1 | #include "EXTERN.h" |
1 | #include "EXTERN.h" |
2 | #include "perl.h" |
2 | #include "perl.h" |
3 | #include "XSUB.h" |
3 | #include "XSUB.h" |
4 | |
4 | |
5 | #include "assert.h" |
5 | #include <assert.h> |
6 | #include "string.h" |
6 | #include <string.h> |
7 | #include "stdlib.h" |
7 | #include <stdlib.h> |
8 | #include "stdio.h" |
8 | #include <stdio.h> |
|
|
9 | #include <float.h> |
9 | |
10 | |
10 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
11 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
11 | # define snprintf _snprintf // C compilers have this in stdio.h |
12 | # define snprintf _snprintf // C compilers have this in stdio.h |
12 | #endif |
13 | #endif |
13 | |
14 | |
|
|
15 | // 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. |
|
|
17 | #ifndef UTF8_MAXBYTES |
|
|
18 | # define UTF8_MAXBYTES 13 |
|
|
19 | #endif |
|
|
20 | |
14 | #define F_ASCII 0x00000001UL |
21 | #define F_ASCII 0x00000001UL |
15 | #define F_LATIN1 0x00000002UL |
22 | #define F_LATIN1 0x00000002UL |
16 | #define F_UTF8 0x00000004UL |
23 | #define F_UTF8 0x00000004UL |
17 | #define F_INDENT 0x00000008UL |
24 | #define F_INDENT 0x00000008UL |
18 | #define F_CANONICAL 0x00000010UL |
25 | #define F_CANONICAL 0x00000010UL |
19 | #define F_SPACE_BEFORE 0x00000020UL |
26 | #define F_SPACE_BEFORE 0x00000020UL |
20 | #define F_SPACE_AFTER 0x00000040UL |
27 | #define F_SPACE_AFTER 0x00000040UL |
21 | #define F_ALLOW_NONREF 0x00000100UL |
28 | #define F_ALLOW_NONREF 0x00000100UL |
22 | #define F_SHRINK 0x00000200UL |
29 | #define F_SHRINK 0x00000200UL |
|
|
30 | #define F_ALLOW_BLESSED 0x00000400UL |
|
|
31 | #define F_CONV_BLESSED 0x00000800UL |
|
|
32 | #define F_RELAXED 0x00001000UL |
|
|
33 | |
23 | #define F_MAXDEPTH 0xf8000000UL |
34 | #define F_MAXDEPTH 0xf8000000UL |
24 | #define S_MAXDEPTH 27 |
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 |
25 | |
39 | |
26 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
40 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
27 | |
41 | #define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE )) |
28 | // F_SELFCONVERT? <=> to_json/toJson |
|
|
29 | // F_BLESSED? <=> { $__class__$ => } |
|
|
30 | |
42 | |
31 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
43 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
32 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
44 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
33 | |
45 | |
34 | #define INIT_SIZE 32 // initial scalar size to be allocated |
46 | #define INIT_SIZE 32 // initial scalar size to be allocated |
35 | #define INDENT_STEP 3 // spaces per indentation level |
47 | #define INDENT_STEP 3 // spaces per indentation level |
36 | |
48 | |
37 | #define SHORT_STRING_LEN 512 // special-case strings of up to this size |
49 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
38 | |
50 | |
39 | #define SB do { |
51 | #define SB do { |
40 | #define SE } while (0) |
52 | #define SE } while (0) |
41 | |
53 | |
|
|
54 | #if __GNUC__ >= 3 |
|
|
55 | # define expect(expr,value) __builtin_expect ((expr),(value)) |
|
|
56 | # define inline inline |
|
|
57 | #else |
|
|
58 | # define expect(expr,value) (expr) |
|
|
59 | # define inline static |
|
|
60 | #endif |
|
|
61 | |
|
|
62 | #define expect_false(expr) expect ((expr) != 0, 0) |
|
|
63 | #define expect_true(expr) expect ((expr) != 0, 1) |
|
|
64 | |
|
|
65 | #ifdef USE_ITHREADS |
|
|
66 | # define JSON_SLOW 1 |
|
|
67 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
|
|
68 | #else |
|
|
69 | # define JSON_SLOW 0 |
|
|
70 | # define JSON_STASH json_stash |
|
|
71 | #endif |
|
|
72 | |
42 | static HV *json_stash; // JSON::XS:: |
73 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
|
|
74 | static SV *json_true, *json_false; |
|
|
75 | |
|
|
76 | typedef struct { |
|
|
77 | U32 flags; |
|
|
78 | SV *cb_object; |
|
|
79 | HV *cb_sk_object; |
|
|
80 | } JSON; |
43 | |
81 | |
44 | ///////////////////////////////////////////////////////////////////////////// |
82 | ///////////////////////////////////////////////////////////////////////////// |
45 | // utility functions |
83 | // utility functions |
46 | |
84 | |
47 | static UV * |
85 | inline void |
48 | SvJSON (SV *sv) |
|
|
49 | { |
|
|
50 | if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) |
|
|
51 | croak ("object is not of type JSON::XS"); |
|
|
52 | |
|
|
53 | return &SvUVX (SvRV (sv)); |
|
|
54 | } |
|
|
55 | |
|
|
56 | static void |
|
|
57 | shrink (SV *sv) |
86 | shrink (SV *sv) |
