475 |
|
|
: i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1476 |
|
|
: i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1477 |
|
|
: -1; |
1478 |
root |
1.1 |
|
1479 |
root |
1.44 |
json_stash = gv_stashpv ("JSON::XS" , 1); |
1480 |
|
|
json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
1481 |
root |
1.43 |
|
1482 |
|
|
json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); |
1483 |
|
|
json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1484 |
root |
1.1 |
} |
1485 |
|
|
|
1486 |
root |
1.4 |
PROTOTYPES: DISABLE |
1487 |
|
|
|
1488 |
root |
1.57 |
void CLONE (...) |
1489 |
|
|
CODE: |
1490 |
|
|
json_stash = 0; |
1491 |
|
|
json_boolean_stash = 0; |
1492 |
|
|
|
1493 |
root |
1.48 |
void new (char *klass) |
1494 |
|
|
PPCODE: |
1495 |
|
|
{ |
1496 |
|
|
SV *pv = NEWSV (0, sizeof (JSON)); |
1497 |
|
|
SvPOK_only (pv); |
1498 |
root |
1.49 |
Zero (SvPVX (pv), 1, JSON); |
1499 |
root |
1.48 |
((JSON *)SvPVX (pv))->flags = F_DEFAULT; |
1500 |
root |
1.68 |
XPUSHs (sv_2mortal (sv_bless ( |
1501 |
|
|
newRV_noinc (pv), |
1502 |
|
|
strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
1503 |
|
|
))); |
1504 |
root |
1.48 |
} |
1505 |
root |
1.1 |
|
1506 |
root |
1.48 |
void ascii (JSON *self, int enable = 1) |
1507 |
root |
1.1 |
ALIAS: |
1508 |
root |
1.44 |
ascii = F_ASCII |
1509 |
|
|
latin1 = F_LATIN1 |
1510 |
|
|
utf8 = F_UTF8 |
1511 |
|
|
indent = F_INDENT |
1512 |
|
|
canonical = F_CANONICAL |
1513 |
|
|
space_before = F_SPACE_BEFORE |
1514 |
|
|
space_after = F_SPACE_AFTER |
1515 |
|
|
pretty = F_PRETTY |
1516 |
|
|
allow_nonref = F_ALLOW_NONREF |
1517 |
|
|
shrink = F_SHRINK |
1518 |
|
|
allow_blessed = F_ALLOW_BLESSED |
1519 |
|
|
convert_blessed = F_CONV_BLESSED |
1520 |
root |
1.63 |
relaxed = F_RELAXED |
1521 |
root |
1.48 |
PPCODE: |
1522 |
root |
1.1 |
{ |
1523 |
|
|
if (enable) |
1524 |
root |
1.48 |
self->flags |= ix; |
1525 |
root |
1.1 |
else |
1526 |
root |
1.48 |
self->flags &= ~ix; |
1527 |
root |
1.1 |
|
1528 |
root |
1.48 |
XPUSHs (ST (0)); |
1529 |
root |
1.1 |
} |
1530 |
|
|
|
1531 |
root |
1.66 |
void get_ascii (JSON *self) |
1532 |
|
|
ALIAS: |
1533 |
|
|
get_ascii = F_ASCII |
1534 |
|
|
get_latin1 = F_LATIN1 |
1535 |
|
|
get_utf8 = F_UTF8 |
1536 |
|
|
get_indent = F_INDENT |
1537 |
|
|
get_canonical = F_CANONICAL |
1538 |
|
|
get_space_before = F_SPACE_BEFORE |
1539 |
|
|
get_space_after = F_SPACE_AFTER |
1540 |
|
|
get_allow_nonref = F_ALLOW_NONREF |
1541 |
|
|
get_shrink = F_SHRINK |
1542 |
|
|
get_allow_blessed = F_ALLOW_BLESSED |
1543 |
|
|
get_convert_blessed = F_CONV_BLESSED |
1544 |
|
|
get_relaxed = F_RELAXED |
1545 |
|
|
PPCODE: |
1546 |
|
|
XPUSHs (boolSV (self->flags & ix)); |
1547 |
|
|
|
1548 |
root |
1.48 |
void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
1549 |
|
|
PPCODE: |
1550 |
root |
1.18 |
{ |
1551 |
|
|
UV log2 = 0; |
1552 |
|
|
|
1553 |
|
|
if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
1554 |
|
|
|
1555 |
|
|
while ((1UL << log2) < max_depth) |
1556 |
|
|
++log2; |
1557 |
|
|
|
1558 |
root |
1.48 |
self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
1559 |
root |
1.18 |
|
1560 |
root |
1.48 |
XPUSHs (ST (0)); |
1561 |
root |
1.18 |
} |
1562 |
|
|
|
1563 |
root |
1.67 |
U32 get_max_depth (JSON *self) |
1564 |
root |
1.66 |
CODE: |
1565 |
|
|
RETVAL = DEC_DEPTH (self->flags); |
1566 |
|
|
OUTPUT: |
1567 |
|
|
RETVAL |
1568 |
|
|
