… | |
… | |
4 | |
4 | |
5 | #include <assert.h> |
5 | #include <assert.h> |
6 | #include <string.h> |
6 | #include <string.h> |
7 | #include <stdlib.h> |
7 | #include <stdlib.h> |
8 | #include <stdio.h> |
8 | #include <stdio.h> |
|
|
9 | #include <limits.h> |
9 | #include <float.h> |
10 | #include <float.h> |
|
|
11 | #include <inttypes.h> |
10 | |
12 | |
11 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
13 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
12 | # define snprintf _snprintf // C compilers have this in stdio.h |
14 | # define snprintf _snprintf // C compilers have this in stdio.h |
13 | #endif |
15 | #endif |
14 | |
16 | |
15 | // some old perls do not have this, try to make it work, no |
17 | // some old perls do not have this, try to make it work, no |
16 | // guarentees, though. if it breaks, you get to keep the pieces. |
18 | // guarantees, though. if it breaks, you get to keep the pieces. |
17 | #ifndef UTF8_MAXBYTES |
19 | #ifndef UTF8_MAXBYTES |
18 | # define UTF8_MAXBYTES 13 |
20 | # define UTF8_MAXBYTES 13 |
19 | #endif |
21 | #endif |
|
|
22 | |
|
|
23 | // compatibility with perl <5.18 |
|
|
24 | #ifndef HvNAMELEN_get |
|
|
25 | # define HvNAMELEN_get(hv) strlen (HvNAME (hv)) |
|
|
26 | #endif |
|
|
27 | #ifndef HvNAMELEN |
|
|
28 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
|
|
29 | #endif |
|
|
30 | #ifndef HvNAMEUTF8 |
|
|
31 | # define HvNAMEUTF8(hv) 0 |
|
|
32 | #endif |
|
|
33 | |
|
|
34 | // three extra for rounding, sign, and end of string |
|
|
35 | #define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3) |
20 | |
36 | |
21 | #define F_ASCII 0x00000001UL |
37 | #define F_ASCII 0x00000001UL |
22 | #define F_LATIN1 0x00000002UL |
38 | #define F_LATIN1 0x00000002UL |
23 | #define F_UTF8 0x00000004UL |
39 | #define F_UTF8 0x00000004UL |
24 | #define F_INDENT 0x00000008UL |
40 | #define F_INDENT 0x00000008UL |
… | |
… | |
28 | #define F_ALLOW_NONREF 0x00000100UL |
44 | #define F_ALLOW_NONREF 0x00000100UL |
29 | #define F_SHRINK 0x00000200UL |
45 | #define F_SHRINK 0x00000200UL |
30 | #define F_ALLOW_BLESSED 0x00000400UL |
46 | #define F_ALLOW_BLESSED 0x00000400UL |
31 | #define F_CONV_BLESSED 0x00000800UL |
47 | #define F_CONV_BLESSED 0x00000800UL |
32 | #define F_RELAXED 0x00001000UL |
48 | #define F_RELAXED 0x00001000UL |
33 | |
49 | #define F_ALLOW_UNKNOWN 0x00002000UL |
34 | #define F_MAXDEPTH 0xf8000000UL |
50 | #define F_ALLOW_TAGS 0x00004000UL |
35 | #define S_MAXDEPTH 27 |
|
|
36 | #define F_MAXSIZE 0x01f00000UL |
|
|
37 | #define S_MAXSIZE 20 |
|
|
38 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
51 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
39 | |
52 | |
40 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
|
|
41 | #define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE )) |
|
|
42 | |
|
|
43 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
53 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
44 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
|
|
45 | |
54 | |
46 | #define INIT_SIZE 32 // initial scalar size to be allocated |
55 | #define INIT_SIZE 32 // initial scalar size to be allocated |
47 | #define INDENT_STEP 3 // spaces per indentation level |
56 | #define INDENT_STEP 3 // spaces per indentation level |
48 | |
57 | |
49 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
58 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
|
|
59 | |
|
|
60 | #define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8) |
50 | |
61 | |
51 | #define SB do { |
62 | #define SB do { |
52 | #define SE } while (0) |
63 | #define SE } while (0) |
53 | |
64 | |
54 | #if __GNUC__ >= 3 |
65 | #if __GNUC__ >= 3 |
… | |
… | |
64 | |
75 | |
65 | #define IN_RANGE_INC(type,val,beg,end) \ |
76 | #define IN_RANGE_INC(type,val,beg,end) \ |
66 | ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
77 | ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
67 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
78 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
68 | |
79 | |
|
|
80 | #define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)" |
|
|
81 | |
69 | #ifdef USE_ITHREADS |
82 | #ifdef USE_ITHREADS |
70 | # define JSON_SLOW 1 |
|
|
71 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
83 | # define JSON_STASH (expect_true (json_stash) ? json_stash : gv_stashpv ("JSON::XS", 1)) |
|
|
84 | # define BOOL_STASH (expect_true (bool_stash) ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1)) |
|
|
85 | # define GET_BOOL(value) (expect_true (bool_ ## value) ? bool_ ## value : get_bool ("Types::Serialiser::" # value)) |
72 | #else |
86 | #else |
73 | # define JSON_SLOW 0 |
|
|
74 | # define JSON_STASH json_stash |
87 | # define JSON_STASH json_stash |
|
|
88 | # define BOOL_STASH bool_stash |
|
|
89 | # define GET_BOOL(value) bool_ ## value |
75 | #endif |
90 | #endif |
76 | |
91 | |
77 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
92 | // the amount of HEs to allocate on the stack, when sorting keys |
78 | static SV *json_true, *json_false; |
93 | #define STACK_HES 64 |
|
|
94 | |
|
|
95 | static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean:: |
|
|
96 | static SV *bool_false, *bool_true; |
|
|
97 | static SV *sv_json; |
|
|
98 | |
|
|
99 | enum { |
|
|
100 | INCR_M_WS = 0, // initial whitespace skipping, must be 0 |
|
|
101 | INCR_M_STR, // inside string |
|
|
102 | INCR_M_BS, // inside backslash |
|
|
103 | INCR_M_C0, // inside comment in initial whitespace sequence |
|
|
104 | INCR_M_C1, // inside comment in other places |
|
|
105 | INCR_M_JSON // outside anything, count nesting |
|
|
106 | }; |
|
|
107 | |
|
|
108 | #define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) |
79 | |
109 | |
80 | typedef struct { |
110 | typedef struct { |
81 | U32 flags; |
111 | U32 flags; |
|
|
112 | U32 max_depth; |
|
|
113 | STRLEN max_size; |
|
|
114 | |
82 | SV *cb_object; |
115 | SV *cb_object; |
83 | HV *cb_sk_object; |
116 | HV *cb_sk_object; |
|
|
117 | |
|
|
118 | // for the incremental parser |
|
|
119 | SV *incr_text; // the source text so far |
|
|
120 | STRLEN incr_pos; // the current offset into the text |
|
|
121 | int incr_nest; // {[]}-nesting level |
|
|
122 | unsigned char incr_mode; |
|
|
123 | |
|
|
124 | SV *v_false, *v_true; |
84 | } JSON; |
125 | } JSON; |
|
|
126 | |
|
|
127 | INLINE void |
|
|
128 | json_init (JSON *json) |
|
|
129 | { |
|
|
130 | Zero (json, 1, JSON); |
|
|
131 | json->max_depth = 512; |
|
|
132 | } |
85 | |
133 | |
86 | ///////////////////////////////////////////////////////////////////////////// |
134 | ///////////////////////////////////////////////////////////////////////////// |
87 | // utility functions |
135 | // utility functions |
88 | |
136 | |
|
|
137 | INLINE SV * |
|
|
138 | get_bool (const char *name) |
|
|
139 | { |
|
|
140 | SV *sv = get_sv (name, 1); |
|
|
141 | |
|
|
142 | SvREADONLY_on (sv); |
|
|
143 | SvREADONLY_on (SvRV (sv)); |
|
|
144 | |
|
|
145 | return sv; |
|
|
146 | } |
|
|
147 | |
89 | INLINE void |
148 | INLINE void |
90 | shrink (SV *sv) |
149 | shrink (SV *sv) |
91 | { |
150 | { |
92 | sv_utf8_downgrade (sv, 1); |
151 | sv_utf8_downgrade (sv, 1); |
|
|
152 | |
93 | if (SvLEN (sv) > SvCUR (sv) + 1) |
153 | if (SvLEN (sv) > SvCUR (sv) + 1) |
94 | { |
154 | { |
95 | #ifdef SvPV_shrink_to_cur |
155 | #ifdef SvPV_shrink_to_cur |
96 | SvPV_shrink_to_cur (sv); |
156 | SvPV_shrink_to_cur (sv); |
97 | #elif defined (SvPV_renew) |
157 | #elif defined (SvPV_renew) |
98 | SvPV_renew (sv, SvCUR (sv) + 1); |
158 | SvPV_renew (sv, SvCUR (sv) + 1); |
99 | #endif |
159 | #endif |
100 | } |
160 | } |
101 | } |
161 | } |
102 | |
162 | |
|
|
163 | /* adds two STRLENs together, slow, and with paranoia */ |
|
|
164 | STRLEN |
|
|
165 | strlen_sum (STRLEN l1, STRLEN l2) |
|
|
166 | { |
|
|
167 | size_t sum = l1 + l2; |
|
|
168 | |
|
|
169 | if (sum < (size_t)l2 || sum != (size_t)(STRLEN)sum) |
|
|
170 | croak ("JSON::XS: string size overflow"); |
|
|
171 | |
|
|
172 | return sum; |
|
|
173 | } |
|
|
174 | |
|
|
175 | /* similar to SvGROW, but somewhat safer and guarantees exponential realloc strategy */ |
|
|
176 | static char * |
|
|
177 | json_sv_grow (SV *sv, size_t len1, size_t len2) |
|
|
178 | { |
|
|
179 | len1 = strlen_sum (len1, len2); |
|
|
180 | len1 = strlen_sum (len1, len1 >> 1); |
|
|
181 | |
|
|
182 | if (len1 > 4096 - 24) |
|
|
183 | len1 = (len1 | 4095) - 24; |
|
|
184 | |
|
|
185 | return SvGROW (sv, len1); |
|
|
186 | } |
|
|
187 | |
103 | // decode an utf-8 character and return it, or (UV)-1 in |
188 | // decode a utf-8 character and return it, or (UV)-1 in |
104 | // case of an error. |
189 | // case of an error. |
105 | // we special-case "safe" characters from U+80 .. U+7FF, |
190 | // we special-case "safe" characters from U+80 .. U+7FF, |
106 | // but use the very good perl function to parse anything else. |
191 | // but use the very good perl function to parse anything else. |
107 | // note that we never call this function for a ascii codepoints |
192 | // note that we never call this function for a ascii codepoints |
108 | INLINE UV |
193 | INLINE UV |
… | |
… | |
123 | // this function takes advantage of this fact, although current gccs |
208 | // this function takes advantage of this fact, although current gccs |
124 | // seem to optimise the check for >= 0x80 away anyways |
209 | // seem to optimise the check for >= 0x80 away anyways |
125 | INLINE unsigned char * |
210 | INLINE unsigned char * |
