… | |
… | |
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 <limits.h> |
10 | #include <float.h> |
10 | #include <float.h> |
|
|
11 | #include <inttypes.h> |
11 | |
12 | |
12 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
13 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
13 | # define snprintf _snprintf // C compilers have this in stdio.h |
14 | # define snprintf _snprintf // C compilers have this in stdio.h |
14 | #endif |
15 | #endif |
15 | |
16 | |
16 | // 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 |
17 | // guarentees, though. if it breaks, you get to keep the pieces. |
18 | // guarantees, though. if it breaks, you get to keep the pieces. |
18 | #ifndef UTF8_MAXBYTES |
19 | #ifndef UTF8_MAXBYTES |
19 | # define UTF8_MAXBYTES 13 |
20 | # define UTF8_MAXBYTES 13 |
20 | #endif |
21 | #endif |
21 | |
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 |
22 | #define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2) |
35 | #define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3) |
23 | |
36 | |
24 | #define F_ASCII 0x00000001UL |
37 | #define F_ASCII 0x00000001UL |
25 | #define F_LATIN1 0x00000002UL |
38 | #define F_LATIN1 0x00000002UL |
26 | #define F_UTF8 0x00000004UL |
39 | #define F_UTF8 0x00000004UL |
27 | #define F_INDENT 0x00000008UL |
40 | #define F_INDENT 0x00000008UL |
… | |
… | |
32 | #define F_SHRINK 0x00000200UL |
45 | #define F_SHRINK 0x00000200UL |
33 | #define F_ALLOW_BLESSED 0x00000400UL |
46 | #define F_ALLOW_BLESSED 0x00000400UL |
34 | #define F_CONV_BLESSED 0x00000800UL |
47 | #define F_CONV_BLESSED 0x00000800UL |
35 | #define F_RELAXED 0x00001000UL |
48 | #define F_RELAXED 0x00001000UL |
36 | #define F_ALLOW_UNKNOWN 0x00002000UL |
49 | #define F_ALLOW_UNKNOWN 0x00002000UL |
|
|
50 | #define F_ALLOW_TAGS 0x00004000UL |
37 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
51 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
38 | |
52 | |
39 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
53 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
40 | |
54 | |
41 | #define INIT_SIZE 32 // initial scalar size to be allocated |
55 | #define INIT_SIZE 64 // initial scalar size to be allocated |
42 | #define INDENT_STEP 3 // spaces per indentation level |
56 | #define INDENT_STEP 3 // spaces per indentation level |
43 | |
57 | |
44 | #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) |
45 | |
61 | |
46 | #define SB do { |
62 | #define SB do { |
47 | #define SE } while (0) |
63 | #define SE } while (0) |
48 | |
64 | |
49 | #if __GNUC__ >= 3 |
65 | #if __GNUC__ >= 3 |
… | |
… | |
62 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
78 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
63 | |
79 | |
64 | #define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)" |
80 | #define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)" |
65 | |
81 | |
66 | #ifdef USE_ITHREADS |
82 | #ifdef USE_ITHREADS |
67 | # define JSON_SLOW 1 |
|
|
68 | # 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)) |
69 | #else |
86 | #else |
70 | # define JSON_SLOW 0 |
|
|
71 | # define JSON_STASH json_stash |
87 | # define JSON_STASH json_stash |
|
|
88 | # define BOOL_STASH bool_stash |
|
|
89 | # define GET_BOOL(value) bool_ ## value |
72 | #endif |
90 | #endif |
73 | |
91 | |
74 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
92 | // the amount of HEs to allocate on the stack, when sorting keys |
75 | 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; |
76 | |
98 | |
77 | enum { |
99 | enum { |
78 | INCR_M_WS = 0, // initial whitespace skipping, must be 0 |
100 | INCR_M_WS = 0, // initial whitespace skipping, must be 0 |
79 | INCR_M_STR, // inside string |
101 | INCR_M_STR, // inside string |
80 | INCR_M_BS, // inside backslash |
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 |
81 | INCR_M_JSON // outside anything, count nesting |
105 | INCR_M_JSON // outside anything, count nesting |
82 | }; |
106 | }; |
83 | |
107 | |
84 | #define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) |
108 | #define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) |
85 | |
109 | |
… | |
… | |
94 | // for the incremental parser |
118 | // for the incremental parser |
95 | SV *incr_text; // the source text so far |
119 | SV *incr_text; // the source text so far |
96 | STRLEN incr_pos; // the current offset into the text |
120 | STRLEN incr_pos; // the current offset into the text |
97 | int incr_nest; // {[]}-nesting level |
121 | int incr_nest; // {[]}-nesting level |
98 | unsigned char incr_mode; |
122 | unsigned char incr_mode; |
|
|
123 | |
|
|
124 | SV *v_false, *v_true; |
99 | } JSON; |
125 | } JSON; |
100 | |
126 | |
101 | INLINE void |
127 | INLINE void |
102 | json_init (JSON *json) |
128 | json_init (JSON *json) |
103 | { |
129 | { |
104 | Zero (json, 1, JSON); |
130 | static const JSON init = { F_ALLOW_NONREF, 512 }; |
105 | json->max_depth = 512; |
131 | |
|
|
132 | *json = init; |
106 | } |
133 | } |
107 | |
134 | |
108 | ///////////////////////////////////////////////////////////////////////////// |
135 | ///////////////////////////////////////////////////////////////////////////// |
109 | // utility functions |
136 | // utility functions |
110 | |
137 | |
… | |
… | |
132 | SvPV_renew (sv, SvCUR (sv) + 1); |
159 | SvPV_renew (sv, SvCUR (sv) + 1); |
133 | #endif |
160 | #endif |
134 | } |
161 | } |
135 | } |
162 | } |
136 | |
163 | |
|
|
164 | /* adds two STRLENs together, slow, and with paranoia */ |
|
|
165 | STRLEN |
|
|
166 | strlen_sum (STRLEN l1, STRLEN l2) |
|
|
167 | { |
|
|
168 | size_t sum = l1 + l2; |
|
|
169 | |
|
|
170 | if (sum < (size_t)l2 || sum != (size_t)(STRLEN)sum) |
|
|
171 | croak ("JSON::XS: string size overflow"); |
|
|
172 | |
|
|
173 | return sum; |
|
|
174 | } |
|
|
175 | |
|
|
176 | /* similar to SvGROW, but somewhat safer and guarantees exponential realloc strategy */ |
|
|
177 | static char * |
|
|
178 | json_sv_grow (SV *sv, size_t len1, size_t len2) |
|
|
179 | { |
|
|
180 | len1 = strlen_sum (len1, len2); |
|
|
181 | len1 = strlen_sum (len1, len1 >> 1); |
|
|
182 | |
|
|
183 | if (len1 > 4096 - 24) |
|
|
184 | len1 = (len1 | 4095) - 24; |
|
|
185 | |
|
|
186 | return SvGROW (sv, len1); |
|
|
187 | } |
|
|
188 | |
137 | // decode an utf-8 character and return it, or (UV)-1 in |
189 | // decode a utf-8 character and return it, or (UV)-1 in |
138 | // case of an error. |
190 | // case of an error. |
139 | // we special-case "safe" characters from U+80 .. U+7FF, |
191 | // we special-case "safe" characters from U+80 .. U+7FF, |
140 | // but use the very good perl function to parse anything else. |
192 | // but use the very good perl function to parse anything else. |
141 | // note that we never call this function for a ascii codepoints |
193 | // note that we never call this function for a ascii codepoints |
142 | INLINE UV |
194 | INLINE UV |
… | |
… | |
175 | *s++ = 0x80 | ( ch & 0x3f); |
227 | *s++ = 0x80 | ( ch & 0x3f); |
176 | |
228 | |
177 | return s; |
229 | return s; |
178 | } |
230 | } |
179 | |
231 | |
|
|
232 | // convert offset pointer to character index, sv must be string |
|
|
233 | static STRLEN |
|
|
234 | ptr_to_index (SV *sv, char *offset) |
|
|
235 | { |
|
|
236 | return SvUTF8 (sv) |
|
|
237 | ? utf8_distance (offset, SvPVX (sv)) |
|
|
238 | : offset - SvPVX (sv); |
|
|
239 | } |
|
|
240 | |
|
|
241 | ///////////////////////////////////////////////////////////////////////////// |
|
|
242 | // fp hell |
|
|
243 | |
|
|
244 | // scan a group of digits, and a trailing exponent |
|
|
245 | static void |
|
|
246 | json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth) |
|
|
247 | { |
|
|
248 | UV uaccum = 0; |
|
|
249 | int eaccum = 0; |
|
|
250 | |
|
|
251 | // if we recurse too deep, skip all remaining digits |
|
|
252 | // to avoid a stack overflow attack |
|
|
253 | if (expect_false (--maxdepth <= 0)) |
|
|
254 | while (((U8)*s - '0') < 10) |
|
|
255 | ++s; |
|
|
256 | |
|
|
257 | for (;;) |
|
|
258 | { |
|
|
259 | U8 dig = (U8)*s - '0'; |
|
|
260 | |
|
|
261 | if (expect_false (dig >= 10)) |