58 | { |
87 | { |
59 | sv_utf8_downgrade (sv, 1); |
88 | sv_utf8_downgrade (sv, 1); |
60 | if (SvLEN (sv) > SvCUR (sv) + 1) |
89 | if (SvLEN (sv) > SvCUR (sv) + 1) |
61 | { |
90 | { |
… | |
… | |
70 | // decode an utf-8 character and return it, or (UV)-1 in |
99 | // decode an utf-8 character and return it, or (UV)-1 in |
71 | // case of an error. |
100 | // case of an error. |
72 | // we special-case "safe" characters from U+80 .. U+7FF, |
101 | // we special-case "safe" characters from U+80 .. U+7FF, |
73 | // but use the very good perl function to parse anything else. |
102 | // but use the very good perl function to parse anything else. |
74 | // note that we never call this function for a ascii codepoints |
103 | // note that we never call this function for a ascii codepoints |
75 | static UV |
104 | inline UV |
76 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
105 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
77 | { |
106 | { |
78 | if (s[0] > 0xdf || s[0] < 0xc2) |
107 | if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) |
79 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
108 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
80 | else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
109 | else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
81 | { |
110 | { |
82 | *clen = 2; |
111 | *clen = 2; |
83 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
112 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
… | |
… | |
96 | typedef struct |
125 | typedef struct |
97 | { |
126 | { |
98 | char *cur; // SvPVX (sv) + current output position |
127 | char *cur; // SvPVX (sv) + current output position |
99 | char *end; // SvEND (sv) |
128 | char *end; // SvEND (sv) |
100 | SV *sv; // result scalar |
129 | SV *sv; // result scalar |
101 | U32 flags; // F_* |
130 | JSON json; |
102 | U32 indent; // indentation level |
131 | U32 indent; // indentation level |
103 | U32 maxdepth; // max. indentation/recursion level |
132 | U32 maxdepth; // max. indentation/recursion level |
104 | } enc_t; |
133 | } enc_t; |
105 | |
134 | |
106 | static void |
135 | inline void |
107 | need (enc_t *enc, STRLEN len) |
136 | need (enc_t *enc, STRLEN len) |
108 | { |
137 | { |
109 | if (enc->cur + len >= enc->end) |
138 | if (expect_false (enc->cur + len >= enc->end)) |
110 | { |
139 | { |
111 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
140 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
112 | SvGROW (enc->sv, cur + len + 1); |
141 | SvGROW (enc->sv, cur + len + 1); |
113 | enc->cur = SvPVX (enc->sv) + cur; |
142 | enc->cur = SvPVX (enc->sv) + cur; |
114 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
143 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
115 | } |
144 | } |
116 | } |
145 | } |
117 | |
146 | |
118 | static void |
147 | inline void |
119 | encode_ch (enc_t *enc, char ch) |
148 | encode_ch (enc_t *enc, char ch) |
120 | { |
149 | { |
121 | need (enc, 1); |
150 | need (enc, 1); |
122 | *enc->cur++ = ch; |
151 | *enc->cur++ = ch; |
123 | } |
152 | } |
… | |
… | |
131 | |
160 | |
132 | while (str < end) |
161 | while (str < end) |
133 | { |
162 | { |
134 | unsigned char ch = *(unsigned char *)str; |
163 | unsigned char ch = *(unsigned char *)str; |
135 | |
164 | |
136 | if (ch >= 0x20 && ch < 0x80) // most common case |
165 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
137 | { |
166 | { |
138 | if (ch == '"') // but with slow exceptions |
167 | if (expect_false (ch == '"')) // but with slow exceptions |
139 | { |
168 | { |
140 | need (enc, len += 1); |
169 | need (enc, len += 1); |
141 | *enc->cur++ = '\\'; |
170 | *enc->cur++ = '\\'; |
142 | *enc->cur++ = '"'; |
171 | *enc->cur++ = '"'; |
143 | } |
172 | } |
144 | else if (ch == '\\') |
173 | else if (expect_false (ch == '\\')) |
145 | { |
174 | { |
146 | need (enc, len += 1); |
175 | need (enc, len += 1); |
147 | *enc->cur++ = '\\'; |
176 | *enc->cur++ = '\\'; |
148 | *enc->cur++ = '\\'; |
177 | *enc->cur++ = '\\'; |
149 | } |
178 | } |
… | |
… | |
167 | STRLEN clen; |
196 | STRLEN clen; |
168 | UV uch; |
197 | UV uch; |
169 | |
198 | |
170 | if (is_utf8) |
199 | if (is_utf8) |
171 | { |
200 | { |
172 | //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); |
|
|
173 | uch = decode_utf8 (str, end - str, &clen); |
201 | uch = decode_utf8 (str, end - str, &clen); |
174 | if (clen == (STRLEN)-1) |
202 | if (clen == (STRLEN)-1) |
175 | croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