126 | encode_utf8 (unsigned char *s, UV ch) |
211 | encode_utf8 (unsigned char *s, UV ch) |
127 | { |
212 | { |
128 | if (ch <= 0x7FF) |
213 | if (expect_false (ch < 0x000080)) |
129 | { |
214 | *s++ = ch; |
130 | *s++ = (ch >> 6) | 0xc0; |
215 | else if (expect_true (ch < 0x000800)) |
131 | *s++ = (ch & 0x3f) | 0x80; |
216 | *s++ = 0xc0 | ( ch >> 6), |
132 | } |
217 | *s++ = 0x80 | ( ch & 0x3f); |
133 | else |
218 | else if ( ch < 0x010000) |
134 | s = uvuni_to_utf8_flags (s, ch, 0); |
219 | *s++ = 0xe0 | ( ch >> 12), |
|
|
220 | *s++ = 0x80 | ((ch >> 6) & 0x3f), |
|
|
221 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
222 | else if ( ch < 0x110000) |
|
|
223 | *s++ = 0xf0 | ( ch >> 18), |
|
|
224 | *s++ = 0x80 | ((ch >> 12) & 0x3f), |
|
|
225 | *s++ = 0x80 | ((ch >> 6) & 0x3f), |
|
|
226 | *s++ = 0x80 | ( ch & 0x3f); |
135 | |
227 | |
136 | return s; |
228 | return s; |
|
|
229 | } |
|
|
230 | |
|
|
231 | // convert offset pointer to character index, sv must be string |
|
|
232 | static STRLEN |
|
|
233 | ptr_to_index (SV *sv, char *offset) |
|
|
234 | { |
|
|
235 | return SvUTF8 (sv) |
|
|
236 | ? utf8_distance (offset, SvPVX (sv)) |
|
|
237 | : offset - SvPVX (sv); |
|
|
238 | } |
|
|
239 | |
|
|
240 | ///////////////////////////////////////////////////////////////////////////// |
|
|
241 | // fp hell |
|
|
242 | |
|
|
243 | // scan a group of digits, and a trailing exponent |
|
|
244 | static void |
|
|
245 | json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth) |
|
|
246 | { |
|
|
247 | UV uaccum = 0; |
|
|
248 | int eaccum = 0; |
|
|
249 | |
|
|
250 | // if we recurse too deep, skip all remaining digits |
|
|
251 | // to avoid a stack overflow attack |
|
|
252 | if (expect_false (--maxdepth <= 0)) |
|
|
253 | while (((U8)*s - '0') < 10) |
|
|
254 | ++s; |
|
|
255 | |
|
|
256 | for (;;) |
|
|
257 | { |
|
|
258 | U8 dig = (U8)*s - '0'; |
|
|
259 | |
|
|
260 | if (expect_false (dig >= 10)) |
|
|
261 | { |
|
|
262 | if (dig == (U8)((U8)'.' - (U8)'0')) |
|
|
263 | { |
|
|
264 | ++s; |
|
|
265 | json_atof_scan1 (s, accum, expo, 1, maxdepth); |
|
|
266 | } |
|
|
267 | else if ((dig | ' ') == 'e' - '0') |
|
|
268 | { |
|
|
269 | int exp2 = 0; |
|
|
270 | int neg = 0; |
|
|
271 | |
|
|
272 | ++s; |
|
|
273 | |
|
|
274 | if (*s == '-') |
|
|
275 | { |
|
|
276 | ++s; |
|
|
277 | neg = 1; |
|
|
278 | } |
|
|
279 | else if (*s == '+') |
|
|
280 | ++s; |
|
|
281 | |
|
|
282 | while ((dig = (U8)*s - '0') < 10) |
|
|
283 | exp2 = exp2 * 10 + *s++ - '0'; |
|
|
284 | |
|
|
285 | *expo += neg ? -exp2 : exp2; |
|
|
286 | } |
|
|
287 | |
|
|
288 | break; |
|
|
289 | } |
|
|
290 | |
|
|
291 | ++s; |
|
|
292 | |
|
|
293 | uaccum = uaccum * 10 + dig; |
|
|
294 | ++eaccum; |
|
|
295 | |
|
|
296 | // if we have too many digits, then recurse for more |
|
|
297 | // we actually do this for rather few digits |
|
|
298 | if (uaccum >= (UV_MAX - 9) / 10) |
|
|
299 | { |
|
|
300 | if (postdp) *expo -= eaccum; |
|
|
301 | json_atof_scan1 (s, accum, expo, postdp, maxdepth); |
|
|
302 | if (postdp) *expo += eaccum; |
|
|
303 | |
|
|
304 | break; |
|
|
305 | } |
|
|
306 | } |
|
|
307 | |
|
|
308 | // this relies greatly on the quality of the pow () |
|
|
309 | // implementation of the platform, but a good |
|
|
310 | // implementation is hard to beat. |
|
|
311 | // (IEEE 754 conformant ones are required to be exact) |
|
|
312 | if (postdp) *expo -= eaccum; |
|
|
313 | *accum += uaccum * Perl_pow (10., *expo); |
|
|
314 | *expo += eaccum; |
|
|
315 | } |
|
|
316 | |
|
|
317 | static NV |
|
|
318 | json_atof (const char *s) |
|
|
319 | { |
|
|
320 | NV accum = 0.; |
|
|
321 | int expo = 0; |
|
|
322 | int neg = 0; |
|
|
323 | |
|
|
324 | if (*s == '-') |
|
|
325 | { |
|
|
326 | ++s; |
|
|
327 | neg = 1; |
|
|
328 | } |
|
|
329 | |
|
|
330 | // a recursion depth of ten gives us >>500 bits |
|
|
331 | json_atof_scan1 (s, &accum, &expo, 0, 10); |
|
|
332 | |
|
|
333 | return neg ? -accum : accum; |
|
|
334 | } |
|
|
335 | |
|
|
336 | // target of scalar reference is bool? -1 == nope, 0 == false, 1 == true |
|
|
337 | static int |
|
|
338 | ref_bool_type (SV *sv) |
|
|
339 | { |
|
|
340 | svtype svt = SvTYPE (sv); |
|
|
341 | |
|
|
342 | if (svt < SVt_PVAV) |
|
|
343 | { |
|
|
344 | STRLEN len = 0; |
|
|
345 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
346 | |
|
|
347 | if (len == 1) |
|
|
348 | if (*pv == '1') |
|
|
349 | return 1; |
|
|
350 | else if (*pv == '0') |
|
|
351 | return 0; |
|
|
352 | } |
|
|
353 | |
|
|
354 | return -1; |
|
|
355 | } |
|
|
356 | |
|
|
357 | // returns whether scalar is not a reference in the sense of allow_nonref |
|
|
358 | static int |
|
|
359 | json_nonref (SV *scalar) |
|
|
360 | { |
|
|
361 | if (!SvROK (scalar)) |
|
|
362 | return 1; |
|
|
363 | |
|
|
364 | scalar = SvRV (scalar); |
|
|
365 | |
|
|
366 | if (SvTYPE (scalar) >= SVt_PVMG) |
|
|
367 | { |
|
|
368 | if (SvSTASH (scalar) == bool_stash) |
|
|
369 | return 1; |
|
|
370 | |
|
|
371 | if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0) |
|
|
372 | return 1; |
|
|
373 | } |
|
|
374 | |
|
|
375 | return 0; |
137 | } |
376 | } |
138 | |
377 | |
139 | ///////////////////////////////////////////////////////////////////////////// |
378 | ///////////////////////////////////////////////////////////////////////////// |
140 | // encoder |
379 | // encoder |
141 | |
380 | |
… | |
… | |
145 | char *cur; // SvPVX (sv) + current output position |
384 | char *cur; // SvPVX (sv) + current output position |
146 | char *end; // SvEND (sv) |
385 | char *end; // SvEND (sv) |
147 | SV *sv; // result scalar |
386 | SV *sv; // result scalar |
148 | JSON json; |
387 | JSON json; |
149 | U32 indent; // indentation level |
388 | U32 indent; // indentation level |
150 | U32 maxdepth; // max. indentation/recursion level |
|
|
151 | UV limit; // escape character values >= this value when encoding |
389 | UV limit; // escape character values >= this value when encoding |
152 | } enc_t; |
390 | } enc_t; |
153 | |
391 | |
154 | INLINE void |
392 | INLINE void |
155 | need (enc_t *enc, STRLEN len) |
393 | need (enc_t *enc, STRLEN len) |
156 | { |
394 | { |
157 | if (expect_false (enc->cur + len >= enc->end)) |
395 | if (expect_false ((uintptr_t)(enc->end - enc->cur) < len)) |
158 | { |
396 | { |
159 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
397 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
160 | SvGROW (enc->sv, cur + len + 1); |
398 | char *buf = json_sv_grow (enc->sv, cur, len); |
161 | enc->cur = SvPVX (enc->sv) + cur; |
399 | enc->cur = buf + cur; |
162 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
400 | enc->end = buf + SvLEN (enc->sv) - 1; |
163 | } |
401 | } |
164 | } |
402 | } |
165 | |
403 | |
166 | INLINE void |
404 | INLINE void |
167 | encode_ch (enc_t *enc, char ch) |
405 | encode_ch (enc_t *enc, char ch) |
… | |
… | |
183 | |
421 | |
184 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
422 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
185 | { |
423 | { |
186 | if (expect_false (ch == '"')) // but with slow exceptions |
424 | if (expect_false (ch == '"')) // but with slow exceptions |
187 | { |
425 | { |
188 | need (enc, len += 1); |
426 | need (enc, len + 1); |
189 | *enc->cur++ = '\\'; |
427 | *enc->cur++ = '\\'; |
190 | *enc->cur++ = '"'; |
428 | *enc->cur++ = '"'; |
191 | } |
429 | } |
192 | else if (expect_false (ch == '\\')) |
430 | else if (expect_false (ch == '\\')) |
193 | { |
431 | { |
194 | need (enc, len += 1); |
432 | need (enc, len + 1); |
195 | *enc->cur++ = '\\'; |
433 | *enc->cur++ = '\\'; |
196 | *enc->cur++ = '\\'; |
434 | *enc->cur++ = '\\'; |
197 | } |
435 | } |
198 | else |
436 | else |
199 | *enc->cur++ = ch; |
437 | *enc->cur++ = ch; |
… | |
… | |
202 | } |
440 | } |
203 | else |
441 | else |
204 | { |
442 | { |
205 | switch (ch) |
443 | switch (ch) |
206 | { |
444 | { |
207 | case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; |
445 | case '\010': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; |
208 | case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; |
446 | case '\011': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; |
209 | case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; |
447 | case '\012': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; |
210 | case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; |
448 | case '\014': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; |
211 | case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; |
449 | case '\015': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; |
212 | |
450 | |
213 | default: |
451 | default: |
214 | { |
452 | { |
215 | STRLEN clen; |
453 | STRLEN clen; |
216 | UV uch; |
454 | UV uch; |
… | |
… | |
227 | clen = 1; |
465 | clen = 1; |
228 | } |
466 | } |
229 | |
467 | |
230 | if (uch < 0x80/*0x20*/ || uch >= enc->limit) |
468 | if (uch < 0x80/*0x20*/ || uch >= enc->limit) |
231 | { |
469 | { |
232 | if (uch > 0xFFFFUL) |
470 | if (uch >= 0x10000UL) |
233 | { |
471 | { |
234 | if (uch > 0x10FFFFUL) |
472 | if (uch >= 0x110000UL) |
235 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
473 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
236 | |
474 | |
237 | need (enc, len += 11); |
475 | need (enc, len + 11); |
238 | sprintf (enc->cur, "\\u%04x\\u%04x", |
476 | sprintf (enc->cur, "\\u%04x\\u%04x", |
239 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
477 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