|
|
262 | { |
|
|
263 | if (dig == (U8)((U8)'.' - (U8)'0')) |
|
|
264 | { |
|
|
265 | ++s; |
|
|
266 | json_atof_scan1 (s, accum, expo, 1, maxdepth); |
|
|
267 | } |
|
|
268 | else if ((dig | ' ') == 'e' - '0') |
|
|
269 | { |
|
|
270 | int exp2 = 0; |
|
|
271 | int neg = 0; |
|
|
272 | |
|
|
273 | ++s; |
|
|
274 | |
|
|
275 | if (*s == '-') |
|
|
276 | { |
|
|
277 | ++s; |
|
|
278 | neg = 1; |
|
|
279 | } |
|
|
280 | else if (*s == '+') |
|
|
281 | ++s; |
|
|
282 | |
|
|
283 | while ((dig = (U8)*s - '0') < 10) |
|
|
284 | exp2 = exp2 * 10 + *s++ - '0'; |
|
|
285 | |
|
|
286 | *expo += neg ? -exp2 : exp2; |
|
|
287 | } |
|
|
288 | |
|
|
289 | break; |
|
|
290 | } |
|
|
291 | |
|
|
292 | ++s; |
|
|
293 | |
|
|
294 | uaccum = uaccum * 10 + dig; |
|
|
295 | ++eaccum; |
|
|
296 | |
|
|
297 | // if we have too many digits, then recurse for more |
|
|
298 | // we actually do this for rather few digits |
|
|
299 | if (uaccum >= (UV_MAX - 9) / 10) |
|
|
300 | { |
|
|
301 | if (postdp) *expo -= eaccum; |
|
|
302 | json_atof_scan1 (s, accum, expo, postdp, maxdepth); |
|
|
303 | if (postdp) *expo += eaccum; |
|
|
304 | |
|
|
305 | break; |
|
|
306 | } |
|
|
307 | } |
|
|
308 | |
|
|
309 | // this relies greatly on the quality of the pow () |
|
|
310 | // implementation of the platform, but a good |
|
|
311 | // implementation is hard to beat. |
|
|
312 | // (IEEE 754 conformant ones are required to be exact) |
|
|
313 | if (postdp) *expo -= eaccum; |
|
|
314 | *accum += uaccum * Perl_pow (10., *expo); |
|
|
315 | *expo += eaccum; |
|
|
316 | } |
|
|
317 | |
|
|
318 | static NV |
|
|
319 | json_atof (const char *s) |
|
|
320 | { |
|
|
321 | NV accum = 0.; |
|
|
322 | int expo = 0; |
|
|
323 | int neg = 0; |
|
|
324 | |
|
|
325 | if (*s == '-') |
|
|
326 | { |
|
|
327 | ++s; |
|
|
328 | neg = 1; |
|
|
329 | } |
|
|
330 | |
|
|
331 | // a recursion depth of ten gives us >>500 bits |
|
|
332 | json_atof_scan1 (s, &accum, &expo, 0, 10); |
|
|
333 | |
|
|
334 | return neg ? -accum : accum; |
|
|
335 | } |
|
|
336 | |
|
|
337 | // target of scalar reference is bool? -1 == nope, 0 == false, 1 == true |
|
|
338 | static int |
|
|
339 | ref_bool_type (SV *sv) |
|
|
340 | { |
|
|
341 | svtype svt = SvTYPE (sv); |
|
|
342 | |
|
|
343 | if (svt < SVt_PVAV) |
|
|
344 | { |
|
|
345 | STRLEN len = 0; |
|
|
346 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
347 | |
|
|
348 | if (len == 1) |
|
|
349 | if (*pv == '1') |
|
|
350 | return 1; |
|
|
351 | else if (*pv == '0') |
|
|
352 | return 0; |
|
|
353 | } |
|
|
354 | |
|
|
355 | return -1; |
|
|
356 | } |
|
|
357 | |
|
|
358 | // returns whether scalar is not a reference in the sense of allow_nonref |
|
|
359 | static int |
|
|
360 | json_nonref (SV *scalar) |
|
|
361 | { |
|
|
362 | if (!SvROK (scalar)) |
|
|
363 | return 1; |
|
|
364 | |
|
|
365 | scalar = SvRV (scalar); |
|
|
366 | |
|
|
367 | if (SvTYPE (scalar) >= SVt_PVMG) |
|
|
368 | { |
|
|
369 | if (SvSTASH (scalar) == bool_stash) |
|
|
370 | return 1; |
|
|
371 | |
|
|
372 | if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0) |
|
|
373 | return 1; |
|
|
374 | } |
|
|
375 | |
|
|
376 | return 0; |
|
|
377 | } |
|
|
378 | |
180 | ///////////////////////////////////////////////////////////////////////////// |
379 | ///////////////////////////////////////////////////////////////////////////// |
181 | // encoder |
380 | // encoder |
182 | |
381 | |
183 | // structure used for encoding JSON |
382 | // structure used for encoding JSON |
184 | typedef struct |
383 | typedef struct |
… | |
… | |
192 | } enc_t; |
391 | } enc_t; |
193 | |
392 | |
194 | INLINE void |
393 | INLINE void |
195 | need (enc_t *enc, STRLEN len) |
394 | need (enc_t *enc, STRLEN len) |
196 | { |
395 | { |
197 | if (expect_false (enc->cur + len >= enc->end)) |
396 | if (expect_false ((uintptr_t)(enc->end - enc->cur) < len)) |
198 | { |
397 | { |
199 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
398 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
200 | SvGROW (enc->sv, cur + len + 1); |
399 | char *buf = json_sv_grow (enc->sv, cur, len); |
201 | enc->cur = SvPVX (enc->sv) + cur; |
400 | enc->cur = buf + cur; |
202 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
401 | enc->end = buf + SvLEN (enc->sv) - 1; |
203 | } |
402 | } |
204 | } |
403 | } |
205 | |
404 | |
206 | INLINE void |
405 | INLINE void |
207 | encode_ch (enc_t *enc, char ch) |
406 | encode_ch (enc_t *enc, char ch) |
… | |
… | |
223 | |
422 | |
224 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
423 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
225 | { |
424 | { |
226 | if (expect_false (ch == '"')) // but with slow exceptions |
425 | if (expect_false (ch == '"')) // but with slow exceptions |
227 | { |
426 | { |
228 | need (enc, len += 1); |
427 | need (enc, len + 1); |
229 | *enc->cur++ = '\\'; |
428 | *enc->cur++ = '\\'; |
230 | *enc->cur++ = '"'; |
429 | *enc->cur++ = '"'; |
231 | } |
430 | } |
232 | else if (expect_false (ch == '\\')) |
431 | else if (expect_false (ch == '\\')) |
233 | { |
432 | { |
234 | need (enc, len += 1); |
433 | need (enc, len + 1); |
235 | *enc->cur++ = '\\'; |
434 | *enc->cur++ = '\\'; |
236 | *enc->cur++ = '\\'; |
435 | *enc->cur++ = '\\'; |
237 | } |
436 | } |
238 | else |
437 | else |
239 | *enc->cur++ = ch; |
438 | *enc->cur++ = ch; |
… | |
… | |
242 | } |
441 | } |
243 | else |
442 | else |
244 | { |
443 | { |
245 | switch (ch) |
444 | switch (ch) |
246 | { |
445 | { |
247 | case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; |
446 | case '\010': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; |
248 | case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; |
447 | case '\011': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; |
249 | case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; |
448 | case '\012': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; |
250 | case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; |
449 | case '\014': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; |
251 | case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; |
450 | case '\015': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; |
252 | |
451 | |
253 | default: |
452 | default: |
254 | { |
453 | { |
255 | STRLEN clen; |
454 | STRLEN clen; |
256 | UV uch; |
455 | UV uch; |
… | |
… | |
272 | if (uch >= 0x10000UL) |
471 | if (uch >= 0x10000UL) |
273 | { |
472 | { |
274 | if (uch >= 0x110000UL) |
473 | if (uch >= 0x110000UL) |
275 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
474 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
276 | |
475 | |
277 | need (enc, len += 11); |
476 | need (enc, len + 11); |
278 | sprintf (enc->cur, "\\u%04x\\u%04x", |
477 | sprintf (enc->cur, "\\u%04x\\u%04x", |
279 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
478 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
280 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
479 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
281 | enc->cur += 12; |
480 | enc->cur += 12; |
282 | } |
481 | } |
283 | else |
482 | else |
284 | { |
483 | { |
285 | static char hexdigit [16] = "0123456789abcdef"; |
|
|
286 | need (enc, len += 5); |
484 | need (enc, len + 5); |
287 | *enc->cur++ = '\\'; |
485 | *enc->cur++ = '\\'; |
288 | *enc->cur++ = 'u'; |
486 | *enc->cur++ = 'u'; |
289 | *enc->cur++ = hexdigit [ uch >> 12 ]; |
487 | *enc->cur++ = PL_hexdigit [ uch >> 12 ]; |
290 | *enc->cur++ = hexdigit [(uch >> 8) & 15]; |
488 | *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; |
291 | *enc->cur++ = hexdigit [(uch >> 4) & 15]; |
489 | *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; |
292 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
490 | *enc->cur++ = PL_hexdigit [(uch >> 0) & 15]; |
293 | } |
491 | } |
294 | |
492 | |
295 | str += clen; |
493 | str += clen; |
296 | } |
494 | } |
297 | else if (enc->json.flags & F_LATIN1) |
495 | else if (enc->json.flags & F_LATIN1) |
… | |
… | |
299 | *enc->cur++ = uch; |
497 | *enc->cur++ = uch; |
300 | str += clen; |
498 | str += clen; |
301 | } |