203 | croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
176 | } |
204 | } |
177 | else |
205 | else |
… | |
… | |
181 | } |
209 | } |
182 | |
210 | |
183 | if (uch > 0x10FFFFUL) |
211 | if (uch > 0x10FFFFUL) |
184 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
212 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
185 | |
213 | |
186 | if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF)) |
214 | if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF)) |
187 | { |
215 | { |
188 | if (uch > 0xFFFFUL) |
216 | if (uch > 0xFFFFUL) |
189 | { |
217 | { |
190 | need (enc, len += 11); |
218 | need (enc, len += 11); |
191 | sprintf (enc->cur, "\\u%04x\\u%04x", |
219 | sprintf (enc->cur, "\\u%04x\\u%04x", |
… | |
… | |
205 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
233 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
206 | } |
234 | } |
207 | |
235 | |
208 | str += clen; |
236 | str += clen; |
209 | } |
237 | } |
210 | else if (enc->flags & F_LATIN1) |
238 | else if (enc->json.flags & F_LATIN1) |
211 | { |
239 | { |
212 | *enc->cur++ = uch; |
240 | *enc->cur++ = uch; |
213 | str += clen; |
241 | str += clen; |
214 | } |
242 | } |
215 | else if (is_utf8) |
243 | else if (is_utf8) |
… | |
… | |
233 | |
261 | |
234 | --len; |
262 | --len; |
235 | } |
263 | } |
236 | } |
264 | } |
237 | |
265 | |
238 | static void |
266 | inline void |
239 | encode_indent (enc_t *enc) |
267 | encode_indent (enc_t *enc) |
240 | { |
268 | { |
241 | if (enc->flags & F_INDENT) |
269 | if (enc->json.flags & F_INDENT) |
242 | { |
270 | { |
243 | int spaces = enc->indent * INDENT_STEP; |
271 | int spaces = enc->indent * INDENT_STEP; |
244 | |
272 | |
245 | need (enc, spaces); |
273 | need (enc, spaces); |
246 | memset (enc->cur, ' ', spaces); |
274 | memset (enc->cur, ' ', spaces); |
247 | enc->cur += spaces; |
275 | enc->cur += spaces; |
248 | } |
276 | } |
249 | } |
277 | } |
250 | |
278 | |
251 | static void |
279 | inline void |
252 | encode_space (enc_t *enc) |
280 | encode_space (enc_t *enc) |
253 | { |
281 | { |
254 | need (enc, 1); |
282 | need (enc, 1); |
255 | encode_ch (enc, ' '); |
283 | encode_ch (enc, ' '); |
256 | } |
284 | } |
257 | |
285 | |
258 | static void |
286 | inline void |
259 | encode_nl (enc_t *enc) |
287 | encode_nl (enc_t *enc) |
260 | { |
288 | { |
261 | if (enc->flags & F_INDENT) |
289 | if (enc->json.flags & F_INDENT) |
262 | { |
290 | { |
263 | need (enc, 1); |
291 | need (enc, 1); |
264 | encode_ch (enc, '\n'); |
292 | encode_ch (enc, '\n'); |
265 | } |
293 | } |
266 | } |
294 | } |
267 | |
295 | |
268 | static void |
296 | inline void |
269 | encode_comma (enc_t *enc) |
297 | encode_comma (enc_t *enc) |
270 | { |
298 | { |
271 | encode_ch (enc, ','); |
299 | encode_ch (enc, ','); |
272 | |
300 | |
273 | if (enc->flags & F_INDENT) |
301 | if (enc->json.flags & F_INDENT) |
274 | encode_nl (enc); |
302 | encode_nl (enc); |
275 | else if (enc->flags & F_SPACE_AFTER) |
303 | else if (enc->json.flags & F_SPACE_AFTER) |
276 | encode_space (enc); |
304 | encode_space (enc); |
277 | } |
305 | } |
278 | |
306 | |
279 | static void encode_sv (enc_t *enc, SV *sv); |
307 | static void encode_sv (enc_t *enc, SV *sv); |
280 | |
308 | |
… | |
… | |
289 | encode_ch (enc, '['); encode_nl (enc); |
317 | encode_ch (enc, '['); encode_nl (enc); |
290 | ++enc->indent; |
318 | ++enc->indent; |
291 | |
319 | |
292 | for (i = 0; i <= len; ++i) |
320 | for (i = 0; i <= len; ++i) |
293 | { |
321 | { |
|
|
322 | SV **svp = av_fetch (av, i, 0); |
|
|
323 | |
294 | encode_indent (enc); |
324 | encode_indent (enc); |
295 | encode_sv (enc, *av_fetch (av, i, 0)); |
325 | |
|
|
326 | if (svp) |
|
|
327 | encode_sv (enc, *svp); |
|
|
328 | else |
|
|
329 | encode_str (enc, "null", 4, 0); |
296 | |
330 | |
297 | if (i < len) |
331 | if (i < len) |
298 | encode_comma (enc); |
332 | encode_comma (enc); |
299 | } |
333 | } |
300 | |
334 | |
… | |
… | |
303 | --enc->indent; |
337 | --enc->indent; |
304 | encode_indent (enc); encode_ch (enc, ']'); |
338 | encode_indent (enc); encode_ch (enc, ']'); |
305 | } |
339 | } |
306 | |
340 | |
307 | static void |
341 | static void |
308 | encode_he (enc_t *enc, HE *he) |
342 | encode_hk (enc_t *enc, HE *he) |
309 | { |
343 | { |
310 | encode_ch (enc, '"'); |
344 | encode_ch (enc, '"'); |
311 | |
345 | |
312 | if (HeKLEN (he) == HEf_SVKEY) |
346 | if (HeKLEN (he) == HEf_SVKEY) |
313 | { |
347 | { |
… | |
… | |
323 | else |
357 | else |
324 | encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