240 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
478 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
241 | enc->cur += 12; |
479 | enc->cur += 12; |
242 | } |
480 | } |
243 | else |
481 | else |
244 | { |
482 | { |
245 | static char hexdigit [16] = "0123456789abcdef"; |
|
|
246 | need (enc, len += 5); |
483 | need (enc, len + 5); |
247 | *enc->cur++ = '\\'; |
484 | *enc->cur++ = '\\'; |
248 | *enc->cur++ = 'u'; |
485 | *enc->cur++ = 'u'; |
249 | *enc->cur++ = hexdigit [ uch >> 12 ]; |
486 | *enc->cur++ = PL_hexdigit [ uch >> 12 ]; |
250 | *enc->cur++ = hexdigit [(uch >> 8) & 15]; |
487 | *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; |
251 | *enc->cur++ = hexdigit [(uch >> 4) & 15]; |
488 | *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; |
252 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
489 | *enc->cur++ = PL_hexdigit [(uch >> 0) & 15]; |
253 | } |
490 | } |
254 | |
491 | |
255 | str += clen; |
492 | str += clen; |
256 | } |
493 | } |
257 | else if (enc->json.flags & F_LATIN1) |
494 | else if (enc->json.flags & F_LATIN1) |
… | |
… | |
259 | *enc->cur++ = uch; |
496 | *enc->cur++ = uch; |
260 | str += clen; |
497 | str += clen; |
261 | } |
498 | } |
262 | else if (is_utf8) |
499 | else if (is_utf8) |
263 | { |
500 | { |
264 | need (enc, len += clen); |
501 | need (enc, len + clen); |
265 | do |
502 | do |
266 | { |
503 | { |
267 | *enc->cur++ = *str++; |
504 | *enc->cur++ = *str++; |
268 | } |
505 | } |
269 | while (--clen); |
506 | while (--clen); |
270 | } |
507 | } |
271 | else |
508 | else |
272 | { |
509 | { |
273 | need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
510 | need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
274 | enc->cur = encode_utf8 (enc->cur, uch); |
511 | enc->cur = encode_utf8 (enc->cur, uch); |
275 | ++str; |
512 | ++str; |
276 | } |
513 | } |
277 | } |
514 | } |
278 | } |
515 | } |
… | |
… | |
328 | static void |
565 | static void |
329 | encode_av (enc_t *enc, AV *av) |
566 | encode_av (enc_t *enc, AV *av) |
330 | { |
567 | { |
331 | int i, len = av_len (av); |
568 | int i, len = av_len (av); |
332 | |
569 | |
333 | if (enc->indent >= enc->maxdepth) |
570 | if (enc->indent >= enc->json.max_depth) |
334 | croak ("data structure too deep (hit recursion limit)"); |
571 | croak (ERR_NESTING_EXCEEDED); |
335 | |
572 | |
336 | encode_ch (enc, '['); |
573 | encode_ch (enc, '['); |
337 | |
574 | |
338 | if (len >= 0) |
575 | if (len >= 0) |
339 | { |
576 | { |
340 | encode_nl (enc); ++enc->indent; |
577 | encode_nl (enc); ++enc->indent; |
341 | |
578 | |
342 | for (i = 0; i <= len; ++i) |
579 | for (i = 0; i <= len; ++i) |
… | |
… | |
354 | encode_comma (enc); |
591 | encode_comma (enc); |
355 | } |
592 | } |
356 | |
593 | |
357 | encode_nl (enc); --enc->indent; encode_indent (enc); |
594 | encode_nl (enc); --enc->indent; encode_indent (enc); |
358 | } |
595 | } |
359 | |
596 | |
360 | encode_ch (enc, ']'); |
597 | encode_ch (enc, ']'); |
361 | } |
598 | } |
362 | |
599 | |
363 | static void |
600 | static void |
364 | encode_hk (enc_t *enc, HE *he) |
601 | encode_hk (enc_t *enc, HE *he) |
… | |
… | |
368 | if (HeKLEN (he) == HEf_SVKEY) |
605 | if (HeKLEN (he) == HEf_SVKEY) |
369 | { |
606 | { |
370 | SV *sv = HeSVKEY (he); |
607 | SV *sv = HeSVKEY (he); |
371 | STRLEN len; |
608 | STRLEN len; |
372 | char *str; |
609 | char *str; |
373 | |
610 | |
374 | SvGETMAGIC (sv); |
611 | SvGETMAGIC (sv); |
375 | str = SvPV (sv, len); |
612 | str = SvPV (sv, len); |
376 | |
613 | |
377 | encode_str (enc, str, len, SvUTF8 (sv)); |
614 | encode_str (enc, str, len, SvUTF8 (sv)); |
378 | } |
615 | } |
… | |
… | |
413 | |
650 | |
414 | static void |
651 | static void |
415 | encode_hv (enc_t *enc, HV *hv) |
652 | encode_hv (enc_t *enc, HV *hv) |
416 | { |
653 | { |
417 | HE *he; |
654 | HE *he; |
418 | int count; |
|
|
419 | |
655 | |
420 | if (enc->indent >= enc->maxdepth) |
656 | if (enc->indent >= enc->json.max_depth) |
421 | croak ("data structure too deep (hit recursion limit)"); |
657 | croak (ERR_NESTING_EXCEEDED); |
422 | |
658 | |
423 | encode_ch (enc, '{'); |
659 | encode_ch (enc, '{'); |
424 | |
660 | |
425 | // for canonical output we have to sort by keys first |
661 | // for canonical output we have to sort by keys first |
426 | // actually, this is mostly due to the stupid so-called |
662 | // actually, this is mostly due to the stupid so-called |
427 | // security workaround added somewhere in 5.8.x. |
663 | // security workaround added somewhere in 5.8.x |
428 | // that randomises hash orderings |
664 | // that randomises hash orderings |
429 | if (enc->json.flags & F_CANONICAL) |
665 | if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv)) |
430 | { |
666 | { |
431 | int count = hv_iterinit (hv); |
667 | int count = hv_iterinit (hv); |
432 | |
668 | |
433 | if (SvMAGICAL (hv)) |
669 | if (SvMAGICAL (hv)) |
434 | { |
670 | { |
… | |
… | |
445 | } |
681 | } |
446 | |
682 | |
447 | if (count) |
683 | if (count) |
448 | { |
684 | { |
449 | int i, fast = 1; |
685 | int i, fast = 1; |
450 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
686 | HE *hes_stack [STACK_HES]; |
451 | HE **hes = _alloca (count * sizeof (HE)); |
687 | HE **hes = hes_stack; |
452 | #else |
688 | |
453 | HE *hes [count]; // if your compiler dies here, you need to enable C99 mode |
689 | // allocate larger arrays on the heap |
454 | #endif |
690 | if (count > STACK_HES) |
|
|
691 | { |
|
|
692 | SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes))); |
|
|
693 | hes = (HE **)SvPVX (sv); |
|
|
694 | } |
455 | |
695 | |
456 | i = 0; |
696 | i = 0; |
457 | while ((he = hv_iternext (hv))) |
697 | while ((he = hv_iternext (hv))) |
458 | { |
698 | { |
459 | hes [i++] = he; |
699 | hes [i++] = he; |
… | |
… | |
528 | // encode objects, arrays and special \0=false and \1=true values. |
768 | // encode objects, arrays and special \0=false and \1=true values. |
529 | static void |
769 | static void |
530 | encode_rv (enc_t *enc, SV *sv) |
770 | encode_rv (enc_t *enc, SV *sv) |
531 | { |
771 | { |
532 | svtype svt; |
772 | svtype svt; |
|
|
773 | GV *method; |
533 | |
774 | |
534 | SvGETMAGIC (sv); |
775 | SvGETMAGIC (sv); |
535 | svt = SvTYPE (sv); |
776 | svt = SvTYPE (sv); |
536 | |
777 | |
537 | if (expect_false (SvOBJECT (sv))) |
778 | if (expect_false (SvOBJECT (sv))) |
538 | { |
779 | { |
539 | HV *stash = !JSON_SLOW || json_boolean_stash |
780 | HV *stash = SvSTASH (sv); |
540 | ? json_boolean_stash |
|
|
541 | : gv_stashpv ("JSON::XS::Boolean", 1); |
|
|
542 | |
781 | |
543 | if (SvSTASH (sv) == stash) |
782 | if (stash == bool_stash) |
544 | { |
783 | { |
545 | if (SvIV (sv)) |
|
|
546 | encode_str (enc, "true", 4, 0); |
784 | if (SvIV (sv)) encode_str (enc, "true" , 4, 0); |
547 | else |
|
|
548 | encode_str (enc, "false", 5, 0); |
785 | else encode_str (enc, "false", 5, 0); |
549 | } |
786 | } |
|
|
787 | else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0))) |
|
|
788 | { |
|
|
789 | int count; |
|
|
790 | dSP; |
|
|
791 | |
|
|
792 | ENTER; SAVETMPS; |
|
|
793 | SAVESTACK_POS (); |
|
|
794 | PUSHMARK (SP); |
|
|
795 | EXTEND (SP, 2); |
|
|
796 | // we re-bless the reference to get overload and other niceties right |
|
|
797 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
|
|
798 | PUSHs (sv_json); |
|
|
799 | |
|
|
800 | PUTBACK; |
|
|
801 | count = call_sv ((SV *)GvCV (method), G_ARRAY); |
|
|
802 | SPAGAIN; |
|
|
803 | |
|
|
804 | // catch this surprisingly common error |
|
|
805 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
806 | croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
807 | |
|
|
808 | encode_ch (enc, '('); |
|
|
809 | encode_ch (enc, '"'); |
|
|
810 | encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash)); |
|
|
811 | encode_ch (enc, '"'); |
|
|
812 | encode_ch (enc, ')'); |
|
|
813 | encode_ch (enc, '['); |
|
|
814 | |
|
|
815 | while (count) |
|
|
816 | { |
|
|
817 | encode_sv (enc, SP[1 - count--]); |
|
|
818 | |
|
|
819 | if (count) |
|
|
820 | encode_ch (enc, ','); |
|
|
821 | } |
|
|
822 | |
|
|
823 | encode_ch (enc, ']'); |
|
|
824 | |
|
|
825 | FREETMPS; LEAVE; |
|
|
826 | } |
|
|
827 | else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0))) |
|
|
828 | { |
|
|
829 | dSP; |
|
|
830 | |
|
|
831 | ENTER; SAVETMPS; |
|
|
832 | PUSHMARK (SP); |
|
|
833 | // we re-bless the reference to get overload and other niceties right |
|
|
834 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
|
|
835 | |
|
|
836 | // calling with G_SCALAR ensures that we always get a 1 return value |
|
|
837 | PUTBACK; |
|
|
838 | call_sv ((SV *)GvCV (method), G_SCALAR); |
|
|
839 | SPAGAIN; |
|
|
840 | |
|
|
841 | // catch this surprisingly common error |
|
|
842 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
843 | croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
844 | |
|
|
845 | sv = POPs; |
|
|
846 | PUTBACK; |
|
|
847 | |
|
|
848 | encode_sv (enc, sv); |
|
|
849 | |
|
|
850 | FREETMPS; LEAVE; |
|
|
851 | } |
|
|
852 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
853 | encode_str (enc, "null", 4, 0); |
550 | else |
854 | else |
551 | { |
|
|
552 | #if 0 |
|
|
553 | if (0 && sv_derived_from (rv, "JSON::Literal")) |
|
|
554 | { |
|
|
555 | // not yet |
|
|
556 | } |
|
|
557 | #endif |
|
|
558 | if (enc->json.flags & F_CONV_BLESSED) |
|
|
559 | { |
|
|
560 | // we re-bless the reference to get overload and other niceties right |
|
|