499 | } |
302 | else if (is_utf8) |
500 | else if (is_utf8) |
303 | { |
501 | { |
304 | need (enc, len += clen); |
502 | need (enc, len + clen); |
305 | do |
503 | do |
306 | { |
504 | { |
307 | *enc->cur++ = *str++; |
505 | *enc->cur++ = *str++; |
308 | } |
506 | } |
309 | while (--clen); |
507 | while (--clen); |
310 | } |
508 | } |
311 | else |
509 | else |
312 | { |
510 | { |
313 | need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
511 | need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
314 | enc->cur = encode_utf8 (enc->cur, uch); |
512 | enc->cur = encode_utf8 (enc->cur, uch); |
315 | ++str; |
513 | ++str; |
316 | } |
514 | } |
317 | } |
515 | } |
318 | } |
516 | } |
… | |
… | |
372 | |
570 | |
373 | if (enc->indent >= enc->json.max_depth) |
571 | if (enc->indent >= enc->json.max_depth) |
374 | croak (ERR_NESTING_EXCEEDED); |
572 | croak (ERR_NESTING_EXCEEDED); |
375 | |
573 | |
376 | encode_ch (enc, '['); |
574 | encode_ch (enc, '['); |
377 | |
575 | |
378 | if (len >= 0) |
576 | if (len >= 0) |
379 | { |
577 | { |
380 | encode_nl (enc); ++enc->indent; |
578 | encode_nl (enc); ++enc->indent; |
381 | |
579 | |
382 | for (i = 0; i <= len; ++i) |
580 | for (i = 0; i <= len; ++i) |
… | |
… | |
394 | encode_comma (enc); |
592 | encode_comma (enc); |
395 | } |
593 | } |
396 | |
594 | |
397 | encode_nl (enc); --enc->indent; encode_indent (enc); |
595 | encode_nl (enc); --enc->indent; encode_indent (enc); |
398 | } |
596 | } |
399 | |
597 | |
400 | encode_ch (enc, ']'); |
598 | encode_ch (enc, ']'); |
401 | } |
599 | } |
402 | |
600 | |
403 | static void |
601 | static void |
404 | encode_hk (enc_t *enc, HE *he) |
602 | encode_hk (enc_t *enc, HE *he) |
… | |
… | |
408 | if (HeKLEN (he) == HEf_SVKEY) |
606 | if (HeKLEN (he) == HEf_SVKEY) |
409 | { |
607 | { |
410 | SV *sv = HeSVKEY (he); |
608 | SV *sv = HeSVKEY (he); |
411 | STRLEN len; |
609 | STRLEN len; |
412 | char *str; |
610 | char *str; |
413 | |
611 | |
414 | SvGETMAGIC (sv); |
612 | SvGETMAGIC (sv); |
415 | str = SvPV (sv, len); |
613 | str = SvPV (sv, len); |
416 | |
614 | |
417 | encode_str (enc, str, len, SvUTF8 (sv)); |
615 | encode_str (enc, str, len, SvUTF8 (sv)); |
418 | } |
616 | } |
… | |
… | |
461 | |
659 | |
462 | encode_ch (enc, '{'); |
660 | encode_ch (enc, '{'); |
463 | |
661 | |
464 | // for canonical output we have to sort by keys first |
662 | // for canonical output we have to sort by keys first |
465 | // actually, this is mostly due to the stupid so-called |
663 | // actually, this is mostly due to the stupid so-called |
466 | // security workaround added somewhere in 5.8.x. |
664 | // security workaround added somewhere in 5.8.x |
467 | // that randomises hash orderings |
665 | // that randomises hash orderings |
468 | if (enc->json.flags & F_CANONICAL) |
666 | if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv)) |
469 | { |
667 | { |
470 | int count = hv_iterinit (hv); |
668 | int count = hv_iterinit (hv); |
471 | |
669 | |
472 | if (SvMAGICAL (hv)) |
670 | if (SvMAGICAL (hv)) |
473 | { |
671 | { |
… | |
… | |
484 | } |
682 | } |
485 | |
683 | |
486 | if (count) |
684 | if (count) |
487 | { |
685 | { |
488 | int i, fast = 1; |
686 | int i, fast = 1; |
489 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
687 | HE *hes_stack [STACK_HES]; |
490 | HE **hes = _alloca (count * sizeof (HE)); |
688 | HE **hes = hes_stack; |
491 | #else |
689 | |
492 | HE *hes [count]; // if your compiler dies here, you need to enable C99 mode |
690 | // allocate larger arrays on the heap |
493 | #endif |
691 | if (count > STACK_HES) |
|
|
692 | { |
|
|
693 | SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes))); |
|
|
694 | hes = (HE **)SvPVX (sv); |
|
|
695 | } |
494 | |
696 | |
495 | i = 0; |
697 | i = 0; |
496 | while ((he = hv_iternext (hv))) |
698 | while ((he = hv_iternext (hv))) |
497 | { |
699 | { |
498 | hes [i++] = he; |
700 | hes [i++] = he; |
… | |
… | |
567 | // encode objects, arrays and special \0=false and \1=true values. |
769 | // encode objects, arrays and special \0=false and \1=true values. |
568 | static void |
770 | static void |
569 | encode_rv (enc_t *enc, SV *sv) |
771 | encode_rv (enc_t *enc, SV *sv) |
570 | { |
772 | { |
571 | svtype svt; |
773 | svtype svt; |
|
|
774 | GV *method; |
572 | |
775 | |
573 | SvGETMAGIC (sv); |
776 | SvGETMAGIC (sv); |
574 | svt = SvTYPE (sv); |
777 | svt = SvTYPE (sv); |
575 | |
778 | |
576 | if (expect_false (SvOBJECT (sv))) |
779 | if (expect_false (SvOBJECT (sv))) |
577 | { |
780 | { |
578 | HV *stash = !JSON_SLOW || json_boolean_stash |
781 | HV *stash = SvSTASH (sv); |
579 | ? json_boolean_stash |
|
|
580 | : gv_stashpv ("JSON::XS::Boolean", 1); |
|
|
581 | |
782 | |
582 | if (SvSTASH (sv) == stash) |
783 | if (stash == bool_stash) |
583 | { |
784 | { |
584 | if (SvIV (sv)) |
|
|
585 | encode_str (enc, "true", 4, 0); |
785 | if (SvIV (sv)) encode_str (enc, "true" , 4, 0); |
586 | else |
|
|
587 | encode_str (enc, "false", 5, 0); |
786 | else encode_str (enc, "false", 5, 0); |
588 | } |
787 | } |
|
|
788 | else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0))) |
|
|
789 | { |
|
|
790 | int count; |
|
|
791 | dSP; |
|
|
792 | |
|
|
793 | ENTER; SAVETMPS; |
|
|
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 | if (count) |
|
|
816 | { |
|
|
817 | int i; |
|
|
818 | |
|
|
819 | for (i = 0; i < count - 1; ++i) |
|
|
820 | { |
|
|
821 | encode_sv (enc, SP[i + 1 - count]); |
|
|
822 | encode_ch (enc, ','); |
|
|
823 | } |
|
|
824 | |
|
|
825 | encode_sv (enc, TOPs); |
|
|
826 | SP -= count; |
|
|
827 | } |
|
|
828 | |
|
|
829 | encode_ch (enc, ']'); |
|
|
830 | |
|
|
831 | FREETMPS; LEAVE; |
|
|
832 | } |
|
|
833 | else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0))) |
|
|
834 | { |
|
|
835 | dSP; |
|
|
836 | |
|
|
837 | ENTER; SAVETMPS; |
|
|
838 | PUSHMARK (SP); |
|
|
839 | // we re-bless the reference to get overload and other niceties right |
|
|
840 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
|
|
841 | |
|
|
842 | // calling with G_SCALAR ensures that we always get a 1 return value |
|
|
843 | PUTBACK; |
|
|
844 | call_sv ((SV *)GvCV (method), G_SCALAR); |
|
|
845 | SPAGAIN; |
|
|
846 | |
|
|
847 | // catch this surprisingly common error |
|
|
848 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
849 | croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
850 | |
|
|
851 | sv = POPs; |
|
|
852 | PUTBACK; |
|
|
853 | |
|
|
854 | encode_sv (enc, sv); |
|
|
855 | |
|
|
856 | FREETMPS; LEAVE; |
|
|
857 | } |
|
|
858 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
859 | encode_str (enc, "null", 4, 0); |
589 | else |
860 | else |
590 | { |
|
|
591 | #if 0 |
|
|
592 | if (0 && sv_derived_from (rv, "JSON::Literal")) |
|
|
593 | { |
|
|
594 | // not yet |
|
|
595 | } |
|
|
596 | #endif |
|
|
597 | if (enc->json.flags & F_CONV_BLESSED) |
|
|
598 | { |
|
|
599 | // we re-bless the reference to get overload and other niceties right |
|
|
600 | GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0); |
|
|
601 | |
|
|
602 | if (to_json) |
|
|
603 | { |
|
|
604 | dSP; |
|
|
605 | |
|
|
606 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
607 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); |
|
|
608 | |
|
|
609 | // calling with G_SCALAR ensures that we always get a 1 return value |
|
|
610 | PUTBACK; |
|
|
611 | call_sv ((SV *)GvCV (to_json), G_SCALAR); |
|
|
612 | SPAGAIN; |
|
|
613 | |
|
|
614 | // catch this surprisingly common error |
|
|
615 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
616 | croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
617 | |
|
|
618 | sv = POPs; |
|
|
619 | PUTBACK; |
|
|
620 | |
|
|
621 | encode_sv (enc, sv); |
|
|
622 | |
|
|
623 | FREETMPS; LEAVE; |
|
|
624 | } |
|
|
625 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
626 | encode_str (enc, "null", 4, 0); |
|
|
627 | else |
|
|
628 | croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it", |
|
|
629 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
630 | } |
|
|
631 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