358 | encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
325 | |
359 | |
326 | encode_ch (enc, '"'); |
360 | encode_ch (enc, '"'); |
327 | |
361 | |
328 | if (enc->flags & F_SPACE_BEFORE) encode_space (enc); |
362 | if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); |
329 | encode_ch (enc, ':'); |
363 | encode_ch (enc, ':'); |
330 | if (enc->flags & F_SPACE_AFTER ) encode_space (enc); |
364 | if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); |
331 | encode_sv (enc, HeVAL (he)); |
|
|
332 | } |
365 | } |
333 | |
366 | |
334 | // compare hash entries, used when all keys are bytestrings |
367 | // compare hash entries, used when all keys are bytestrings |
335 | static int |
368 | static int |
336 | he_cmp_fast (const void *a_, const void *b_) |
369 | he_cmp_fast (const void *a_, const void *b_) |
… | |
… | |
341 | HE *b = *(HE **)b_; |
374 | HE *b = *(HE **)b_; |
342 | |
375 | |
343 | STRLEN la = HeKLEN (a); |
376 | STRLEN la = HeKLEN (a); |
344 | STRLEN lb = HeKLEN (b); |
377 | STRLEN lb = HeKLEN (b); |
345 | |
378 | |
346 | if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) |
379 | if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la))) |
347 | cmp = la - lb; |
380 | cmp = lb - la; |
348 | |
381 | |
349 | return cmp; |
382 | return cmp; |
350 | } |
383 | } |
351 | |
384 | |
352 | // compare hash entries, used when some keys are sv's or utf-x |
385 | // compare hash entries, used when some keys are sv's or utf-x |
353 | static int |
386 | static int |
354 | he_cmp_slow (const void *a, const void *b) |
387 | he_cmp_slow (const void *a, const void *b) |
355 | { |
388 | { |
356 | return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); |
389 | return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a)); |
357 | } |
390 | } |
358 | |
391 | |
359 | static void |
392 | static void |
360 | encode_hv (enc_t *enc, HV *hv) |
393 | encode_hv (enc_t *enc, HV *hv) |
361 | { |
394 | { |
|
|
395 | HE *he; |
362 | int count, i; |
396 | int count; |
363 | |
397 | |
364 | if (enc->indent >= enc->maxdepth) |
398 | if (enc->indent >= enc->maxdepth) |
365 | croak ("data structure too deep (hit recursion limit)"); |
399 | croak ("data structure too deep (hit recursion limit)"); |
366 | |
400 | |
367 | encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; |
401 | encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; |
368 | |
402 | |
369 | if ((count = hv_iterinit (hv))) |
|
|
370 | { |
|
|
371 | // for canonical output we have to sort by keys first |
403 | // for canonical output we have to sort by keys first |
372 | // actually, this is mostly due to the stupid so-called |
404 | // actually, this is mostly due to the stupid so-called |
373 | // security workaround added somewhere in 5.8.x. |
405 | // security workaround added somewhere in 5.8.x. |
374 | // that randomises hash orderings |
406 | // that randomises hash orderings |
375 | if (enc->flags & F_CANONICAL) |
407 | if (enc->json.flags & F_CANONICAL) |
|
|
408 | { |
|
|
409 | int count = hv_iterinit (hv); |
|
|
410 | |
|
|
411 | if (SvMAGICAL (hv)) |
376 | { |
412 | { |
|
|
413 | // need to count by iterating. could improve by dynamically building the vector below |
|
|
414 | // but I don't care for the speed of this special case. |
|
|
415 | // note also that we will run into undefined behaviour when the two iterations |
|
|
416 | // do not result in the same count, something I might care for in some later release. |
|
|
417 | |
|
|
418 | count = 0; |
|
|
419 | while (hv_iternext (hv)) |
|
|
420 | ++count; |
|
|
421 | |
|
|
422 | hv_iterinit (hv); |
|
|
423 | } |
|
|
424 | |
|
|
425 | if (count) |
|
|
426 | { |
|
|
427 | int i, fast = 1; |
|
|
428 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
|
|
429 | HE **hes = _alloca (count * sizeof (HE)); |
|
|
430 | #else |
377 | HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode |
431 | HE *hes [count]; // if your compiler dies here, you need to enable C99 mode |
378 | int fast = 1; |
432 | #endif |
379 | |
433 | |
380 | i = 0; |
434 | i = 0; |
381 | while ((he = hv_iternext (hv))) |
435 | while ((he = hv_iternext (hv))) |
382 | { |
436 | { |
383 | hes [i++] = he; |
437 | hes [i++] = he; |
… | |
… | |
405 | |
459 | |
406 | FREETMPS; |
460 | FREETMPS; |
407 | LEAVE; |
461 | LEAVE; |
408 | } |
462 | } |
409 | |
463 | |
410 | for (i = 0; i < count; ++i) |
464 | while (count--) |
411 | { |
465 | { |
412 | encode_indent (enc); |
466 | encode_indent (enc); |
|
|