561 | GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0); |
|
|
562 | |
|
|
563 | if (to_json) |
|
|
564 | { |
|
|
565 | dSP; |
|
|
566 | |
|
|
567 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
568 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); |
|
|
569 | |
|
|
570 | // calling with G_SCALAR ensures that we always get a 1 return value |
|
|
571 | PUTBACK; |
|
|
572 | call_sv ((SV *)GvCV (to_json), G_SCALAR); |
|
|
573 | SPAGAIN; |
|
|
574 | |
|
|
575 | // catch this surprisingly common error |
|
|
576 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
577 | croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
578 | |
|
|
579 | sv = POPs; |
|
|
580 | PUTBACK; |
|
|
581 | |
|
|
582 | encode_sv (enc, sv); |
|
|
583 | |
|
|
584 | FREETMPS; LEAVE; |
|
|
585 | } |
|
|
586 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
587 | encode_str (enc, "null", 4, 0); |
|
|
588 | else |
|
|
589 | croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it", |
|
|
590 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
591 | } |
|
|
592 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
593 | encode_str (enc, "null", 4, 0); |
|
|
594 | else |
|
|
595 | croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", |
855 | croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)", |
596 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
856 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
597 | } |
|
|
598 | } |
857 | } |
599 | else if (svt == SVt_PVHV) |
858 | else if (svt == SVt_PVHV) |
600 | encode_hv (enc, (HV *)sv); |
859 | encode_hv (enc, (HV *)sv); |
601 | else if (svt == SVt_PVAV) |
860 | else if (svt == SVt_PVAV) |
602 | encode_av (enc, (AV *)sv); |
861 | encode_av (enc, (AV *)sv); |
603 | else if (svt < SVt_PVAV) |
862 | else if (svt < SVt_PVAV) |
604 | { |
863 | { |
605 | STRLEN len = 0; |
864 | int bool_type = ref_bool_type (sv); |
606 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
607 | |
865 | |
608 | if (len == 1 && *pv == '1') |
866 | if (bool_type == 1) |
609 | encode_str (enc, "true", 4, 0); |
867 | encode_str (enc, "true", 4, 0); |
610 | else if (len == 1 && *pv == '0') |
868 | else if (bool_type == 0) |
611 | encode_str (enc, "false", 5, 0); |
869 | encode_str (enc, "false", 5, 0); |
|
|
870 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
871 | encode_str (enc, "null", 4, 0); |
612 | else |
872 | else |
613 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
873 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
614 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
874 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
615 | } |
875 | } |
|
|
876 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
877 | encode_str (enc, "null", 4, 0); |
616 | else |
878 | else |
617 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
879 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
618 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
880 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
619 | } |
881 | } |
620 | |
882 | |
… | |
… | |
638 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
900 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
639 | enc->cur += strlen (enc->cur); |
901 | enc->cur += strlen (enc->cur); |
640 | } |
902 | } |
641 | else if (SvIOKp (sv)) |
903 | else if (SvIOKp (sv)) |
642 | { |
904 | { |
643 | // we assume we can always read an IV as a UV |
905 | // we assume we can always read an IV as a UV and vice versa |
644 | if (SvUV (sv) & ~(UV)0x7fff) |
906 | // we assume two's complement |
645 | { |
907 | // we assume no aliasing issues in the union |
646 | // large integer, use the (rather slow) snprintf way. |
908 | if (SvIsUV (sv) ? SvUVX (sv) <= 59000 |
647 | need (enc, sizeof (UV) * 3); |
909 | : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000) |
648 | enc->cur += |
|
|
649 | SvIsUV(sv) |
|
|
650 | ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv)) |
|
|
651 | : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv)); |
|
|
652 | } |
|
|
653 | else |
|
|
654 | { |
910 | { |
655 | // optimise the "small number case" |
911 | // optimise the "small number case" |
656 | // code will likely be branchless and use only a single multiplication |
912 | // code will likely be branchless and use only a single multiplication |
|
|
913 | // works for numbers up to 59074 |
657 | I32 i = SvIV (sv); |
914 | I32 i = SvIVX (sv); |
658 | U32 u; |
915 | U32 u; |
659 | char digit, nz = 0; |
916 | char digit, nz = 0; |
660 | |
917 | |
661 | need (enc, 6); |
918 | need (enc, 6); |
662 | |
919 | |
… | |
… | |
668 | |
925 | |
669 | // now output digit by digit, each time masking out the integer part |
926 | // now output digit by digit, each time masking out the integer part |
670 | // and multiplying by 5 while moving the decimal point one to the right, |
927 | // and multiplying by 5 while moving the decimal point one to the right, |
671 | // resulting in a net multiplication by 10. |
928 | // resulting in a net multiplication by 10. |
672 | // we always write the digit to memory but conditionally increment |
929 | // we always write the digit to memory but conditionally increment |
673 | // the pointer, to ease the usage of conditional move instructions. |
930 | // the pointer, to enable the use of conditional move instructions. |
674 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; |
931 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5; |
675 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; |
932 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5; |
676 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; |
933 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5; |
677 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; |
934 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5; |
678 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
935 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
679 | } |
936 | } |
|
|
937 | else |
|
|
938 | { |
|
|
939 | // large integer, use the (rather slow) snprintf way. |
|
|
940 | need (enc, IVUV_MAXCHARS); |
|
|
941 | enc->cur += |
|
|
942 | SvIsUV(sv) |
|
|
943 | ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) |
|
|
944 | : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); |
|
|
945 | } |
680 | } |
946 | } |
681 | else if (SvROK (sv)) |
947 | else if (SvROK (sv)) |
682 | encode_rv (enc, SvRV (sv)); |
948 | encode_rv (enc, SvRV (sv)); |
683 | else if (!SvOK (sv)) |
949 | else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) |
684 | encode_str (enc, "null", 4, 0); |
950 | encode_str (enc, "null", 4, 0); |
685 | else |
951 | else |
686 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
952 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data", |
687 | SvPV_nolen (sv), SvFLAGS (sv)); |
953 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
688 | } |
954 | } |
689 | |
955 | |
690 | static SV * |
956 | static SV * |
691 | encode_json (SV *scalar, JSON *json) |
957 | encode_json (SV *scalar, JSON *json) |
692 | { |
958 | { |
693 | enc_t enc; |
959 | enc_t enc; |
694 | |
960 | |
695 | if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
961 | if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar)) |
696 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
962 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
697 | |
963 | |
698 | enc.json = *json; |
964 | enc.json = *json; |
699 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
965 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
700 | enc.cur = SvPVX (enc.sv); |
966 | enc.cur = SvPVX (enc.sv); |
701 | enc.end = SvEND (enc.sv); |
967 | enc.end = SvEND (enc.sv); |
702 | enc.indent = 0; |
968 | enc.indent = 0; |
703 | enc.maxdepth = DEC_DEPTH (enc.json.flags); |
|
|
704 | enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
969 | enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
705 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
970 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
706 | : 0x10FFFFUL; |
971 | : 0x110000UL; |
707 | |
972 | |
708 | SvPOK_only (enc.sv); |
973 | SvPOK_only (enc.sv); |
709 | encode_sv (&enc, scalar); |
974 | encode_sv (&enc, scalar); |
|
|
975 | encode_nl (&enc); |
710 | |
976 | |
711 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
977 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
712 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
978 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
713 | |
979 | |
714 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
980 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
… | |
… | |
775 | if (*dec->cur != ch) \ |
1041 | if (*dec->cur != ch) \ |
776 | ERR (# ch " expected"); \ |
1042 | ERR (# ch " expected"); \ |
777 | ++dec->cur; \ |
1043 | ++dec->cur; \ |
778 | SE |
1044 | SE |
779 | |
1045 | |
780 | #define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") |
1046 | #define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED) |
781 | #define DEC_DEC_DEPTH --dec->depth |
1047 | #define DEC_DEC_DEPTH --dec->depth |
782 | |
1048 | |
783 | static SV *decode_sv (dec_t *dec); |
1049 | static SV *decode_sv (dec_t *dec); |
784 | |
1050 | |
785 | static signed char decode_hexdigit[256]; |
1051 | static signed char decode_hexdigit[256]; |
… | |
… | |
889 | default: |
1155 | default: |
890 | --dec_cur; |
1156 | --dec_cur; |
891 | ERR ("illegal backslash escape sequence in string"); |
1157 | ERR ("illegal backslash escape sequence in string"); |