632 | encode_str (enc, "null", 4, 0); |
|
|
633 | else |
|
|
634 | croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", |
861 | croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)", |
635 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
862 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
636 | } |
|
|
637 | } |
863 | } |
638 | else if (svt == SVt_PVHV) |
864 | else if (svt == SVt_PVHV) |
639 | encode_hv (enc, (HV *)sv); |
865 | encode_hv (enc, (HV *)sv); |
640 | else if (svt == SVt_PVAV) |
866 | else if (svt == SVt_PVAV) |
641 | encode_av (enc, (AV *)sv); |
867 | encode_av (enc, (AV *)sv); |
642 | else if (svt < SVt_PVAV) |
868 | else if (svt < SVt_PVAV) |
643 | { |
869 | { |
644 | STRLEN len = 0; |
870 | int bool_type = ref_bool_type (sv); |
645 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
646 | |
871 | |
647 | if (len == 1 && *pv == '1') |
872 | if (bool_type == 1) |
648 | encode_str (enc, "true", 4, 0); |
873 | encode_str (enc, "true", 4, 0); |
649 | else if (len == 1 && *pv == '0') |
874 | else if (bool_type == 0) |
650 | encode_str (enc, "false", 5, 0); |
875 | encode_str (enc, "false", 5, 0); |
651 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
876 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
652 | encode_str (enc, "null", 4, 0); |
877 | encode_str (enc, "null", 4, 0); |
653 | else |
878 | else |
654 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
879 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
… | |
… | |
717 | } |
942 | } |
718 | else |
943 | else |
719 | { |
944 | { |
720 | // large integer, use the (rather slow) snprintf way. |
945 | // large integer, use the (rather slow) snprintf way. |
721 | need (enc, IVUV_MAXCHARS); |
946 | need (enc, IVUV_MAXCHARS); |
722 | enc->cur += |
947 | enc->cur += |
723 | SvIsUV(sv) |
948 | SvIsUV(sv) |
724 | ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) |
949 | ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) |
725 | : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); |
950 | : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); |
726 | } |
951 | } |
727 | } |
952 | } |
728 | else if (SvROK (sv)) |
953 | else if (SvROK (sv)) |
729 | encode_rv (enc, SvRV (sv)); |
954 | encode_rv (enc, SvRV (sv)); |
730 | else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) |
955 | else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) |
731 | encode_str (enc, "null", 4, 0); |
956 | encode_str (enc, "null", 4, 0); |
732 | else |
957 | else |
733 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
958 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data", |
734 | SvPV_nolen (sv), SvFLAGS (sv)); |
959 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
735 | } |
960 | } |
736 | |
961 | |
737 | static SV * |
962 | static SV * |
738 | encode_json (SV *scalar, JSON *json) |
963 | encode_json (SV *scalar, JSON *json) |
739 | { |
964 | { |
740 | enc_t enc; |
965 | enc_t enc; |
741 | |
966 | |
742 | if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
967 | if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar)) |
743 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
968 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
744 | |
969 | |
745 | enc.json = *json; |
970 | enc.json = *json; |
746 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
971 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
747 | enc.cur = SvPVX (enc.sv); |
972 | enc.cur = SvPVX (enc.sv); |
… | |
… | |
751 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
976 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
752 | : 0x110000UL; |
977 | : 0x110000UL; |
753 | |
978 | |
754 | SvPOK_only (enc.sv); |
979 | SvPOK_only (enc.sv); |
755 | encode_sv (&enc, scalar); |
980 | encode_sv (&enc, scalar); |
|
|
981 | encode_nl (&enc); |
756 | |
982 | |
757 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
983 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
758 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
984 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
759 | |
985 | |
760 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
986 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
… | |
… | |
940 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
1166 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
941 | *cur++ = ch; |
1167 | *cur++ = ch; |
942 | else if (ch >= 0x80) |
1168 | else if (ch >= 0x80) |
943 | { |
1169 | { |
944 | STRLEN clen; |
1170 | STRLEN clen; |
945 | UV uch; |
|
|
946 | |
1171 | |
947 | --dec_cur; |
1172 | --dec_cur; |
948 | |
1173 | |
949 | uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
1174 | decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
950 | if (clen == (STRLEN)-1) |
1175 | if (clen == (STRLEN)-1) |
951 | ERR ("malformed UTF-8 character in JSON string"); |
1176 | ERR ("malformed UTF-8 character in JSON string"); |
952 | |
1177 | |
953 | do |
1178 | do |
954 | *cur++ = *dec_cur++; |
1179 | *cur++ = *dec_cur++; |
955 | while (--clen); |
1180 | while (--clen); |
956 | |
1181 | |
957 | utf8 = 1; |
1182 | utf8 = 1; |
958 | } |
1183 | } |
|
|
1184 | else if (ch == '\t' && dec->json.flags & F_RELAXED) |
|
|
1185 | *cur++ = ch; |
959 | else |
1186 | else |
960 | { |
1187 | { |
961 | --dec_cur; |
1188 | --dec_cur; |
962 | |
1189 | |
963 | if (!ch) |
1190 | if (!ch) |
… | |
… | |
971 | { |
1198 | { |
972 | STRLEN len = cur - buf; |
1199 | STRLEN len = cur - buf; |
973 | |
1200 | |
974 | if (sv) |
1201 | if (sv) |
975 | { |
1202 | { |
976 | SvGROW (sv, SvCUR (sv) + len + 1); |
1203 | STRLEN cur = SvCUR (sv); |
|
|
1204 | |
|
|
1205 | if (SvLEN (sv) - cur <= len) |
|
|
1206 | json_sv_grow (sv, cur, len); |
|
|
1207 | |
977 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
1208 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
978 | SvCUR_set (sv, SvCUR (sv) + len); |
1209 | SvCUR_set (sv, SvCUR (sv) + len); |
979 | } |
1210 | } |
980 | else |
1211 | else |
981 | sv = newSVpvn (buf, len); |
1212 | sv = newSVpvn (buf, len); |
… | |
… | |
1072 | |
1303 | |
1073 | // special case the rather common 1..5-digit-int case |
1304 | // special case the rather common 1..5-digit-int case |
1074 | if (*start == '-') |
1305 | if (*start == '-') |
1075 | switch (len) |
1306 | switch (len) |
1076 | { |
1307 | { |
1077 | case 2: return newSViv (-( start [1] - '0' * 1)); |
1308 | case 2: return newSViv (-(IV)( start [1] - '0' * 1)); |
1078 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
1309 | case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11)); |
1079 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
1310 | case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
1080 | case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
1311 | case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
1081 | case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
1312 | case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
1082 | } |
1313 | } |
1083 | else |
1314 | else |
1084 | switch (len) |
1315 | switch (len) |
1085 | { |
1316 | { |
1086 | case 1: return newSViv ( start [0] - '0' * 1); |
1317 | case 1: return newSViv ( start [0] - '0' * 1); |
1087 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1318 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1088 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
1319 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
1089 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
1320 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
1090 | case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
1321 | case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
1091 | } |
1322 | } |
1092 | |
1323 | |
1093 | { |
1324 | { |
1094 | UV uv; |
1325 | UV uv; |
1095 | int numtype = grok_number (start, len, &uv); |
1326 | int numtype = grok_number (start, len, &uv); |
… | |
… | |
1104 | } |
1335 | } |
1105 | |
1336 | |
1106 | len -= *start == '-' ? 1 : 0; |
1337 | len -= *start == '-' ? 1 : 0; |
1107 | |
1338 | |
1108 | // does not fit into IV or UV, try NV |
1339 | // does not fit into IV or UV, try NV |