467 | he = hes [count]; |
413 | encode_he (enc, hes [i]); |
468 | encode_hk (enc, he); |
|
|
469 | encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
414 | |
470 | |
415 | if (i < count - 1) |
471 | if (count) |
416 | encode_comma (enc); |
472 | encode_comma (enc); |
417 | } |
473 | } |
418 | |
474 | } |
|
|
475 | } |
|
|
476 | else |
|
|
477 | { |
|
|
478 | if (hv_iterinit (hv) || SvMAGICAL (hv)) |
|
|
479 | if ((he = hv_iternext (hv))) |
|
|
480 | for (;;) |
|
|
481 | { |
419 | encode_nl (enc); |
482 | encode_indent (enc); |
|
|
483 | encode_hk (enc, he); |
|
|
484 | encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
|
|
485 | |
|
|
486 | if (!(he = hv_iternext (hv))) |
|
|
487 | break; |
|
|
488 | |
|
|
489 | encode_comma (enc); |
|
|
490 | } |
|
|
491 | } |
|
|
492 | |
|
|
493 | encode_nl (enc); |
|
|
494 | |
|
|
495 | --enc->indent; encode_indent (enc); encode_ch (enc, '}'); |
|
|
496 | } |
|
|
497 | |
|
|
498 | // encode objects, arrays and special \0=false and \1=true values. |
|
|
499 | static void |
|
|
500 | encode_rv (enc_t *enc, SV *sv) |
|
|
501 | { |
|
|
502 | svtype svt; |
|
|
503 | |
|
|
504 | SvGETMAGIC (sv); |
|
|
505 | svt = SvTYPE (sv); |
|
|
506 | |
|
|
507 | if (expect_false (SvOBJECT (sv))) |
|
|
508 | { |
|
|
509 | HV *stash = !JSON_SLOW || json_boolean_stash |
|
|
510 | ? json_boolean_stash |
|
|
511 | : gv_stashpv ("JSON::XS::Boolean", 1); |
|
|
512 | |
|
|
513 | if (SvSTASH (sv) == stash) |
|
|
514 | { |
|
|
515 | if (SvIV (sv)) |
|
|
516 | encode_str (enc, "true", 4, 0); |
|
|
517 | else |
|
|
518 | encode_str (enc, "false", 5, 0); |
420 | } |
519 | } |
421 | else |
520 | else |
422 | { |
521 | { |
423 | HE *he = hv_iternext (hv); |
522 | #if 0 |
424 | |
523 | if (0 && sv_derived_from (rv, "JSON::Literal")) |
425 | for (;;) |
|
|
426 | { |
|
|
427 | encode_indent (enc); |
|
|
428 | encode_he (enc, he); |
|
|
429 | |
|
|
430 | if (!(he = hv_iternext (hv))) |
|
|
431 | break; |
|
|
432 | |
|
|
433 | encode_comma (enc); |
|
|
434 | } |
524 | { |
|
|
525 | // not yet |
|
|
526 | } |
|
|
527 | #endif |
|
|
528 | if (enc->json.flags & F_CONV_BLESSED) |
|
|
529 | { |
|
|
530 | // we re-bless the reference to get overload and other niceties right |
|
|
531 | GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0); |
435 | |
532 | |
|
|
533 | if (to_json) |
|
|
534 | { |
|
|
535 | dSP; |
|
|
536 | |
|
|
537 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
538 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); |
|
|
539 | |
|
|
540 | // calling with G_SCALAR ensures that we always get a 1 return value |
|
|
541 | PUTBACK; |
|
|
542 | call_sv ((SV *)GvCV (to_json), G_SCALAR); |
|
|
543 | SPAGAIN; |
|
|
544 | |
|
|
545 | // catch this surprisingly common error |
|
|
546 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
547 | croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
548 | |
|
|
549 | sv = POPs; |
|
|
550 | PUTBACK; |
|
|
551 | |
436 | encode_nl (enc); |
552 | encode_sv (enc, sv); |
|
|
553 | |
|
|
554 | FREETMPS; LEAVE; |
|
|
555 | } |
|
|
556 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
557 | encode_str (enc, "null", 4, 0); |
|
|
558 | else |
|
|
559 | croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it", |
|
|
560 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
561 | } |
|
|
562 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
563 | encode_str (enc, "null", 4, 0); |
|
|
564 | else |
|
|
565 | croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", |
|
|
566 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
437 | } |
567 | } |
438 | } |
568 | } |
439 | |
|
|
440 | --enc->indent; encode_indent (enc); encode_ch (enc, '}'); |
|
|
441 | } |
|
|
442 | |
|
|
443 | // encode objects, arrays and special \0=false and \1=true values. |
|
|
444 | static void |
|
|
445 | encode_rv (enc_t *enc, SV *sv) |
|
|
446 | { |
|
|
447 | svtype svt; |
|
|
448 | |
|
|
449 | SvGETMAGIC (sv); |
|
|
450 | svt = SvTYPE (sv); |
|
|
451 | |
|
|
452 | if (svt == SVt_PVHV) |
569 | else if (svt == SVt_PVHV) |
453 | encode_hv (enc, (HV *)sv); |
570 | encode_hv (enc, (HV *)sv); |
454 | else if (svt == SVt_PVAV) |
571 | else if (svt == SVt_PVAV) |
455 | encode_av (enc, (AV *)sv); |
572 | encode_av (enc, (AV *)sv); |
456 | else if (svt < SVt_PVAV) |
573 | else if (svt < SVt_PVAV) |
457 | { |
574 | { |
458 | if (SvNIOK (sv) && SvIV (sv) == 0) |
575 | STRLEN len = 0; |
|
|
576 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
577 | |
|
|