|
1569 |
root |
1.48 |
void max_size (JSON *self, UV max_size = 0) |
1570 |
|
|
PPCODE: |
1571 |
root |
1.45 |
{ |
1572 |
|
|
UV log2 = 0; |
1573 |
|
|
|
1574 |
|
|
if (max_size > 0x80000000UL) max_size = 0x80000000UL; |
1575 |
|
|
if (max_size == 1) max_size = 2; |
1576 |
|
|
|
1577 |
|
|
while ((1UL << log2) < max_size) |
1578 |
|
|
++log2; |
1579 |
|
|
|
1580 |
root |
1.48 |
self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); |
1581 |
root |
1.45 |
|
1582 |
root |
1.48 |
XPUSHs (ST (0)); |
1583 |
root |
1.45 |
} |
1584 |
|
|
|
1585 |
root |
1.66 |
int get_max_size (JSON *self) |
1586 |
|
|
CODE: |
1587 |
|
|
RETVAL = DEC_SIZE (self->flags); |
1588 |
|
|
OUTPUT: |
1589 |
|
|
RETVAL |
1590 |
|
|
|
1591 |
root |
1.51 |
void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1592 |
root |
1.49 |
PPCODE: |
1593 |
|
|
{ |
1594 |
root |
1.52 |
SvREFCNT_dec (self->cb_object); |
1595 |
|
|
self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; |
1596 |
|
|
|
1597 |
|
|
XPUSHs (ST (0)); |
1598 |
|
|
} |
1599 |
|
|
|
1600 |
|
|
void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) |
1601 |
|
|
PPCODE: |
1602 |
|
|
{ |
1603 |
|
|
if (!self->cb_sk_object) |
1604 |
|
|
self->cb_sk_object = newHV (); |
1605 |
root |
1.51 |
|
1606 |
root |
1.52 |
if (SvOK (cb)) |
1607 |
|
|
hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
1608 |
root |
1.51 |
else |
1609 |
root |
1.52 |
{ |
1610 |
|
|
hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0); |
1611 |
root |
1.49 |
|
1612 |
root |
1.52 |
if (!HvKEYS (self->cb_sk_object)) |
1613 |
|
|
{ |
1614 |
|
|
SvREFCNT_dec (self->cb_sk_object); |
1615 |
|
|
self->cb_sk_object = 0; |
1616 |
|
|
} |
1617 |
root |
1.49 |
} |
1618 |
|
|
|
1619 |
|
|
XPUSHs (ST (0)); |
1620 |
|
|
} |
1621 |
|
|
|
1622 |
root |
1.48 |
void encode (JSON *self, SV *scalar) |
1623 |
root |
1.1 |
PPCODE: |
1624 |
root |
1.48 |
XPUSHs (encode_json (scalar, self)); |
1625 |
root |
1.1 |
|
1626 |
root |
1.48 |
void decode (JSON *self, SV *jsonstr) |
1627 |
root |
1.1 |
PPCODE: |
1628 |
root |
1.48 |
XPUSHs (decode_json (jsonstr, self, 0)); |
1629 |
root |
1.31 |
|
1630 |
root |
1.48 |
void decode_prefix (JSON *self, SV *jsonstr) |
1631 |
root |
1.31 |
PPCODE: |
1632 |
|
|
{ |
1633 |
|
|
UV offset; |
1634 |
|
|
EXTEND (SP, 2); |
1635 |
root |
1.48 |
PUSHs (decode_json (jsonstr, self, &offset)); |
1636 |
root |
1.31 |
PUSHs (sv_2mortal (newSVuv (offset))); |
1637 |
|
|
} |
1638 |
root |
1.2 |
|
1639 |
root |
1.52 |
void DESTROY (JSON *self) |
1640 |
|
|
CODE: |
1641 |
|
|
SvREFCNT_dec (self->cb_sk_object); |
1642 |
|
|
SvREFCNT_dec (self->cb_object); |
1643 |
|
|
|
1644 |
root |
1.4 |
PROTOTYPES: ENABLE |
1645 |
|
|
|
1646 |
root |
1.69 |
void encode_json (SV *scalar) |
1647 |
root |
1.76 |
ALIAS: |
1648 |
|
|
to_json_ = 0 |
1649 |
|
|
encode_json = F_UTF8 |
1650 |
root |
1.2 |
PPCODE: |
1651 |
root |
1.47 |
{ |
1652 |
root |
1.76 |
JSON json = { F_DEFAULT | ix }; |
1653 |
root |
1.47 |
XPUSHs (encode_json (scalar, &json)); |
1654 |
|
|
} |
1655 |
root |
1.2 |
|
1656 |
root |
1.69 |
void decode_json (SV *jsonstr) |
1657 |
root |
1.76 |
ALIAS: |
1658 |
|
|
from_json_ = 0 |
1659 |
|
|
decode_json = F_UTF8 |
1660 |
root |
1.2 |
PPCODE: |
1661 |
root |
1.47 |
{ |
1662 |
root |
1.76 |
JSON json = { F_DEFAULT | ix }; |
1663 |
root |
1.47 |
XPUSHs (decode_json (jsonstr, &json, 0)); |
1664 |
|
|
} |
1665 |
root |
1.1 |
|