1109 | if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) |
1340 | if (len <= NV_DIG) |
1110 | #if defined (LDBL_DIG) |
|
|
1111 | || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len) |
|
|
1112 | #endif |
|
|
1113 | ) |
|
|
1114 | // fits into NV without loss of precision |
1341 | // fits into NV without loss of precision |
1115 | return newSVnv (Atof (start)); |
1342 | return newSVnv (json_atof (start)); |
1116 | |
1343 | |
1117 | // everything else fails, convert it to a string |
1344 | // everything else fails, convert it to a string |
1118 | return newSVpvn (start, dec->cur - start); |
1345 | return newSVpvn (start, dec->cur - start); |
1119 | } |
1346 | } |
1120 | |
1347 | |
1121 | // loss of precision here |
1348 | // loss of precision here |
1122 | return newSVnv (Atof (start)); |
1349 | return newSVnv (json_atof (start)); |
1123 | |
1350 | |
1124 | fail: |
1351 | fail: |
1125 | return 0; |
1352 | return 0; |
1126 | } |
1353 | } |
1127 | |
1354 | |
… | |
… | |
1151 | if (*dec->cur == ']') |
1378 | if (*dec->cur == ']') |
1152 | { |
1379 | { |
1153 | ++dec->cur; |
1380 | ++dec->cur; |
1154 | break; |
1381 | break; |
1155 | } |
1382 | } |
1156 | |
1383 | |
1157 | if (*dec->cur != ',') |
1384 | if (*dec->cur != ',') |
1158 | ERR (", or ] expected while parsing array"); |
1385 | ERR (", or ] expected while parsing array"); |
1159 | |
1386 | |
1160 | ++dec->cur; |
1387 | ++dec->cur; |
1161 | |
1388 | |
… | |
… | |
1285 | |
1512 | |
1286 | hv_iterinit (hv); |
1513 | hv_iterinit (hv); |
1287 | he = hv_iternext (hv); |
1514 | he = hv_iternext (hv); |
1288 | hv_iterinit (hv); |
1515 | hv_iterinit (hv); |
1289 | |
1516 | |
1290 | // the next line creates a mortal sv each time its called. |
1517 | // the next line creates a mortal sv each time it's called. |
1291 | // might want to optimise this for common cases. |
1518 | // might want to optimise this for common cases. |
1292 | cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); |
1519 | cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); |
1293 | |
1520 | |
1294 | if (cb) |
1521 | if (cb) |
1295 | { |
1522 | { |
1296 | dSP; |
1523 | dSP; |
1297 | int count; |
1524 | int count; |
1298 | |
1525 | |
1299 | ENTER; SAVETMPS; PUSHMARK (SP); |
1526 | ENTER; SAVETMPS; |
|
|
1527 | PUSHMARK (SP); |
1300 | XPUSHs (HeVAL (he)); |
1528 | XPUSHs (HeVAL (he)); |
|
|
1529 | sv_2mortal (sv); |
1301 | |
1530 | |
1302 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1531 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1303 | |
1532 | |
1304 | if (count == 1) |
1533 | if (count == 1) |
1305 | { |
1534 | { |
1306 | sv = newSVsv (POPs); |
1535 | sv = newSVsv (POPs); |
1307 | FREETMPS; LEAVE; |
1536 | FREETMPS; LEAVE; |
1308 | return sv; |
1537 | return sv; |
1309 | } |
1538 | } |
|
|
1539 | else if (count) |
|
|
1540 | croak ("filter_json_single_key_object callbacks must not return more than one scalar"); |
1310 | |
1541 | |
|
|
1542 | SvREFCNT_inc (sv); |
1311 | FREETMPS; LEAVE; |
1543 | FREETMPS; LEAVE; |
1312 | } |
1544 | } |
1313 | } |
1545 | } |
1314 | |
1546 | |
1315 | if (dec->json.cb_object) |
1547 | if (dec->json.cb_object) |
1316 | { |
1548 | { |
1317 | dSP; |
1549 | dSP; |
1318 | int count; |
1550 | int count; |
1319 | |
1551 | |
1320 | ENTER; SAVETMPS; PUSHMARK (SP); |
1552 | ENTER; SAVETMPS; |
|
|
1553 | PUSHMARK (SP); |
1321 | XPUSHs (sv_2mortal (sv)); |
1554 | XPUSHs (sv_2mortal (sv)); |
1322 | |
1555 | |
1323 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
1556 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
1324 | |
1557 | |
1325 | if (count == 1) |
1558 | if (count == 1) |
1326 | { |
1559 | { |
1327 | sv = newSVsv (POPs); |
1560 | sv = newSVsv (POPs); |
1328 | FREETMPS; LEAVE; |
1561 | FREETMPS; LEAVE; |
1329 | return sv; |
1562 | return sv; |
1330 | } |
1563 | } |
|
|
1564 | else if (count) |
|
|
1565 | croak ("filter_json_object callbacks must not return more than one scalar"); |
1331 | |
1566 | |
1332 | SvREFCNT_inc (sv); |
1567 | SvREFCNT_inc (sv); |
1333 | FREETMPS; LEAVE; |
1568 | FREETMPS; LEAVE; |
1334 | } |
1569 | } |
1335 | } |
1570 | } |
… | |
… | |
1341 | DEC_DEC_DEPTH; |
1576 | DEC_DEC_DEPTH; |
1342 | return 0; |
1577 | return 0; |
1343 | } |
1578 | } |
1344 | |
1579 | |
1345 | static SV * |
1580 | static SV * |
|
|
1581 | decode_tag (dec_t *dec) |
|
|
1582 | { |
|
|
1583 | SV *tag = 0; |
|
|
1584 | SV *val = 0; |
|
|
1585 | |
|
|
1586 | if (!(dec->json.flags & F_ALLOW_TAGS)) |
|
|
1587 | ERR ("malformed JSON string, neither array, object, number, string or atom"); |
|
|
1588 | |
|
|
1589 | ++dec->cur; |
|
|
1590 | |
|
|
1591 | decode_ws (dec); |
|
|
1592 | |
|
|
1593 | tag = decode_sv (dec); |
|
|
1594 | if (!tag) |
|
|
1595 | goto fail; |
|
|
1596 | |
|
|
1597 | if (!SvPOK (tag)) |
|
|
1598 | ERR ("malformed JSON string, (tag) must be a string"); |
|
|
1599 | |
|
|
1600 | decode_ws (dec); |
|
|
1601 | |
|
|
1602 | if (*dec->cur != ')') |
|
|
1603 | ERR (") expected after tag"); |
|
|
1604 | |
|
|
1605 | ++dec->cur; |
|
|
1606 | |
|
|
1607 | decode_ws (dec); |
|
|
1608 | |
|
|
1609 | val = decode_sv (dec); |
|
|
1610 | if (!val) |
|
|
1611 | goto fail; |
|
|
1612 | |
|
|
1613 | if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV) |
|
|
1614 | ERR ("malformed JSON string, tag value must be an array"); |
|
|
1615 | |
|
|
1616 | { |
|
|
1617 | AV *av = (AV *)SvRV (val); |
|
|
1618 | int i, len = av_len (av) + 1; |
|
|
1619 | HV *stash = gv_stashsv (tag, 0); |
|
|
1620 | SV *sv; |
|
|
1621 | |
|
|
1622 | if (!stash) |
|
|
1623 | ERR ("cannot decode perl-object (package does not exist)"); |
|
|
1624 | |
|
|
1625 | GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); |
|
|
1626 | |
|
|
1627 | if (!method) |
|
|
1628 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
|
|
1629 | |
|
|
1630 | dSP; |
|
|
1631 | |
|
|
1632 | ENTER; SAVETMPS; |
|
|
1633 | PUSHMARK (SP); |
|
|
1634 | EXTEND (SP, len + 2); |
|
|
1635 | // we re-bless the reference to get overload and other niceties right |
|
|
1636 | PUSHs (tag); |
|
|
1637 | PUSHs (sv_json); |
|
|
1638 | |
|
|
1639 | for (i = 0; i < len; ++i) |
|
|
1640 | PUSHs (*av_fetch (av, i, 1)); |
|
|
1641 | |
|
|
1642 | PUTBACK; |
|
|
1643 | call_sv ((SV *)GvCV (method), G_SCALAR); |
|
|
1644 | SPAGAIN; |
|
|
1645 | |
|
|
1646 | SvREFCNT_dec (tag); |
|
|
1647 | SvREFCNT_dec (val); |
|
|
1648 | sv = SvREFCNT_inc (POPs); |
|
|
1649 | |
|
|
1650 | PUTBACK; |
|
|
1651 | |
|
|
1652 | FREETMPS; LEAVE; |
|
|
1653 | |
|
|
1654 | return sv; |
|
|
1655 | } |
|
|
1656 | |
|
|
1657 | fail: |
|
|
1658 | SvREFCNT_dec (tag); |
|
|
1659 | SvREFCNT_dec (val); |
|
|
1660 | return 0; |
|
|
1661 | } |
|
|
1662 | |
|
|
1663 | static SV * |
1346 | decode_sv (dec_t *dec) |
1664 | decode_sv (dec_t *dec) |
1347 | { |
1665 | { |
1348 | // the beauty of JSON: you need exactly one character lookahead |
1666 | // the beauty of JSON: you need exactly one character lookahead |
1349 | // to parse everything. |
1667 | // to parse everything. |
1350 | switch (*dec->cur) |
1668 | switch (*dec->cur) |
1351 | { |
1669 | { |
1352 | case '"': ++dec->cur; return decode_str (dec); |
1670 | case '"': ++dec->cur; return decode_str (dec); |
1353 | case '[': ++dec->cur; return decode_av (dec); |
1671 | case '[': ++dec->cur; return decode_av (dec); |
1354 | case '{': ++dec->cur; return decode_hv (dec); |
1672 | case '{': ++dec->cur; return decode_hv (dec); |
|
|
1673 | case '(': return decode_tag (dec); |
1355 | |
1674 | |
1356 | case '-': |
1675 | case '-': |
1357 | case '0': case '1': case '2': case '3': case '4': |
1676 | case '0': case '1': case '2': case '3': case '4': |
1358 | case '5': case '6': case '7': case '8': case '9': |
1677 | case '5': case '6': case '7': case '8': case '9': |
1359 | return decode_num (dec); |
1678 | return decode_num (dec); |
1360 | |
1679 | |
|
|
1680 | case 'f': |
|
|
1681 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
|
|
1682 | { |
|
|
1683 | dec->cur += 5; |
|
|
1684 | |
|
|
1685 | if (expect_false (!dec->json.v_false)) |
|
|
1686 | dec->json.v_false = GET_BOOL (false); |
|