578 | if (len == 1 && *pv == '1') |
|
|
579 | encode_str (enc, "true", 4, 0); |
|
|
580 | else if (len == 1 && *pv == '0') |
459 | encode_str (enc, "false", 5, 0); |
581 | encode_str (enc, "false", 5, 0); |
460 | else if (SvNIOK (sv) && SvIV (sv) == 1) |
|
|
461 | encode_str (enc, "true", 4, 0); |
|
|
462 | else |
582 | else |
463 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
583 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
464 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
584 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
465 | } |
585 | } |
466 | else |
586 | else |
… | |
… | |
481 | encode_str (enc, str, len, SvUTF8 (sv)); |
601 | encode_str (enc, str, len, SvUTF8 (sv)); |
482 | encode_ch (enc, '"'); |
602 | encode_ch (enc, '"'); |
483 | } |
603 | } |
484 | else if (SvNOKp (sv)) |
604 | else if (SvNOKp (sv)) |
485 | { |
605 | { |
|
|
606 | // trust that perl will do the right thing w.r.t. JSON syntax. |
486 | need (enc, NV_DIG + 32); |
607 | need (enc, NV_DIG + 32); |
487 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
608 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
488 | enc->cur += strlen (enc->cur); |
609 | enc->cur += strlen (enc->cur); |
489 | } |
610 | } |
490 | else if (SvIOKp (sv)) |
611 | else if (SvIOKp (sv)) |
491 | { |
612 | { |
492 | need (enc, 64); |
613 | // we assume we can always read an IV as a UV |
|
|
614 | if (SvUV (sv) & ~(UV)0x7fff) |
|
|
615 | { |
|
|
616 | // large integer, use the (rather slow) snprintf way. |
|
|
617 | need (enc, sizeof (UV) * 3); |
493 | enc->cur += |
618 | enc->cur += |
494 | SvIsUV(sv) |
619 | SvIsUV(sv) |
495 | ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) |
620 | ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv)) |
496 | : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); |
621 | : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv)); |
|
|
622 | } |
|
|
623 | else |
|
|
624 | { |
|
|
625 | // optimise the "small number case" |
|
|
626 | // code will likely be branchless and use only a single multiplication |
|
|
627 | I32 i = SvIV (sv); |
|
|
628 | U32 u; |
|
|
629 | char digit, nz = 0; |
|
|
630 | |
|
|
631 | need (enc, 6); |
|
|
632 | |
|
|
633 | *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0; |
|
|
634 | u = i < 0 ? -i : i; |
|
|
635 | |
|
|
636 | // convert to 4.28 fixed-point representation |
|
|
637 | u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits |
|
|
638 | |
|
|
639 | // now output digit by digit, each time masking out the integer part |
|
|
640 | // and multiplying by 5 while moving the decimal point one to the right, |
|
|
641 | // resulting in a net multiplication by 10. |
|
|
642 | // we always write the digit to memory but conditionally increment |
|
|
643 | // the pointer, to ease the usage of conditional move instructions. |
|
|
644 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; |
|
|
645 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; |
|
|
646 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; |
|
|
647 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; |
|
|
648 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
|
|
649 | } |
497 | } |
650 | } |
498 | else if (SvROK (sv)) |
651 | else if (SvROK (sv)) |
499 | encode_rv (enc, SvRV (sv)); |
652 | encode_rv (enc, SvRV (sv)); |
500 | else if (!SvOK (sv)) |
653 | else if (!SvOK (sv)) |
501 | encode_str (enc, "null", 4, 0); |
654 | encode_str (enc, "null", 4, 0); |
… | |
… | |
503 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
656 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
504 | SvPV_nolen (sv), SvFLAGS (sv)); |
657 | SvPV_nolen (sv), SvFLAGS (sv)); |
505 | } |
658 | } |
506 | |
659 | |
507 | static SV * |
660 | static SV * |
508 | encode_json (SV *scalar, U32 flags) |
661 | encode_json (SV *scalar, JSON *json) |
509 | { |
662 | { |
510 | enc_t enc; |
663 | enc_t enc; |
511 | |
664 | |
512 | if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
665 | if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
513 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
666 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
514 | |
667 | |
515 | enc.flags = flags; |
668 | enc.json = *json; |
516 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
669 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
517 | enc.cur = SvPVX (enc.sv); |
670 | enc.cur = SvPVX (enc.sv); |
518 | enc.end = SvEND (enc.sv); |
671 | enc.end = SvEND (enc.sv); |
519 | enc.indent = 0; |
672 | enc.indent = 0; |