892 | } |
1158 | } |
893 | } |
1159 | } |
894 | else if (expect_true (ch >= 0x20 && ch <= 0x7f)) |
1160 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
895 | *cur++ = ch; |
1161 | *cur++ = ch; |
896 | else if (ch >= 0x80) |
1162 | else if (ch >= 0x80) |
897 | { |
1163 | { |
898 | STRLEN clen; |
1164 | STRLEN clen; |
899 | UV uch; |
|
|
900 | |
1165 | |
901 | --dec_cur; |
1166 | --dec_cur; |
902 | |
1167 | |
903 | uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
1168 | decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
904 | if (clen == (STRLEN)-1) |
1169 | if (clen == (STRLEN)-1) |
905 | ERR ("malformed UTF-8 character in JSON string"); |
1170 | ERR ("malformed UTF-8 character in JSON string"); |
906 | |
1171 | |
907 | do |
1172 | do |
908 | *cur++ = *dec_cur++; |
1173 | *cur++ = *dec_cur++; |
909 | while (--clen); |
1174 | while (--clen); |
910 | |
1175 | |
911 | utf8 = 1; |
1176 | utf8 = 1; |
912 | } |
1177 | } |
|
|
1178 | else if (ch == '\t' && dec->json.flags & F_RELAXED) |
|
|
1179 | *cur++ = ch; |
913 | else |
1180 | else |
914 | { |
1181 | { |
915 | --dec_cur; |
1182 | --dec_cur; |
916 | |
1183 | |
917 | if (!ch) |
1184 | if (!ch) |
… | |
… | |
925 | { |
1192 | { |
926 | STRLEN len = cur - buf; |
1193 | STRLEN len = cur - buf; |
927 | |
1194 | |
928 | if (sv) |
1195 | if (sv) |
929 | { |
1196 | { |
930 | SvGROW (sv, SvCUR (sv) + len + 1); |
1197 | STRLEN cur = SvCUR (sv); |
|
|
1198 | |
|
|
1199 | if (SvLEN (sv) - cur <= len) |
|
|
1200 | json_sv_grow (sv, cur, len); |
|
|
1201 | |
931 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
1202 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
932 | SvCUR_set (sv, SvCUR (sv) + len); |
1203 | SvCUR_set (sv, SvCUR (sv) + len); |
933 | } |
1204 | } |
934 | else |
1205 | else |
935 | sv = newSVpvn (buf, len); |
1206 | sv = newSVpvn (buf, len); |
… | |
… | |
1026 | |
1297 | |
1027 | // special case the rather common 1..5-digit-int case |
1298 | // special case the rather common 1..5-digit-int case |
1028 | if (*start == '-') |
1299 | if (*start == '-') |
1029 | switch (len) |
1300 | switch (len) |
1030 | { |
1301 | { |
1031 | case 2: return newSViv (-( start [1] - '0' * 1)); |
1302 | case 2: return newSViv (-(IV)( start [1] - '0' * 1)); |
1032 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
1303 | case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11)); |
1033 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
1304 | case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
1034 | case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
1305 | case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
1035 | case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
1306 | case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
1036 | } |
1307 | } |
1037 | else |
1308 | else |
1038 | switch (len) |
1309 | switch (len) |
1039 | { |
1310 | { |
1040 | case 1: return newSViv ( start [0] - '0' * 1); |
1311 | case 1: return newSViv ( start [0] - '0' * 1); |
1041 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1312 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1042 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
1313 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
1043 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
1314 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
1044 | case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
1315 | case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
1045 | } |
1316 | } |
1046 | |
1317 | |
1047 | { |
1318 | { |
1048 | UV uv; |
1319 | UV uv; |
1049 | int numtype = grok_number (start, len, &uv); |
1320 | int numtype = grok_number (start, len, &uv); |
… | |
… | |
1058 | } |
1329 | } |
1059 | |
1330 | |
1060 | len -= *start == '-' ? 1 : 0; |
1331 | len -= *start == '-' ? 1 : 0; |
1061 | |
1332 | |
1062 | // does not fit into IV or UV, try NV |
1333 | // does not fit into IV or UV, try NV |
1063 | if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) |
1334 | if (len <= NV_DIG) |
1064 | #if defined (LDBL_DIG) |
|
|
1065 | || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len) |
|
|
1066 | #endif |
|
|
1067 | ) |
|
|
1068 | // fits into NV without loss of precision |
1335 | // fits into NV without loss of precision |
1069 | return newSVnv (Atof (start)); |
1336 | return newSVnv (json_atof (start)); |
1070 | |
1337 | |
1071 | // everything else fails, convert it to a string |
1338 | // everything else fails, convert it to a string |
1072 | return newSVpvn (start, dec->cur - start); |
1339 | return newSVpvn (start, dec->cur - start); |
1073 | } |
1340 | } |
1074 | |
1341 | |
1075 | // loss of precision here |
1342 | // loss of precision here |
1076 | return newSVnv (Atof (start)); |
1343 | return newSVnv (json_atof (start)); |
1077 | |
1344 | |
1078 | fail: |
1345 | fail: |
1079 | return 0; |
1346 | return 0; |
1080 | } |
1347 | } |
1081 | |
1348 | |
… | |
… | |
1105 | if (*dec->cur == ']') |
1372 | if (*dec->cur == ']') |
1106 | { |
1373 | { |
1107 | ++dec->cur; |
1374 | ++dec->cur; |
1108 | break; |
1375 | break; |
1109 | } |
1376 | } |
1110 | |
1377 | |
1111 | if (*dec->cur != ',') |
1378 | if (*dec->cur != ',') |
1112 | ERR (", or ] expected while parsing array"); |
1379 | ERR (", or ] expected while parsing array"); |
1113 | |
1380 | |
1114 | ++dec->cur; |
1381 | ++dec->cur; |
1115 | |
1382 | |
… | |
… | |
1157 | char *p = dec->cur; |
1424 | char *p = dec->cur; |
1158 | char *e = p + 24; // only try up to 24 bytes |
1425 | char *e = p + 24; // only try up to 24 bytes |
1159 | |
1426 | |
1160 | for (;;) |
1427 | for (;;) |
1161 | { |
1428 | { |
1162 | // the >= 0x80 is true on most architectures |
1429 | // the >= 0x80 is false on most architectures |
1163 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1430 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1164 | { |
1431 | { |
1165 | // slow path, back up and use decode_str |
1432 | // slow path, back up and use decode_str |
1166 | SV *key = decode_str (dec); |
1433 | SV *key = decode_str (dec); |
1167 | if (!key) |
1434 | if (!key) |
… | |
… | |
1239 | |
1506 | |
1240 | hv_iterinit (hv); |
1507 | hv_iterinit (hv); |
1241 | he = hv_iternext (hv); |
1508 | he = hv_iternext (hv); |
1242 | hv_iterinit (hv); |
1509 | hv_iterinit (hv); |
1243 | |
1510 | |
1244 | // the next line creates a mortal sv each time its called. |
1511 | // the next line creates a mortal sv each time it's called. |
1245 | // might want to optimise this for common cases. |
1512 | // might want to optimise this for common cases. |
1246 | cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); |
1513 | cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); |
1247 | |
1514 | |
1248 | if (cb) |
1515 | if (cb) |
1249 | { |
1516 | { |
1250 | dSP; |
1517 | dSP; |
1251 | int count; |
1518 | int count; |
1252 | |
1519 | |
1253 | ENTER; SAVETMPS; PUSHMARK (SP); |
1520 | ENTER; SAVETMPS; |
|
|
1521 | SAVESTACK_POS (); |
|
|
1522 | PUSHMARK (SP); |
1254 | XPUSHs (HeVAL (he)); |
1523 | XPUSHs (HeVAL (he)); |
|
|
1524 | sv_2mortal (sv); |
1255 | |
1525 | |
1256 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1526 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1257 | |
1527 | |
1258 | if (count == 1) |
1528 | if (count == 1) |
1259 | { |
1529 | { |
1260 | sv = newSVsv (POPs); |
1530 | sv = newSVsv (POPs); |
1261 | FREETMPS; LEAVE; |
1531 | FREETMPS; LEAVE; |
1262 | return sv; |
1532 | return sv; |
1263 | } |
1533 | } |
1264 | |
1534 | |
|
|
1535 | SvREFCNT_inc (sv); |
1265 | FREETMPS; LEAVE; |
1536 | FREETMPS; LEAVE; |
1266 | } |
1537 | } |
1267 | } |
1538 | } |
1268 | |
1539 | |
1269 | if (dec->json.cb_object) |
1540 | if (dec->json.cb_object) |
1270 | { |
1541 | { |
1271 | dSP; |
1542 | dSP; |
1272 | int count; |
1543 | int count; |
1273 | |
1544 | |
1274 | ENTER; SAVETMPS; PUSHMARK (SP); |
1545 | ENTER; SAVETMPS; |
|
|
1546 | SAVESTACK_POS (); |
|
|
1547 | PUSHMARK (SP); |
1275 | XPUSHs (sv_2mortal (sv)); |
1548 | XPUSHs (sv_2mortal (sv)); |
1276 | |
1549 | |
1277 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
1550 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
1278 | |
1551 | |
1279 | if (count == 1) |
1552 | if (count == 1) |
… | |
… | |
1295 | DEC_DEC_DEPTH; |
1568 | DEC_DEC_DEPTH; |
1296 | return 0; |
1569 | return 0; |
1297 | } |
1570 | } |
1298 | |
1571 | |
1299 | static SV * |
1572 | static SV * |
|
|
1573 | decode_tag (dec_t *dec) |
|
|
1574 | { |
|
|
1575 | SV *tag = 0; |
|
|
1576 | SV *val = 0; |
|
|
1577 | |
|
|
1578 | if (!(dec->json.flags & F_ALLOW_TAGS)) |
|
|
1579 | ERR ("malformed JSON string, neither array, object, number, string or atom"); |
|
|
1580 | |
|
|
1581 | ++dec->cur; |
|
|
1582 | |
|
|
1583 | decode_ws (dec); |
|
|
1584 | |
|
|
1585 | tag = decode_sv (dec); |
|
|
1586 | if (!tag) |
|
|
1587 | goto fail; |
|
|
1588 | |
|
|
1589 | if (!SvPOK (tag)) |
|
|
1590 | ERR ("malformed JSON string, (tag) must be a string"); |
|
|
1591 | |
|
|
1592 | decode_ws (dec); |
|
|
1593 | |
|
|
1594 | if (*dec->cur != ')') |
|
|
1595 | ERR (") expected after tag"); |
|
|
1596 | |
|
|
1597 | ++dec->cur; |
|
|
1598 | |
|
|
1599 | decode_ws (dec); |
|
|
1600 | |
|
|
1601 | val = decode_sv (dec); |
|
|
1602 | if (!val) |
|
|
1603 | goto fail; |
|
|
1604 | |
|
|
1605 | if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV) |
|
|
1606 | ERR ("malformed JSON string, tag value must be an array"); |
|
|
1607 | |
|
|
1608 | { |
|
|
1609 | AV *av = (AV *)SvRV (val); |