|
1687 | |
|
|
1688 | return newSVsv (dec->json.v_false); |
|
|
1689 | } |
|
|
1690 | else |
|
|
1691 | ERR ("'false' expected"); |
|
|
1692 | |
|
|
1693 | break; |
|
|
1694 | |
1361 | case 't': |
1695 | case 't': |
1362 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1696 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1363 | { |
1697 | { |
1364 | dec->cur += 4; |
1698 | dec->cur += 4; |
1365 | #if JSON_SLOW |
1699 | |
1366 | json_true = get_bool ("JSON::XS::true"); |
1700 | if (expect_false (!dec->json.v_true)) |
1367 | #endif |
1701 | dec->json.v_true = GET_BOOL (true); |
|
|
1702 | |
1368 | return newSVsv (json_true); |
1703 | return newSVsv (dec->json.v_true); |
1369 | } |
1704 | } |
1370 | else |
1705 | else |
1371 | ERR ("'true' expected"); |
1706 | ERR ("'true' expected"); |
1372 | |
|
|
1373 | break; |
|
|
1374 | |
|
|
1375 | case 'f': |
|
|
1376 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
|
|
1377 | { |
|
|
1378 | dec->cur += 5; |
|
|
1379 | #if JSON_SLOW |
|
|
1380 | json_false = get_bool ("JSON::XS::false"); |
|
|
1381 | #endif |
|
|
1382 | return newSVsv (json_false); |
|
|
1383 | } |
|
|
1384 | else |
|
|
1385 | ERR ("'false' expected"); |
|
|
1386 | |
1707 | |
1387 | break; |
1708 | break; |
1388 | |
1709 | |
1389 | case 'n': |
1710 | case 'n': |
1390 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
1711 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
… | |
… | |
1396 | ERR ("'null' expected"); |
1717 | ERR ("'null' expected"); |
1397 | |
1718 | |
1398 | break; |
1719 | break; |
1399 | |
1720 | |
1400 | default: |
1721 | default: |
1401 | ERR ("malformed JSON string, neither array, object, number, string or atom"); |
1722 | ERR ("malformed JSON string, neither tag, array, object, number, string or atom"); |
1402 | break; |
1723 | break; |
1403 | } |
1724 | } |
1404 | |
1725 | |
1405 | fail: |
1726 | fail: |
1406 | return 0; |
1727 | return 0; |
… | |
… | |
1408 | |
1729 | |
1409 | static SV * |
1730 | static SV * |
1410 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1731 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1411 | { |
1732 | { |
1412 | dec_t dec; |
1733 | dec_t dec; |
1413 | STRLEN offset; |
|
|
1414 | SV *sv; |
1734 | SV *sv; |
1415 | |
1735 | |
|
|
1736 | /* work around bugs in 5.10 where manipulating magic values |
|
|
1737 | * makes perl ignore the magic in subsequent accesses. |
|
|
1738 | * also make a copy of non-PV values, to get them into a clean |
|
|
1739 | * state (SvPV should do that, but it's buggy, see below). |
|
|
1740 | * |
|
|
1741 | * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16), |
|
|
1742 | * as reported by Reini Urban. |
|
|
1743 | */ |
1416 | SvGETMAGIC (string); |
1744 | /*SvGETMAGIC (string);*/ |
|
|
1745 | if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string)) |
|
|
1746 | string = sv_2mortal (newSVsv (string)); |
|
|
1747 | |
1417 | SvUPGRADE (string, SVt_PV); |
1748 | SvUPGRADE (string, SVt_PV); |
1418 | |
1749 | |
1419 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
1750 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
1420 | * assertion with -DDEBUGGING, although SvCUR is documented to |
1751 | * assertion with -DDEBUGGING, although SvCUR is documented to |
1421 | * return the xpv_cur field which certainly exists after upgrading. |
1752 | * return the xpv_cur field which certainly exists after upgrading. |
1422 | * according to nicholas clark, calling SvPOK fixes this. |
1753 | * according to nicholas clark, calling SvPOK fixes this. |
1423 | * But it doesn't fix it, so try another workaround, call SvPV_nolen |
1754 | * But it doesn't fix it, so try another workaround, call SvPV_nolen |
1424 | * and hope for the best. |
1755 | * and hope for the best. |
|
|
1756 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
|
|
1757 | * assertion business is seriously broken, try yet another workaround |
|
|
1758 | * for the broken -DDEBUGGING. |
1425 | */ |
1759 | */ |
|
|
1760 | { |
1426 | #ifdef DEBUGGING |
1761 | #ifdef DEBUGGING |
1427 | SvPV_nolen (string); |
1762 | STRLEN offset = SvOK (string) ? sv_len (string) : 0; |
|
|
1763 | #else |
|
|
1764 | STRLEN offset = SvCUR (string); |
1428 | #endif |
1765 | #endif |
1429 | |
1766 | |
1430 | if (SvCUR (string) > json->max_size && json->max_size) |
1767 | if (offset > json->max_size && json->max_size) |
1431 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1768 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1432 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
1769 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
|
|
1770 | } |
1433 | |
1771 | |
1434 | if (json->flags & F_UTF8) |
1772 | if (DECODE_WANTS_OCTETS (json)) |
1435 | sv_utf8_downgrade (string, 0); |
1773 | sv_utf8_downgrade (string, 0); |
1436 | else |
1774 | else |
1437 | sv_utf8_upgrade (string); |
1775 | sv_utf8_upgrade (string); |
1438 | |
1776 | |
1439 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1777 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
… | |
… | |
1450 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1788 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1451 | |
1789 | |
1452 | decode_ws (&dec); |
1790 | decode_ws (&dec); |
1453 | sv = decode_sv (&dec); |
1791 | sv = decode_sv (&dec); |
1454 | |
1792 | |
1455 | if (!(offset_return || !sv)) |
1793 | if (offset_return) |
|
|
1794 | *offset_return = dec.cur - SvPVX (string); |
|
|
1795 | else if (sv) |
1456 | { |
1796 | { |
1457 | // check for trailing garbage |
1797 | // check for trailing garbage |
1458 | decode_ws (&dec); |
1798 | decode_ws (&dec); |
1459 | |
1799 | |
1460 | if (*dec.cur) |
1800 | if (dec.cur != dec.end) |
1461 | { |
1801 | { |
1462 | dec.err = "garbage after JSON object"; |
1802 | dec.err = "garbage after JSON object"; |
1463 | SvREFCNT_dec (sv); |
1803 | SvREFCNT_dec (sv); |
1464 | sv = 0; |
1804 | sv = 0; |
1465 | } |
1805 | } |
1466 | } |
|
|
1467 | |
|
|
1468 | if (offset_return || !sv) |
|
|
1469 | { |
|
|
1470 | offset = dec.json.flags & F_UTF8 |
|
|
1471 | ? dec.cur - SvPVX (string) |
|
|
1472 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
1473 | |
|
|
1474 | if (offset_return) |
|
|
1475 | *offset_return = offset; |
|
|
1476 | } |
1806 | } |
1477 | |
1807 | |
1478 | if (!sv) |
1808 | if (!sv) |
1479 | { |
1809 | { |
1480 | SV *uni = sv_newmortal (); |
1810 | SV *uni = sv_newmortal (); |
… | |
… | |
1486 | SAVEVPTR (PL_curcop); |
1816 | SAVEVPTR (PL_curcop); |
1487 | PL_curcop = &cop; |
1817 | PL_curcop = &cop; |
1488 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1818 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1489 | LEAVE; |
1819 | LEAVE; |
1490 | |
1820 | |
1491 | croak ("%s, at character offset %d [\"%s\"]", |
1821 | croak ("%s, at character offset %d (before \"%s\")", |
1492 | dec.err, |
1822 | dec.err, |
1493 | (int)offset, |
1823 | (int)ptr_to_index (string, dec.cur), |
1494 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1824 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1495 | } |
1825 | } |
1496 | |
1826 | |
1497 | sv = sv_2mortal (sv); |
1827 | sv = sv_2mortal (sv); |
1498 | |
1828 | |
1499 | if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1829 | if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv)) |
1500 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1830 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1501 | |
1831 | |
1502 | return sv; |
1832 | return sv; |
1503 | } |
1833 | } |
1504 | |
1834 | |
… | |
… | |
1507 | |
1837 | |
1508 | static void |
1838 | static void |
1509 | incr_parse (JSON *self) |
1839 | incr_parse (JSON *self) |
1510 | { |
1840 | { |
1511 | const char *p = SvPVX (self->incr_text) + self->incr_pos; |
1841 | const char *p = SvPVX (self->incr_text) + self->incr_pos; |
|
|
1842 | |
|
|
1843 | // the state machine here is a bit convoluted and could be simplified a lot |
|
|
1844 | // but this would make it slower, so... |
1512 | |
1845 | |
1513 | for (;;) |
1846 | for (;;) |
1514 | { |
1847 | { |
1515 | //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 |