520 | enc.maxdepth = DEC_DEPTH (flags); |
673 | enc.maxdepth = DEC_DEPTH (enc.json.flags); |
521 | |
674 | |
522 | SvPOK_only (enc.sv); |
675 | SvPOK_only (enc.sv); |
523 | encode_sv (&enc, scalar); |
676 | encode_sv (&enc, scalar); |
524 | |
677 | |
525 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
678 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
526 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
679 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
527 | |
680 | |
528 | if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
681 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
529 | SvUTF8_on (enc.sv); |
682 | SvUTF8_on (enc.sv); |
530 | |
683 | |
531 | if (enc.flags & F_SHRINK) |
684 | if (enc.json.flags & F_SHRINK) |
532 | shrink (enc.sv); |
685 | shrink (enc.sv); |
533 | |
686 | |
534 | return enc.sv; |
687 | return enc.sv; |
535 | } |
688 | } |
536 | |
689 | |
… | |
… | |
541 | typedef struct |
694 | typedef struct |
542 | { |
695 | { |
543 | char *cur; // current parser pointer |
696 | char *cur; // current parser pointer |
544 | char *end; // end of input string |
697 | char *end; // end of input string |
545 | const char *err; // parse error, if != 0 |
698 | const char *err; // parse error, if != 0 |
546 | U32 flags; // F_* |
699 | JSON json; |
547 | U32 depth; // recursion depth |
700 | U32 depth; // recursion depth |
548 | U32 maxdepth; // recursion depth limit |
701 | U32 maxdepth; // recursion depth limit |
549 | } dec_t; |
702 | } dec_t; |
550 | |
703 | |
551 | static void |
704 | inline void |
|
|
705 | decode_comment (dec_t *dec) |
|
|
706 | { |
|
|
707 | // only '#'-style comments allowed a.t.m. |
|
|
708 | |
|
|
709 | while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) |
|
|
710 | ++dec->cur; |
|
|
711 | } |
|
|
712 | |
|
|
713 | inline void |
552 | decode_ws (dec_t *dec) |
714 | decode_ws (dec_t *dec) |
553 | { |
715 | { |
554 | for (;;) |
716 | for (;;) |
555 | { |
717 | { |
556 | char ch = *dec->cur; |
718 | char ch = *dec->cur; |
557 | |
719 | |
558 | if (ch > 0x20 |
720 | if (ch > 0x20) |
|
|
721 | { |
|
|
722 | if (expect_false (ch == '#')) |
|
|
723 | { |
|
|
724 | if (dec->json.flags & F_RELAXED) |
|
|
725 | decode_comment (dec); |
|
|
726 | else |
|
|
727 | break; |
|
|
728 | } |
|
|
729 | else |
|
|
730 | break; |
|
|
731 | } |
559 | || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) |
732 | else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09) |
560 | break; |
733 | break; // parse error, but let higher level handle it, gives better error messages |
561 | |
734 | |
562 | ++dec->cur; |
735 | ++dec->cur; |
563 | } |
736 | } |
564 | } |
737 | } |
565 | |
738 | |
… | |
… | |
582 | decode_4hex (dec_t *dec) |
755 | decode_4hex (dec_t *dec) |
583 | { |
756 | { |
584 | signed char d1, d2, d3, d4; |
757 | signed char d1, d2, d3, d4; |
585 | unsigned char *cur = (unsigned char *)dec->cur; |
758 | unsigned char *cur = (unsigned char *)dec->cur; |
586 | |
759 | |
587 | d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); |
760 | d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected"); |
588 | d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); |
761 | d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected"); |
589 | d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); |
762 | d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected"); |
590 | d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); |
763 | d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected"); |
591 | |
764 | |
592 | dec->cur += 4; |
765 | dec->cur += 4; |
593 | |
766 | |
594 | return ((UV)d1) << 12 |
767 | return ((UV)d1) << 12 |
595 | | ((UV)d2) << 8 |
768 | | ((UV)d2) << 8 |
… | |
… | |
603 | static SV * |
776 | static SV * |
604 | decode_str (dec_t *dec) |
777 | decode_str (dec_t *dec) |
605 | { |
778 | { |
606 | SV *sv = 0; |
779 | SV *sv = 0; |
607 | int utf8 = 0; |
780 | int utf8 = 0; |
|
|
781 | char *dec_cur = dec->cur; |
608 | |
782 | |
609 | do |
783 | do |
610 | { |
784 | { |
611 | char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; |
785 | char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; |
612 | char *cur = buf; |
786 | char *cur = buf; |
613 | |
787 | |
614 | do |
788 | do |
615 | { |
789 | { |
616 | unsigned char ch = *(unsigned char *)dec->cur++; |
790 | unsigned char ch = *(unsigned char *)dec_cur++; |
617 | |
791 | |
618 | if (ch == '"') |
792 | if (expect_false (ch == '"')) |
619 | { |
793 | { |