|
|
1610 | int i, len = av_len (av) + 1; |
|
|
1611 | HV *stash = gv_stashsv (tag, 0); |
|
|
1612 | SV *sv; |
|
|
1613 | |
|
|
1614 | if (!stash) |
|
|
1615 | ERR ("cannot decode perl-object (package does not exist)"); |
|
|
1616 | |
|
|
1617 | GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); |
|
|
1618 | |
|
|
1619 | if (!method) |
|
|
1620 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
|
|
1621 | |
|
|
1622 | dSP; |
|
|
1623 | |
|
|
1624 | ENTER; SAVETMPS; |
|
|
1625 | PUSHMARK (SP); |
|
|
1626 | EXTEND (SP, len + 2); |
|
|
1627 | // we re-bless the reference to get overload and other niceties right |
|
|
1628 | PUSHs (tag); |
|
|
1629 | PUSHs (sv_json); |
|
|
1630 | |
|
|
1631 | for (i = 0; i < len; ++i) |
|
|
1632 | PUSHs (*av_fetch (av, i, 1)); |
|
|
1633 | |
|
|
1634 | PUTBACK; |
|
|
1635 | call_sv ((SV *)GvCV (method), G_SCALAR); |
|
|
1636 | SPAGAIN; |
|
|
1637 | |
|
|
1638 | SvREFCNT_dec (tag); |
|
|
1639 | SvREFCNT_dec (val); |
|
|
1640 | sv = SvREFCNT_inc (POPs); |
|
|
1641 | |
|
|
1642 | PUTBACK; |
|
|
1643 | |
|
|
1644 | FREETMPS; LEAVE; |
|
|
1645 | |
|
|
1646 | return sv; |
|
|
1647 | } |
|
|
1648 | |
|
|
1649 | fail: |
|
|
1650 | SvREFCNT_dec (tag); |
|
|
1651 | SvREFCNT_dec (val); |
|
|
1652 | return 0; |
|
|
1653 | } |
|
|
1654 | |
|
|
1655 | static SV * |
1300 | decode_sv (dec_t *dec) |
1656 | decode_sv (dec_t *dec) |
1301 | { |
1657 | { |
1302 | // the beauty of JSON: you need exactly one character lookahead |
1658 | // the beauty of JSON: you need exactly one character lookahead |
1303 | // to parse anything. |
1659 | // to parse everything. |
1304 | switch (*dec->cur) |
1660 | switch (*dec->cur) |
1305 | { |
1661 | { |
1306 | case '"': ++dec->cur; return decode_str (dec); |
1662 | case '"': ++dec->cur; return decode_str (dec); |
1307 | case '[': ++dec->cur; return decode_av (dec); |
1663 | case '[': ++dec->cur; return decode_av (dec); |
1308 | case '{': ++dec->cur; return decode_hv (dec); |
1664 | case '{': ++dec->cur; return decode_hv (dec); |
|
|
1665 | case '(': return decode_tag (dec); |
1309 | |
1666 | |
1310 | case '-': |
1667 | case '-': |
1311 | case '0': case '1': case '2': case '3': case '4': |
1668 | case '0': case '1': case '2': case '3': case '4': |
1312 | case '5': case '6': case '7': case '8': case '9': |
1669 | case '5': case '6': case '7': case '8': case '9': |
1313 | return decode_num (dec); |
1670 | return decode_num (dec); |
1314 | |
1671 | |
|
|
1672 | case 'f': |
|
|
1673 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
|
|
1674 | { |
|
|
1675 | dec->cur += 5; |
|
|
1676 | |
|
|
1677 | if (expect_false (!dec->json.v_false)) |
|
|
1678 | dec->json.v_false = GET_BOOL (false); |
|
|
1679 | |
|
|
1680 | return newSVsv (dec->json.v_false); |
|
|
1681 | } |
|
|
1682 | else |
|
|
1683 | ERR ("'false' expected"); |
|
|
1684 | |
|
|
1685 | break; |
|
|
1686 | |
1315 | case 't': |
1687 | case 't': |
1316 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1688 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1317 | { |
1689 | { |
1318 | dec->cur += 4; |
1690 | dec->cur += 4; |
1319 | #if JSON_SLOW |
1691 | |
1320 | json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); |
1692 | if (expect_false (!dec->json.v_true)) |
1321 | #endif |
1693 | dec->json.v_true = GET_BOOL (true); |
1322 | return SvREFCNT_inc (json_true); |
1694 | |
|
|
1695 | return newSVsv (dec->json.v_true); |
1323 | } |
1696 | } |
1324 | else |
1697 | else |
1325 | ERR ("'true' expected"); |
1698 | ERR ("'true' expected"); |
1326 | |
|
|
1327 | break; |
|
|
1328 | |
|
|
1329 | case 'f': |
|
|
1330 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
|
|
1331 | { |
|
|
1332 | dec->cur += 5; |
|
|
1333 | #if JSON_SLOW |
|
|
1334 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
|
|
1335 | #endif |
|
|
1336 | return SvREFCNT_inc (json_false); |
|
|
1337 | } |
|
|
1338 | else |
|
|
1339 | ERR ("'false' expected"); |
|
|
1340 | |
1699 | |
1341 | break; |
1700 | break; |
1342 | |
1701 | |
1343 | case 'n': |
1702 | case 'n': |
1344 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
1703 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
… | |
… | |
1350 | ERR ("'null' expected"); |
1709 | ERR ("'null' expected"); |
1351 | |
1710 | |
1352 | break; |
1711 | break; |
1353 | |
1712 | |
1354 | default: |
1713 | default: |
1355 | ERR ("malformed JSON string, neither array, object, number, string or atom"); |
1714 | ERR ("malformed JSON string, neither tag, array, object, number, string or atom"); |
1356 | break; |
1715 | break; |
1357 | } |
1716 | } |
1358 | |
1717 | |
1359 | fail: |
1718 | fail: |
1360 | return 0; |
1719 | return 0; |
1361 | } |
1720 | } |
1362 | |
1721 | |
1363 | static SV * |
1722 | static SV * |
1364 | decode_json (SV *string, JSON *json, UV *offset_return) |
1723 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1365 | { |
1724 | { |
1366 | dec_t dec; |
1725 | dec_t dec; |
1367 | UV offset; |
|
|
1368 | SV *sv; |
1726 | SV *sv; |
1369 | |
1727 | |
|
|
1728 | /* work around bugs in 5.10 where manipulating magic values |
|
|
1729 | * makes perl ignore the magic in subsequent accesses. |
|
|
1730 | * also make a copy of non-PV values, to get them into a clean |
|
|
1731 | * state (SvPV should do that, but it's buggy, see below). |
|
|
1732 | * |
|
|
1733 | * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16), |
|
|
1734 | * as reported by Reini Urban. |
|
|
1735 | */ |
1370 | SvGETMAGIC (string); |
1736 | /*SvGETMAGIC (string);*/ |
|
|
1737 | if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string)) |
|
|
1738 | string = sv_2mortal (newSVsv (string)); |
|
|
1739 | |
1371 | SvUPGRADE (string, SVt_PV); |
1740 | SvUPGRADE (string, SVt_PV); |
1372 | |
1741 | |
1373 | if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) |
1742 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
|
|
1743 | * assertion with -DDEBUGGING, although SvCUR is documented to |
|
|
1744 | * return the xpv_cur field which certainly exists after upgrading. |
|
|
1745 | * according to nicholas clark, calling SvPOK fixes this. |
|
|
1746 | * But it doesn't fix it, so try another workaround, call SvPV_nolen |
|
|
1747 | * and hope for the best. |
|
|
1748 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
|
|
1749 | * assertion business is seriously broken, try yet another workaround |
|
|
1750 | * for the broken -DDEBUGGING. |
|
|
1751 | */ |
|
|
1752 | { |
|
|
1753 | #ifdef DEBUGGING |
|
|
1754 | STRLEN offset = SvOK (string) ? sv_len (string) : 0; |
|
|
1755 | #else |
|
|
1756 | STRLEN offset = SvCUR (string); |
|
|
1757 | #endif |
|
|
1758 | |
|
|
1759 | if (offset > json->max_size && json->max_size) |
1374 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1760 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1375 | (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); |
1761 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
|
|
1762 | } |
1376 | |
1763 | |
1377 | if (json->flags & F_UTF8) |
1764 | if (DECODE_WANTS_OCTETS (json)) |
1378 | sv_utf8_downgrade (string, 0); |
1765 | sv_utf8_downgrade (string, 0); |
1379 | else |
1766 | else |
1380 | sv_utf8_upgrade (string); |
1767 | sv_utf8_upgrade (string); |
1381 | |
1768 | |
1382 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1769 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1383 | |
1770 | |
1384 | dec.json = *json; |
1771 | dec.json = *json; |
1385 | dec.cur = SvPVX (string); |
1772 | dec.cur = SvPVX (string); |
1386 | dec.end = SvEND (string); |
1773 | dec.end = SvEND (string); |
1387 | dec.err = 0; |
1774 | dec.err = 0; |
1388 | dec.depth = 0; |
1775 | dec.depth = 0; |
1389 | dec.maxdepth = DEC_DEPTH (dec.json.flags); |
|
|
1390 | |
1776 | |
1391 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1777 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1392 | dec.json.flags |= F_HOOK; |
1778 | dec.json.flags |= F_HOOK; |
1393 | |
1779 | |
1394 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1780 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1395 | |
1781 | |
1396 | decode_ws (&dec); |
1782 | decode_ws (&dec); |
1397 | sv = decode_sv (&dec); |
1783 | sv = decode_sv (&dec); |
1398 | |
1784 | |
1399 | if (!(offset_return || !sv)) |
1785 | if (offset_return) |
|
|
1786 | *offset_return = dec.cur - SvPVX (string); |
|
|
1787 | else if (sv) |
1400 | { |
1788 | { |
1401 | // check for trailing garbage |
1789 | // check for trailing garbage |
1402 | decode_ws (&dec); |
1790 | decode_ws (&dec); |
1403 | |
1791 | |
1404 | if (*dec.cur) |
1792 | if (*dec.cur) |
1405 | { |
1793 | { |
1406 | dec.err = "garbage after JSON object"; |
1794 | dec.err = "garbage after JSON object"; |
1407 | SvREFCNT_dec (sv); |
1795 | SvREFCNT_dec (sv); |
1408 | sv = 0; |
1796 | sv = 0; |
1409 | } |
1797 | } |
1410 | } |
|
|
1411 | |
|
|
1412 | if (offset_return || !sv) |
|
|
1413 | { |
|
|
1414 | offset = dec.json.flags & F_UTF8 |
|
|
1415 | ? dec.cur - SvPVX (string) |
|
|
1416 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
1417 | |
|
|
1418 | if (offset_return) |
|
|
1419 | *offset_return = offset; |
|
|
1420 | } |
1798 | } |
1421 | |
1799 | |
1422 | if (!sv) |
1800 | if (!sv) |
1423 | { |
1801 | { |
1424 | SV *uni = sv_newmortal (); |
1802 | SV *uni = sv_newmortal (); |
… | |
… | |
1430 | SAVEVPTR (PL_curcop); |
1808 | SAVEVPTR (PL_curcop); |
1431 | PL_curcop = &cop; |
1809 | PL_curcop = &cop; |
1432 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1810 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1433 | LEAVE; |