1848 | //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 |
1516 | switch (self->incr_mode) |
1849 | switch (self->incr_mode) |
1517 | { |
1850 | { |
1518 | // only used for intiial whitespace skipping |
1851 | // only used for initial whitespace skipping |
1519 | case INCR_M_WS: |
1852 | case INCR_M_WS: |
1520 | for (;;) |
1853 | for (;;) |
1521 | { |
1854 | { |
1522 | if (*p > 0x20) |
1855 | if (*p > 0x20) |
1523 | { |
1856 | { |
|
|
1857 | if (*p == '#') |
|
|
1858 | { |
|
|
1859 | self->incr_mode = INCR_M_C0; |
|
|
1860 | goto incr_m_c; |
|
|
1861 | } |
|
|
1862 | else |
|
|
1863 | { |
1524 | self->incr_mode = INCR_M_JSON; |
1864 | self->incr_mode = INCR_M_JSON; |
1525 | goto incr_m_json; |
1865 | goto incr_m_json; |
|
|
1866 | } |
1526 | } |
1867 | } |
1527 | else if (!*p) |
1868 | else if (!*p) |
1528 | goto interrupt; |
1869 | goto interrupt; |
1529 | |
1870 | |
1530 | ++p; |
1871 | ++p; |
… | |
… | |
1536 | goto interrupt; |
1877 | goto interrupt; |
1537 | |
1878 | |
1538 | ++p; |
1879 | ++p; |
1539 | self->incr_mode = INCR_M_STR; |
1880 | self->incr_mode = INCR_M_STR; |
1540 | goto incr_m_str; |
1881 | goto incr_m_str; |
|
|
1882 | |
|
|
1883 | // inside #-style comments |
|
|
1884 | case INCR_M_C0: |
|
|
1885 | case INCR_M_C1: |
|
|
1886 | incr_m_c: |
|
|
1887 | for (;;) |
|
|
1888 | { |
|
|
1889 | if (*p == '\n') |
|
|
1890 | { |
|
|
1891 | self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON; |
|
|
1892 | break; |
|
|
1893 | } |
|
|
1894 | else if (!*p) |
|
|
1895 | goto interrupt; |
|
|
1896 | |
|
|
1897 | ++p; |
|
|
1898 | } |
|
|
1899 | |
|
|
1900 | break; |
1541 | |
1901 | |
1542 | // inside a string |
1902 | // inside a string |
1543 | case INCR_M_STR: |
1903 | case INCR_M_STR: |
1544 | incr_m_str: |
1904 | incr_m_str: |
1545 | for (;;) |
1905 | for (;;) |
… | |
… | |
1596 | self->incr_mode = INCR_M_STR; |
1956 | self->incr_mode = INCR_M_STR; |
1597 | goto incr_m_str; |
1957 | goto incr_m_str; |
1598 | |
1958 | |
1599 | case '[': |
1959 | case '[': |
1600 | case '{': |
1960 | case '{': |
|
|
1961 | case '(': |
1601 | if (++self->incr_nest > self->max_depth) |
1962 | if (++self->incr_nest > self->max_depth) |
1602 | croak (ERR_NESTING_EXCEEDED); |
1963 | croak (ERR_NESTING_EXCEEDED); |
1603 | break; |
1964 | break; |
1604 | |
1965 | |
1605 | case ']': |
1966 | case ']': |
1606 | case '}': |
1967 | case '}': |
1607 | if (--self->incr_nest <= 0) |
1968 | if (--self->incr_nest <= 0) |
1608 | goto interrupt; |
1969 | goto interrupt; |
|
|
1970 | break; |
|
|
1971 | |
|
|
1972 | case ')': |
|
|
1973 | --self->incr_nest; |
|
|
1974 | break; |
|
|
1975 | |
|
|
1976 | case '#': |
|
|
1977 | self->incr_mode = INCR_M_C1; |
|
|
1978 | goto incr_m_c; |
1609 | } |
1979 | } |
1610 | } |
1980 | } |
1611 | } |
1981 | } |
1612 | |
1982 | |
1613 | modechange: |
1983 | modechange: |
1614 | ; |
1984 | ; |
1615 | } |
1985 | } |
1616 | |
1986 | |
1617 | interrupt: |
1987 | interrupt: |
1618 | self->incr_pos = p - SvPVX (self->incr_text); |
1988 | self->incr_pos = p - SvPVX (self->incr_text); |
|
|
1989 | //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D |
1619 | //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D |
1990 | //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D |
1620 | } |
1991 | } |
1621 | |
1992 | |
1622 | ///////////////////////////////////////////////////////////////////////////// |
1993 | ///////////////////////////////////////////////////////////////////////////// |
1623 | // XS interface functions |
1994 | // XS interface functions |
… | |
… | |
1633 | i >= '0' && i <= '9' ? i - '0' |
2004 | i >= '0' && i <= '9' ? i - '0' |
1634 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
2005 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1635 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
2006 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1636 | : -1; |
2007 | : -1; |
1637 | |
2008 | |
1638 | json_stash = gv_stashpv ("JSON::XS" , 1); |
2009 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1639 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
2010 | bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); |
1640 | |
|
|
1641 | json_true = get_bool ("JSON::XS::true"); |
|
|
1642 | json_false = get_bool ("JSON::XS::false"); |
2011 | bool_false = get_bool ("Types::Serialiser::false"); |
|
|
2012 | bool_true = get_bool ("Types::Serialiser::true"); |
|
|
2013 | |
|
|
2014 | sv_json = newSVpv ("JSON", 0); |
|
|
2015 | SvREADONLY_on (sv_json); |
|
|
2016 | |
|
|
2017 | CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ |
1643 | } |
2018 | } |
1644 | |
2019 | |
1645 | PROTOTYPES: DISABLE |
2020 | PROTOTYPES: DISABLE |
1646 | |
2021 | |
1647 | void CLONE (...) |
2022 | void CLONE (...) |
1648 | CODE: |
2023 | CODE: |
|
|
2024 | // as long as these writes are atomic, the race should not matter |
|
|
2025 | // as existing threads either already use 0, or use the old value, |
|
|
2026 | // which is sitll correct for the initial thread. |
1649 | json_stash = 0; |
2027 | json_stash = 0; |
1650 | json_boolean_stash = 0; |
2028 | bool_stash = 0; |
|
|
2029 | bool_false = 0; |
|
|
2030 | bool_true = 0; |
1651 | |
2031 | |
1652 | void new (char *klass) |
2032 | void new (char *klass) |
1653 | PPCODE: |
2033 | PPCODE: |
1654 | { |
2034 | { |
1655 | SV *pv = NEWSV (0, sizeof (JSON)); |
2035 | SV *pv = NEWSV (0, sizeof (JSON)); |
1656 | SvPOK_only (pv); |
2036 | SvPOK_only (pv); |
1657 | json_init ((JSON *)SvPVX (pv)); |
2037 | json_init ((JSON *)SvPVX (pv)); |
1658 | XPUSHs (sv_2mortal (sv_bless ( |
2038 | XPUSHs (sv_2mortal (sv_bless ( |
1659 | newRV_noinc (pv), |
2039 | newRV_noinc (pv), |
1660 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
2040 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
1661 | ))); |
2041 | ))); |
1662 | } |
2042 | } |
|
|
2043 | |
|
|
2044 | void boolean_values (JSON *self, SV *v_false = 0, SV *v_true = 0) |
|
|
2045 | PPCODE: |
|
|
2046 | self->v_false = newSVsv (v_false); |
|
|
2047 | self->v_true = newSVsv (v_true); |
|
|
2048 | XPUSHs (ST (0)); |
|
|
2049 | |
|
|
2050 | void get_boolean_values (JSON *self) |
|
|
2051 | PPCODE: |
|
|
2052 | if (self->v_false && self->v_true) |
|
|
2053 | { |
|
|
2054 | EXTEND (SP, 2); |
|
|
2055 | PUSHs (self->v_false); |
|
|
2056 | PUSHs (self->v_true); |
|
|
2057 | } |
1663 | |
2058 | |
1664 | void ascii (JSON *self, int enable = 1) |
2059 | void ascii (JSON *self, int enable = 1) |
1665 | ALIAS: |
2060 | ALIAS: |
1666 | ascii = F_ASCII |
2061 | ascii = F_ASCII |
1667 | latin1 = F_LATIN1 |
2062 | latin1 = F_LATIN1 |
… | |
… | |
1675 | shrink = F_SHRINK |
2070 | shrink = F_SHRINK |
1676 | allow_blessed = F_ALLOW_BLESSED |
2071 | allow_blessed = F_ALLOW_BLESSED |
1677 | convert_blessed = F_CONV_BLESSED |
2072 | convert_blessed = F_CONV_BLESSED |
1678 | relaxed = F_RELAXED |
2073 | relaxed = F_RELAXED |
1679 | allow_unknown = F_ALLOW_UNKNOWN |
2074 | allow_unknown = F_ALLOW_UNKNOWN |
|
|
2075 | allow_tags = F_ALLOW_TAGS |
1680 | PPCODE: |
2076 | PPCODE: |
1681 | { |
2077 | { |
1682 | if (enable) |
2078 | if (enable) |
1683 | self->flags |= ix; |
2079 | self->flags |= ix; |
1684 | else |
2080 | else |
… | |
… | |
1700 | get_shrink = F_SHRINK |
2096 | get_shrink = F_SHRINK |
1701 | get_allow_blessed = F_ALLOW_BLESSED |
2097 | get_allow_blessed = F_ALLOW_BLESSED |
1702 | get_convert_blessed = F_CONV_BLESSED |
2098 | get_convert_blessed = F_CONV_BLESSED |
1703 | get_relaxed = F_RELAXED |
2099 | get_relaxed = F_RELAXED |
1704 | get_allow_unknown = F_ALLOW_UNKNOWN |
2100 | get_allow_unknown = F_ALLOW_UNKNOWN |
|
|
2101 | get_allow_tags = F_ALLOW_TAGS |
1705 | PPCODE: |
2102 | PPCODE: |
1706 | XPUSHs (boolSV (self->flags & ix)); |
2103 | XPUSHs (boolSV (self->flags & ix)); |
1707 | |
2104 | |
1708 | void max_depth (JSON *self, U32 max_depth = 0x80000000UL) |
2105 | void max_depth (JSON *self, U32 max_depth = 0x80000000UL) |
1709 | PPCODE: |
2106 | PPCODE: |
… | |
… | |
1737 | } |
2134 | } |
1738 | |
2135 | |
1739 | void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) |
2136 | void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) |