1811 | LEAVE; |
1434 | |
1812 | |
1435 | croak ("%s, at character offset %d [\"%s\"]", |
1813 | croak ("%s, at character offset %d (before \"%s\")", |
1436 | dec.err, |
1814 | dec.err, |
1437 | (int)offset, |
1815 | (int)ptr_to_index (string, dec.cur), |
1438 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1816 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1439 | } |
1817 | } |
1440 | |
1818 | |
1441 | sv = sv_2mortal (sv); |
1819 | sv = sv_2mortal (sv); |
1442 | |
1820 | |
1443 | if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1821 | if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv)) |
1444 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1822 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1445 | |
1823 | |
1446 | return sv; |
1824 | return sv; |
|
|
1825 | } |
|
|
1826 | |
|
|
1827 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1828 | // incremental parser |
|
|
1829 | |
|
|
1830 | static void |
|
|
1831 | incr_parse (JSON *self) |
|
|
1832 | { |
|
|
1833 | const char *p = SvPVX (self->incr_text) + self->incr_pos; |
|
|
1834 | |
|
|
1835 | // the state machine here is a bit convoluted and could be simplified a lot |
|
|
1836 | // but this would make it slower, so... |
|
|
1837 | |
|
|
1838 | for (;;) |
|
|
1839 | { |
|
|
1840 | //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 |
|
|
1841 | switch (self->incr_mode) |
|
|
1842 | { |
|
|
1843 | // only used for initial whitespace skipping |
|
|
1844 | case INCR_M_WS: |
|
|
1845 | for (;;) |
|
|
1846 | { |
|
|
1847 | if (*p > 0x20) |
|
|
1848 | { |
|
|
1849 | if (*p == '#') |
|
|
1850 | { |
|
|
1851 | self->incr_mode = INCR_M_C0; |
|
|
1852 | goto incr_m_c; |
|
|
1853 | } |
|
|
1854 | else |
|
|
1855 | { |
|
|
1856 | self->incr_mode = INCR_M_JSON; |
|
|
1857 | goto incr_m_json; |
|
|
1858 | } |
|
|
1859 | } |
|
|
1860 | else if (!*p) |
|
|
1861 | goto interrupt; |
|
|
1862 | |
|
|
1863 | ++p; |
|
|
1864 | } |
|
|
1865 | |
|
|
1866 | // skip a single char inside a string (for \\-processing) |
|
|
1867 | case INCR_M_BS: |
|
|
1868 | if (!*p) |
|
|
1869 | goto interrupt; |
|
|
1870 | |
|
|
1871 | ++p; |
|
|
1872 | self->incr_mode = INCR_M_STR; |
|
|
1873 | goto incr_m_str; |
|
|
1874 | |
|
|
1875 | // inside #-style comments |
|
|
1876 | case INCR_M_C0: |
|
|
1877 | case INCR_M_C1: |
|
|
1878 | incr_m_c: |
|
|
1879 | for (;;) |
|
|
1880 | { |
|
|
1881 | if (*p == '\n') |
|
|
1882 | { |
|
|
1883 | self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON; |
|
|
1884 | break; |
|
|
1885 | } |
|
|
1886 | else if (!*p) |
|
|
1887 | goto interrupt; |
|
|
1888 | |
|
|
1889 | ++p; |
|
|
1890 | } |
|
|
1891 | |
|
|
1892 | break; |
|
|
1893 | |
|
|
1894 | // inside a string |
|
|
1895 | case INCR_M_STR: |
|
|
1896 | incr_m_str: |
|
|
1897 | for (;;) |
|
|
1898 | { |
|
|
1899 | if (*p == '"') |
|
|
1900 | { |
|
|
1901 | ++p; |
|
|
1902 | self->incr_mode = INCR_M_JSON; |
|
|
1903 | |
|
|
1904 | if (!self->incr_nest) |
|
|
1905 | goto interrupt; |
|
|
1906 | |
|
|
1907 | goto incr_m_json; |
|
|
1908 | } |
|
|
1909 | else if (*p == '\\') |
|
|
1910 | { |
|
|
1911 | ++p; // "virtually" consumes character after \ |
|
|
1912 | |
|
|
1913 | if (!*p) // if at end of string we have to switch modes |
|
|
1914 | { |
|
|
1915 | self->incr_mode = INCR_M_BS; |
|
|
1916 | goto interrupt; |
|
|
1917 | } |
|
|
1918 | } |
|
|
1919 | else if (!*p) |
|
|
1920 | goto interrupt; |
|
|
1921 | |
|
|
1922 | ++p; |
|
|
1923 | } |
|
|
1924 | |
|
|
1925 | // after initial ws, outside string |
|
|
1926 | case INCR_M_JSON: |
|
|
1927 | incr_m_json: |
|
|
1928 | for (;;) |
|
|
1929 | { |
|
|
1930 | switch (*p++) |
|
|
1931 | { |
|
|
1932 | case 0: |
|
|
1933 | --p; |
|
|
1934 | goto interrupt; |
|
|
1935 | |
|
|
1936 | case 0x09: |
|
|
1937 | case 0x0a: |
|
|
1938 | case 0x0d: |
|
|
1939 | case 0x20: |
|
|
1940 | if (!self->incr_nest) |
|
|
1941 | { |
|
|
1942 | --p; // do not eat the whitespace, let the next round do it |
|
|
1943 | goto interrupt; |
|
|
1944 | } |
|
|
1945 | break; |
|
|
1946 | |
|
|
1947 | case '"': |
|
|
1948 | self->incr_mode = INCR_M_STR; |
|
|
1949 | goto incr_m_str; |
|
|
1950 | |
|
|
1951 | case '[': |
|
|
1952 | case '{': |
|
|
1953 | case '(': |
|
|
1954 | if (++self->incr_nest > self->max_depth) |
|
|
1955 | croak (ERR_NESTING_EXCEEDED); |
|
|
1956 | break; |
|
|
1957 | |
|
|
1958 | case ']': |
|
|
1959 | case '}': |
|
|
1960 | if (--self->incr_nest <= 0) |
|
|
1961 | goto interrupt; |
|
|
1962 | break; |
|
|
1963 | |
|
|
1964 | case ')': |
|
|
1965 | --self->incr_nest; |
|
|
1966 | break; |
|
|
1967 | |
|
|
1968 | case '#': |
|
|
1969 | self->incr_mode = INCR_M_C1; |
|
|
1970 | goto incr_m_c; |
|
|
1971 | } |
|
|
1972 | } |
|
|
1973 | } |
|
|
1974 | |
|
|
1975 | modechange: |
|
|
1976 | ; |
|
|
1977 | } |
|
|
1978 | |
|
|
1979 | interrupt: |
|
|
1980 | self->incr_pos = p - SvPVX (self->incr_text); |
|
|
1981 | //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D |
|
|
1982 | //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D |
1447 | } |
1983 | } |
1448 | |
1984 | |
1449 | ///////////////////////////////////////////////////////////////////////////// |
1985 | ///////////////////////////////////////////////////////////////////////////// |
1450 | // XS interface functions |
1986 | // XS interface functions |
1451 | |
1987 | |
… | |
… | |
1460 | i >= '0' && i <= '9' ? i - '0' |
1996 | i >= '0' && i <= '9' ? i - '0' |
1461 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1997 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1462 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1998 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1463 | : -1; |
1999 | : -1; |
1464 | |
2000 | |
1465 | json_stash = gv_stashpv ("JSON::XS" , 1); |
2001 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1466 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
2002 | bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); |
|
|
2003 | bool_false = get_bool ("Types::Serialiser::false"); |
|
|
2004 | bool_true = get_bool ("Types::Serialiser::true"); |
1467 | |
2005 | |
1468 | json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); |
2006 | sv_json = newSVpv ("JSON", 0); |
1469 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
2007 | SvREADONLY_on (sv_json); |
|
|
2008 | |
|
|
2009 | CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ |
1470 | } |
2010 | } |
1471 | |
2011 | |
1472 | PROTOTYPES: DISABLE |
2012 | PROTOTYPES: DISABLE |
1473 | |
2013 | |
1474 | void CLONE (...) |
2014 | void CLONE (...) |
1475 | CODE: |
2015 | CODE: |
|
|
2016 | // as long as these writes are atomic, the race should not matter |
|
|
2017 | // as existing threads either already use 0, or use the old value, |
|
|
2018 | // which is sitll correct for the initial thread. |
1476 | json_stash = 0; |
2019 | json_stash = 0; |
1477 | json_boolean_stash = 0; |
2020 | bool_stash = 0; |
|
|
2021 | bool_false = 0; |
|
|
2022 | bool_true = 0; |
1478 | |
2023 | |
1479 | void new (char *klass) |
2024 | void new (char *klass) |
1480 | PPCODE: |
2025 | PPCODE: |
1481 | { |
2026 | { |
1482 | SV *pv = NEWSV (0, sizeof (JSON)); |
2027 | SV *pv = NEWSV (0, sizeof (JSON)); |
1483 | SvPOK_only (pv); |
2028 | SvPOK_only (pv); |
1484 | Zero (SvPVX (pv), 1, JSON); |
2029 | json_init ((JSON *)SvPVX (pv)); |
1485 | ((JSON *)SvPVX (pv))->flags = F_DEFAULT; |
|
|
1486 | XPUSHs (sv_2mortal (sv_bless ( |
2030 | XPUSHs (sv_2mortal (sv_bless ( |
1487 | newRV_noinc (pv), |
2031 | newRV_noinc (pv), |
1488 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
2032 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
1489 | ))); |
2033 | ))); |
1490 | } |
2034 | } |
|
|
2035 | |
|
|
2036 | void boolean_values (JSON *self, SV *v_false = 0, SV *v_true = 0) |
|
|
2037 | PPCODE: |
|
|
2038 | self->v_false = newSVsv (v_false); |
|
|
2039 | self->v_true = newSVsv (v_true); |
|
|
2040 | XPUSHs (ST (0)); |
|
|
2041 | |
|
|
2042 | void get_boolean_values (JSON *self) |
|
|
2043 | PPCODE: |
|
|
2044 | if (self->v_false && self->v_true) |
|
|
2045 | { |
|
|
2046 | EXTEND (SP, 2); |
|
|
2047 | PUSHs (self->v_false); |
|
|
2048 | PUSHs (self->v_true); |
|
|
2049 | } |
1491 | |
2050 | |
1492 | void ascii (JSON *self, int enable = 1) |
2051 | void ascii (JSON *self, int enable = 1) |
1493 | ALIAS: |
2052 | ALIAS: |
1494 | ascii = F_ASCII |
2053 | ascii = F_ASCII |
1495 | latin1 = F_LATIN1 |
2054 | latin1 = F_LATIN1 |
… | |
… | |
1502 | allow_nonref = F_ALLOW_NONREF |
2061 | allow_nonref = F_ALLOW_NONREF |
1503 | shrink = F_SHRINK |
2062 | shrink = F_SHRINK |
1504 | allow_blessed = F_ALLOW_BLESSED |
2063 | allow_blessed = F_ALLOW_BLESSED |
1505 | convert_blessed = F_CONV_BLESSED |
2064 | convert_blessed = F_CONV_BLESSED |
1506 | relaxed = F_RELAXED |
2065 | relaxed = F_RELAXED |
|
|
2066 | allow_unknown = F_ALLOW_UNKNOWN |
|
|
2067 | allow_tags = F_ALLOW_TAGS |
1507 | PPCODE: |
2068 | PPCODE: |
1508 | { |
2069 | { |
1509 | if (enable) |
2070 | if (enable) |
1510 | self->flags |= ix; |
2071 | self->flags |= ix; |
1511 | else |
2072 | else |
… | |
… | |
1526 | get_allow_nonref = F_ALLOW_NONREF |
2087 | get_allow_nonref = F_ALLOW_NONREF |
1527 | get_shrink = F_SHRINK |
2088 | get_shrink = F_SHRINK |
1528 | get_allow_blessed = F_ALLOW_BLESSED |
2089 | get_allow_blessed = F_ALLOW_BLESSED |
1529 | get_convert_blessed = F_CONV_BLESSED |
2090 | get_convert_blessed = F_CONV_BLESSED |
1530 | get_relaxed = F_RELAXED |
2091 | get_relaxed = F_RELAXED |
|
|
2092 | get_allow_unknown = F_ALLOW_UNKNOWN |
|
|
2093 | get_allow_tags = F_ALLOW_TAGS |
1531 | PPCODE: |
2094 | PPCODE: |
1532 | XPUSHs (boolSV (self->flags & ix)); |
2095 | XPUSHs (boolSV (self->flags & ix)); |
1533 | |
2096 | |
1534 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
2097 | void max_depth (JSON *self, U32 max_depth = 0x80000000UL) |
1535 | PPCODE: |
2098 | PPCODE: |
1536 | { |
2099 | self->max_depth = max_depth; |
1537 | UV log2 = 0; |
|
|
1538 | |
|
|
1539 | if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
|
|
1540 | |
|
|
1541 | while ((1UL << log2) < max_depth) |
|
|
1542 | ++log2; |
|
|
1543 | |
|
|
1544 | self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
|
|
1545 | |
|
|
1546 | XPUSHs (ST (0)); |
2100 | XPUSHs (ST (0)); |
1547 | } |
|
|
1548 | |
2101 | |
1549 | U32 get_max_depth (JSON *self) |
2102 | U32 get_max_depth (JSON *self) |
1550 | CODE: |
2103 | CODE: |
1551 | RETVAL = DEC_DEPTH (self->flags); |
2104 | RETVAL = self->max_depth; |
1552 | OUTPUT: |
2105 | OUTPUT: |
1553 | RETVAL |
2106 | RETVAL |
1554 | |
2107 | |
1555 | void max_size (JSON *self, UV max_size = 0) |
2108 | void max_size (JSON *self, U32 max_size = 0) |
1556 | PPCODE: |
2109 | PPCODE: |
1557 | { |
2110 | self->max_size = max_size; |
1558 | UV log2 = 0; |
|
|
1559 | |
|
|
1560 | if (max_size > 0x80000000UL) max_size = 0x80000000UL; |
|
|
1561 | if (max_size == 1) max_size = 2; |
|
|
1562 | |
|
|
1563 | while ((1UL << log2) < max_size) |
|
|
1564 | ++log2; |
|
|
1565 | |
|
|
1566 | self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); |
|
|
1567 | |
|
|
1568 | XPUSHs (ST (0)); |
2111 | XPUSHs (ST (0)); |
1569 | } |
|
|
1570 | |
2112 | |
1571 | int get_max_size (JSON *self) |
2113 | int get_max_size (JSON *self) |
1572 | CODE: |
2114 | CODE: |
1573 | RETVAL = DEC_SIZE (self->flags); |
2115 | RETVAL = self->max_size; |
1574 | OUTPUT: |
2116 | OUTPUT: |
1575 | RETVAL |
2117 | RETVAL |
1576 | |
2118 | |
1577 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
2119 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1578 | PPCODE: |
2120 | PPCODE: |
… | |
… | |
1584 | } |
2126 | } |
1585 | |
2127 | |
1586 | void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) |
2128 | void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) |
1587 | PPCODE: |
2129 | PPCODE: |
1588 | { |
2130 | { |
1589 | if (!self->cb_sk_object) |
2131 | if (!self->cb_sk_object) |
1590 | self->cb_sk_object = newHV (); |
2132 | self->cb_sk_object = newHV (); |
1591 | |
2133 | |
1592 | if (SvOK (cb)) |
2134 | if (SvOK (cb)) |
1593 | hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
2135 | hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
1594 | else |
2136 | else |
… | |
… | |
1605 | XPUSHs (ST (0)); |
2147 | XPUSHs (ST (0)); |
1606 | } |
2148 | } |
1607 | |
2149 | |
1608 | void encode (JSON *self, SV *scalar) |
2150 | void encode (JSON *self, SV *scalar) |
1609 | PPCODE: |
2151 | PPCODE: |
1610 | XPUSHs (encode_json (scalar, self)); |
2152 | PUTBACK; scalar = encode_json (scalar, self); SPAGAIN; |
|
|
2153 | XPUSHs (scalar); |
1611 | |
2154 | |
1612 | void decode (JSON *self, SV *jsonstr) |
2155 | void decode (JSON *self, SV *jsonstr) |
1613 | PPCODE: |
2156 | PPCODE: |
1614 | XPUSHs (decode_json (jsonstr, self, 0)); |
2157 | PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN; |
|
|
2158 | XPUSHs (jsonstr); |
1615 | |
2159 | |
1616 | void decode_prefix (JSON *self, SV *jsonstr) |
2160 | void decode_prefix (JSON *self, SV *jsonstr) |
1617 | PPCODE: |
2161 | PPCODE: |
1618 | { |
2162 | { |
1619 | UV offset; |
2163 | SV *sv; |
|
|
2164 | STRLEN offset; |
|
|
2165 | PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; |
1620 | EXTEND (SP, 2); |
2166 | EXTEND (SP, 2); |
1621 | PUSHs (decode_json (jsonstr, self, &offset)); |
2167 | PUSHs (sv); |
1622 | PUSHs (sv_2mortal (newSVuv (offset))); |
2168 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset)))); |
|
|
2169 | } |
|
|
2170 | |
|
|
2171 | void incr_parse (JSON *self, SV *jsonstr = 0) |
|
|
2172 | PPCODE: |
|
|
2173 | { |
|
|
2174 | if (!self->incr_text) |
|
|
2175 | self->incr_text = newSVpvn ("", 0); |
|
|
2176 | |
|
|
2177 | /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */ |
|
|
2178 | if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text)) |
|
|
2179 | if (DECODE_WANTS_OCTETS (self)) |
|
|
2180 | { |
|
|
2181 | if (self->incr_pos) |
|
|
2182 | self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text), |
|
|
2183 | (U8 *)SvPVX (self->incr_text) + self->incr_pos); |
|
|
2184 | |
|
|
2185 | sv_utf8_downgrade (self->incr_text, 0); |
|
|
2186 | } |
|
|
2187 | else |
|
|
2188 | { |
|
|
2189 | sv_utf8_upgrade (self->incr_text); |
|
|
2190 | |
|
|
2191 | if (self->incr_pos) |
|
|
2192 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
|
|
2193 | - (U8 *)SvPVX (self->incr_text); |
|
|
2194 | } |
|
|
2195 | |
|
|
2196 | // append data, if any |
|
|
2197 | if (jsonstr) |
|
|
2198 | { |
|
|
2199 | /* make sure both strings have same encoding */ |
|
|
2200 | if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text)) |
|
|
2201 | if (SvUTF8 (jsonstr)) |
|
|
2202 | sv_utf8_downgrade (jsonstr, 0); |
|
|
2203 | else |
|
|
2204 | sv_utf8_upgrade (jsonstr); |
|
|
2205 | |
|
|
2206 | /* and then just blindly append */ |
|
|
2207 | { |
|
|
2208 | STRLEN len; |
|
|
2209 | const char *str = SvPV (jsonstr, len); |
|
|
2210 | STRLEN cur = SvCUR (self->incr_text); |
|
|
2211 | |
|
|
2212 | if (SvLEN (self->incr_text) - cur <= len) |
|
|
2213 | json_sv_grow (self->incr_text, cur, len); |
|
|
2214 | |
|
|
2215 | Move (str, SvEND (self->incr_text), len, char); |
|
|
2216 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
|
|
2217 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
|
|
2218 | } |
|
|
2219 | } |
|
|
2220 | |
|
|
2221 | if (GIMME_V != G_VOID) |
|
|
2222 | do |
|
|
2223 | { |
|
|
2224 | SV *sv; |
|
|
2225 | STRLEN offset; |
|
|
2226 | |
|
|
2227 | if (!INCR_DONE (self)) |
|
|
2228 | { |
|
|
2229 | incr_parse (self); |
|
|
2230 | |
|
|
2231 | if (self->incr_pos > self->max_size && self->max_size) |
|
|
2232 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
|
|
2233 | (unsigned long)self->incr_pos, (unsigned long)self->max_size); |
|
|
2234 | |
|
|
2235 | if (!INCR_DONE (self)) |
|
|
2236 | { |
|
|
2237 | // as an optimisation, do not accumulate white space in the incr buffer |
|
|
2238 | if (self->incr_mode == INCR_M_WS && self->incr_pos) |
|
|
2239 | { |
|
|
2240 | self->incr_pos = 0; |
|
|
2241 | SvCUR_set (self->incr_text, 0); |
|
|
2242 | } |
|
|
2243 | |
|
|
2244 | break; |
|
|
2245 | } |
|
|
2246 | } |
|
|
2247 | |
|
|
2248 | PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; |
|
|
2249 | XPUSHs (sv); |
|
|
2250 | |
|
|
2251 | self->incr_pos -= offset; |
|
|
2252 | self->incr_nest = 0; |
|
|
2253 | self->incr_mode = 0; |
|
|
2254 | |
|
|
2255 | sv_chop (self->incr_text, SvPVX (self->incr_text) + offset); |
|
|
2256 | } |
|
|
2257 | while (GIMME_V == G_ARRAY); |
|
|
2258 | } |
|
|
2259 | |
|
|
2260 | SV *incr_text (JSON *self) |
|
|
2261 | ATTRS: lvalue |
|
|
2262 | CODE: |
|
|
2263 | { |
|
|
2264 | if (self->incr_pos) |
|
|
2265 | croak ("incr_text can not be called when the incremental parser already started parsing"); |
|
|
2266 | |
|
|
2267 | RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef; |
|
|
2268 | } |
|
|
2269 | OUTPUT: |
|
|
2270 | RETVAL |
|
|
2271 | |
|
|
2272 | void incr_skip (JSON *self) |
|
|
2273 | CODE: |
|
|
2274 | { |
|
|
2275 | if (self->incr_pos) |
|
|
2276 | { |
|
|
2277 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos); |
|
|
2278 | self->incr_pos = 0; |
|
|
2279 | self->incr_nest = 0; |
|
|
2280 | self->incr_mode = 0; |
|
|
2281 | } |
|
|
2282 | } |
|
|
2283 | |
|
|
2284 | void incr_reset (JSON *self) |
|
|
2285 | CODE: |
|
|
2286 | { |
|
|
2287 | SvREFCNT_dec (self->incr_text); |
|
|
2288 | self->incr_text = 0; |
|
|
2289 | self->incr_pos = 0; |
|
|
2290 | self->incr_nest = 0; |
|
|
2291 | self->incr_mode = 0; |
1623 | } |
2292 | } |
1624 | |
2293 | |
1625 | void DESTROY (JSON *self) |
2294 | void DESTROY (JSON *self) |
1626 | CODE: |
2295 | CODE: |
|
|
2296 | SvREFCNT_dec (self->v_false); |
|
|
2297 | SvREFCNT_dec (self->v_true); |
1627 | SvREFCNT_dec (self->cb_sk_object); |
2298 | SvREFCNT_dec (self->cb_sk_object); |
1628 | SvREFCNT_dec (self->cb_object); |
2299 | SvREFCNT_dec (self->cb_object); |
|
|
2300 | SvREFCNT_dec (self->incr_text); |
1629 | |
2301 | |
1630 | PROTOTYPES: ENABLE |
2302 | PROTOTYPES: ENABLE |
1631 | |
2303 | |
1632 | void encode_json (SV *scalar) |
2304 | void encode_json (SV *scalar) |
1633 | PPCODE: |
2305 | PPCODE: |
1634 | { |
2306 | { |
1635 | JSON json = { F_DEFAULT | F_UTF8 }; |
2307 | JSON json; |
1636 | XPUSHs (encode_json (scalar, &json)); |
2308 | json_init (&json); |
|
|
2309 | json.flags |= F_UTF8; |
|
|
2310 | PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN; |
|
|
2311 | XPUSHs (scalar); |
1637 | } |
2312 | } |
1638 | |
2313 | |
1639 | void decode_json (SV *jsonstr) |
2314 | void decode_json (SV *jsonstr) |
1640 | PPCODE: |
2315 | PPCODE: |
1641 | { |
2316 | { |
1642 | JSON json = { F_DEFAULT | F_UTF8 }; |
2317 | JSON json; |
1643 | XPUSHs (decode_json (jsonstr, &json, 0)); |
2318 | json_init (&json); |
|
|
2319 | json.flags |= F_UTF8; |
|
|
2320 | PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN; |
|
|
2321 | XPUSHs (jsonstr); |
1644 | } |
2322 | } |
1645 | |
2323 | |