1740 | PPCODE: |
2137 | PPCODE: |
1741 | { |
2138 | { |
1742 | if (!self->cb_sk_object) |
2139 | if (!self->cb_sk_object) |
1743 | self->cb_sk_object = newHV (); |
2140 | self->cb_sk_object = newHV (); |
1744 | |
2141 | |
1745 | if (SvOK (cb)) |
2142 | if (SvOK (cb)) |
1746 | hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
2143 | hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
1747 | else |
2144 | else |
… | |
… | |
1758 | XPUSHs (ST (0)); |
2155 | XPUSHs (ST (0)); |
1759 | } |
2156 | } |
1760 | |
2157 | |
1761 | void encode (JSON *self, SV *scalar) |
2158 | void encode (JSON *self, SV *scalar) |
1762 | PPCODE: |
2159 | PPCODE: |
1763 | XPUSHs (encode_json (scalar, self)); |
2160 | PUTBACK; scalar = encode_json (scalar, self); SPAGAIN; |
|
|
2161 | XPUSHs (scalar); |
1764 | |
2162 | |
1765 | void decode (JSON *self, SV *jsonstr) |
2163 | void decode (JSON *self, SV *jsonstr) |
1766 | PPCODE: |
2164 | PPCODE: |
1767 | XPUSHs (decode_json (jsonstr, self, 0)); |
2165 | PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN; |
|
|
2166 | XPUSHs (jsonstr); |
1768 | |
2167 | |
1769 | void decode_prefix (JSON *self, SV *jsonstr) |
2168 | void decode_prefix (JSON *self, SV *jsonstr) |
1770 | PPCODE: |
2169 | PPCODE: |
1771 | { |
2170 | { |
|
|
2171 | SV *sv; |
1772 | STRLEN offset; |
2172 | STRLEN offset; |
|
|
2173 | PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; |
1773 | EXTEND (SP, 2); |
2174 | EXTEND (SP, 2); |
1774 | PUSHs (decode_json (jsonstr, self, &offset)); |
2175 | PUSHs (sv); |
1775 | PUSHs (sv_2mortal (newSVuv (offset))); |
2176 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset)))); |
1776 | } |
2177 | } |
1777 | |
2178 | |
1778 | void incr_parse (JSON *self, SV *jsonstr = 0) |
2179 | void incr_parse (JSON *self, SV *jsonstr = 0) |
1779 | PPCODE: |
2180 | PPCODE: |
1780 | { |
2181 | { |
1781 | if (!self->incr_text) |
2182 | if (!self->incr_text) |
1782 | self->incr_text = newSVpvn ("", 0); |
2183 | self->incr_text = newSVpvn ("", 0); |
|
|
2184 | |
|
|
2185 | /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */ |
|
|
2186 | if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text)) |
|
|
2187 | if (DECODE_WANTS_OCTETS (self)) |
|
|
2188 | { |
|
|
2189 | if (self->incr_pos) |
|
|
2190 | self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text), |
|
|
2191 | (U8 *)SvPVX (self->incr_text) + self->incr_pos); |
|
|
2192 | |
|
|
2193 | sv_utf8_downgrade (self->incr_text, 0); |
|
|
2194 | } |
|
|
2195 | else |
|
|
2196 | { |
|
|
2197 | sv_utf8_upgrade (self->incr_text); |
|
|
2198 | |
|
|
2199 | if (self->incr_pos) |
|
|
2200 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
|
|
2201 | - (U8 *)SvPVX (self->incr_text); |
|
|
2202 | } |
1783 | |
2203 | |
1784 | // append data, if any |
2204 | // append data, if any |
1785 | if (jsonstr) |
2205 | if (jsonstr) |
1786 | { |
2206 | { |
|
|
2207 | /* make sure both strings have same encoding */ |
1787 | if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) |
2208 | if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text)) |
|
|
2209 | if (SvUTF8 (jsonstr)) |
|
|
2210 | sv_utf8_downgrade (jsonstr, 0); |
1788 | { |
2211 | else |
1789 | /* utf-8-ness differs, need to upgrade */ |
|
|
1790 | sv_utf8_upgrade (self->incr_text); |
2212 | sv_utf8_upgrade (jsonstr); |
1791 | |
2213 | |
1792 | if (self->incr_pos) |
2214 | /* and then just blindly append */ |
1793 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
|
|
1794 | - (U8 *)SvPVX (self->incr_text); |
|
|
1795 | } |
|
|
1796 | |
|
|
1797 | { |
2215 | { |
1798 | STRLEN len; |
2216 | STRLEN len; |
1799 | const char *str = SvPV (jsonstr, len); |
2217 | const char *str = SvPV (jsonstr, len); |
1800 | SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); |
2218 | STRLEN cur = SvCUR (self->incr_text); |
|
|
2219 | |
|
|
2220 | if (SvLEN (self->incr_text) - cur <= len) |
|
|
2221 | json_sv_grow (self->incr_text, cur, len); |
|
|
2222 | |
1801 | Move (str, SvEND (self->incr_text), len, char); |
2223 | Move (str, SvEND (self->incr_text), len, char); |
1802 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
2224 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
1803 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
2225 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
1804 | } |
2226 | } |
1805 | } |
2227 | } |
1806 | |
2228 | |
1807 | if (GIMME_V != G_VOID) |
2229 | if (GIMME_V != G_VOID) |
1808 | do |
2230 | do |
1809 | { |
2231 | { |
|
|
2232 | SV *sv; |
1810 | STRLEN offset; |
2233 | STRLEN offset; |
1811 | |
2234 | |
1812 | if (!INCR_DONE (self)) |
2235 | if (!INCR_DONE (self)) |
1813 | { |
2236 | { |
1814 | incr_parse (self); |
2237 | incr_parse (self); |
… | |
… | |
1816 | if (self->incr_pos > self->max_size && self->max_size) |
2239 | if (self->incr_pos > self->max_size && self->max_size) |
1817 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
2240 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1818 | (unsigned long)self->incr_pos, (unsigned long)self->max_size); |
2241 | (unsigned long)self->incr_pos, (unsigned long)self->max_size); |
1819 | |
2242 | |
1820 | if (!INCR_DONE (self)) |
2243 | if (!INCR_DONE (self)) |
|
|
2244 | { |
|
|
2245 | // as an optimisation, do not accumulate white space in the incr buffer |
|
|
2246 | if (self->incr_mode == INCR_M_WS && self->incr_pos) |
|
|
2247 | { |
|
|
2248 | self->incr_pos = 0; |
|
|
2249 | SvCUR_set (self->incr_text, 0); |
|
|
2250 | } |
|
|
2251 | |
1821 | break; |
2252 | break; |
|
|
2253 | } |
1822 | } |
2254 | } |
1823 | |
2255 | |
1824 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
2256 | PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; |
|
|
2257 | XPUSHs (sv); |
1825 | |
2258 | |
1826 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset); |
|
|
1827 | self->incr_pos -= offset; |
2259 | self->incr_pos -= offset; |
1828 | self->incr_nest = 0; |
2260 | self->incr_nest = 0; |
1829 | self->incr_mode = 0; |
2261 | self->incr_mode = 0; |
|
|
2262 | |
|
|
2263 | sv_chop (self->incr_text, SvPVX (self->incr_text) + offset); |
1830 | } |
2264 | } |
1831 | while (GIMME_V == G_ARRAY); |
2265 | while (GIMME_V == G_ARRAY); |
1832 | } |
2266 | } |
1833 | |
2267 | |
1834 | SV *incr_text (JSON *self) |
2268 | SV *incr_text (JSON *self) |
… | |
… | |
1865 | self->incr_mode = 0; |
2299 | self->incr_mode = 0; |
1866 | } |
2300 | } |
1867 | |
2301 | |
1868 | void DESTROY (JSON *self) |
2302 | void DESTROY (JSON *self) |
1869 | CODE: |
2303 | CODE: |
|
|
2304 | SvREFCNT_dec (self->v_false); |
|
|
2305 | SvREFCNT_dec (self->v_true); |
1870 | SvREFCNT_dec (self->cb_sk_object); |
2306 | SvREFCNT_dec (self->cb_sk_object); |
1871 | SvREFCNT_dec (self->cb_object); |
2307 | SvREFCNT_dec (self->cb_object); |
1872 | SvREFCNT_dec (self->incr_text); |
2308 | SvREFCNT_dec (self->incr_text); |
1873 | |
2309 | |
1874 | PROTOTYPES: ENABLE |
2310 | PROTOTYPES: ENABLE |
1875 | |
2311 | |
1876 | void encode_json (SV *scalar) |
2312 | void encode_json (SV *scalar) |
1877 | ALIAS: |
|
|
1878 | to_json_ = 0 |
|
|
1879 | encode_json = F_UTF8 |
|
|
1880 | PPCODE: |
2313 | PPCODE: |
1881 | { |
2314 | { |
1882 | JSON json; |
2315 | JSON json; |
1883 | json_init (&json); |
2316 | json_init (&json); |
1884 | json.flags |= ix; |
2317 | json.flags |= F_UTF8; |
1885 | XPUSHs (encode_json (scalar, &json)); |
2318 | PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN; |
|
|
2319 | XPUSHs (scalar); |
1886 | } |
2320 | } |
1887 | |
2321 | |
1888 | void decode_json (SV *jsonstr) |
2322 | void decode_json (SV *jsonstr) |
1889 | ALIAS: |
|
|
1890 | from_json_ = 0 |
|
|
1891 | decode_json = F_UTF8 |
|
|
1892 | PPCODE: |
2323 | PPCODE: |
1893 | { |
2324 | { |
1894 | JSON json; |
2325 | JSON json; |
1895 | json_init (&json); |
2326 | json_init (&json); |
1896 | json.flags |= ix; |
2327 | json.flags |= F_UTF8; |
1897 | XPUSHs (decode_json (jsonstr, &json, 0)); |
2328 | PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN; |
|
|
2329 | XPUSHs (jsonstr); |
1898 | } |
2330 | } |
1899 | |
2331 | |
1900 | |
|
|