620 | --dec->cur; |
794 | --dec_cur; |
621 | break; |
795 | break; |
622 | } |
796 | } |
623 | else if (ch == '\\') |
797 | else if (expect_false (ch == '\\')) |
624 | { |
798 | { |
625 | switch (*dec->cur) |
799 | switch (*dec_cur) |
626 | { |
800 | { |
627 | case '\\': |
801 | case '\\': |
628 | case '/': |
802 | case '/': |
629 | case '"': *cur++ = *dec->cur++; break; |
803 | case '"': *cur++ = *dec_cur++; break; |
630 | |
804 | |
631 | case 'b': ++dec->cur; *cur++ = '\010'; break; |
805 | case 'b': ++dec_cur; *cur++ = '\010'; break; |
632 | case 't': ++dec->cur; *cur++ = '\011'; break; |
806 | case 't': ++dec_cur; *cur++ = '\011'; break; |
633 | case 'n': ++dec->cur; *cur++ = '\012'; break; |
807 | case 'n': ++dec_cur; *cur++ = '\012'; break; |
634 | case 'f': ++dec->cur; *cur++ = '\014'; break; |
808 | case 'f': ++dec_cur; *cur++ = '\014'; break; |
635 | case 'r': ++dec->cur; *cur++ = '\015'; break; |
809 | case 'r': ++dec_cur; *cur++ = '\015'; break; |
636 | |
810 | |
637 | case 'u': |
811 | case 'u': |
638 | { |
812 | { |
639 | UV lo, hi; |
813 | UV lo, hi; |
640 | ++dec->cur; |
814 | ++dec_cur; |
641 | |
815 | |
|
|
816 | dec->cur = dec_cur; |
642 | hi = decode_4hex (dec); |
817 | hi = decode_4hex (dec); |
|
|
818 | dec_cur = dec->cur; |
643 | if (hi == (UV)-1) |
819 | if (hi == (UV)-1) |
644 | goto fail; |
820 | goto fail; |
645 | |
821 | |
646 | // possibly a surrogate pair |
822 | // possibly a surrogate pair |
647 | if (hi >= 0xd800) |
823 | if (hi >= 0xd800) |
648 | if (hi < 0xdc00) |
824 | if (hi < 0xdc00) |
649 | { |
825 | { |
650 | if (dec->cur [0] != '\\' || dec->cur [1] != 'u') |
826 | if (dec_cur [0] != '\\' || dec_cur [1] != 'u') |
651 | ERR ("missing low surrogate character in surrogate pair"); |
827 | ERR ("missing low surrogate character in surrogate pair"); |
652 | |
828 | |
653 | dec->cur += 2; |
829 | dec_cur += 2; |
654 | |
830 | |
|
|
831 | dec->cur = dec_cur; |
655 | lo = decode_4hex (dec); |
832 | lo = decode_4hex (dec); |
|
|
833 | dec_cur = dec->cur; |
656 | if (lo == (UV)-1) |
834 | if (lo == (UV)-1) |
657 | goto fail; |
835 | goto fail; |
658 | |
836 | |
659 | if (lo < 0xdc00 || lo >= 0xe000) |
837 | if (lo < 0xdc00 || lo >= 0xe000) |
660 | ERR ("surrogate pair expected"); |
838 | ERR ("surrogate pair expected"); |
… | |
… | |
674 | *cur++ = hi; |
852 | *cur++ = hi; |
675 | } |
853 | } |
676 | break; |
854 | break; |
677 | |
855 | |
678 | default: |
856 | default: |
679 | --dec->cur; |
857 | --dec_cur; |
680 | ERR ("illegal backslash escape sequence in string"); |
858 | ERR ("illegal backslash escape sequence in string"); |
681 | } |
859 | } |
682 | } |
860 | } |
683 | else if (ch >= 0x20 && ch <= 0x7f) |
861 | else if (expect_true (ch >= 0x20 && ch <= 0x7f)) |
684 | *cur++ = ch; |
862 | *cur++ = ch; |
685 | else if (ch >= 0x80) |
863 | else if (ch >= 0x80) |
686 | { |
864 | { |
687 | STRLEN clen; |
865 | STRLEN clen; |
688 | UV uch; |
866 | UV uch; |
689 | |
867 | |
690 | --dec->cur; |
868 | --dec_cur; |
691 | |
869 | |
692 | uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); |
870 | uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
693 | if (clen == (STRLEN)-1) |
871 | if (clen == (STRLEN)-1) |
694 | ERR ("malformed UTF-8 character in JSON string"); |
872 | ERR ("malformed UTF-8 character in JSON string"); |
695 | |
873 | |
696 | do |
874 | do |
697 | *cur++ = *dec->cur++; |
875 | *cur++ = *dec_cur++; |
698 | while (--clen); |
876 | while (--clen); |
699 | |
877 | |
700 | utf8 = 1; |
878 | utf8 = 1; |
701 | } |
879 | } |
702 | else |
880 | else |
703 | { |
881 | { |
704 | --dec->cur; |
882 | --dec_cur; |
705 | |
883 | |
706 | if (!ch) |
884 | if (!ch) |
707 | ERR ("unexpected end of string while parsing JSON string"); |
885 | ERR ("unexpected end of string while parsing JSON string"); |
708 | else |
886 | else |
709 | ERR ("invalid character encountered while parsing JSON string"); |
887 | ERR ("invalid character encountered while parsing JSON string"); |
… | |
… | |
722 | } |
900 | } |
723 | else |
901 | else |
724 | sv = newSVpvn (buf, len); |
902 | sv = newSVpvn (buf, len); |
725 | } |
903 | } |
726 | } |
904 | } |
727 | while (*dec->cur != '"'); |
905 | while (*dec_cur != '"'); |
728 | |
906 | |
729 | ++dec->cur; |
907 | ++dec_cur; |
730 | |
908 | |
731 | if (sv) |
909 | if (sv) |
732 | { |
910 | { |
733 | SvPOK_only (sv); |
911 | SvPOK_only (sv); |
734 | *SvEND (sv) = 0; |
912 | *SvEND (sv) = 0; |
… | |
… | |
737 | SvUTF8_on (sv); |
915 | SvUTF8_on (sv); |
738 | } |
916 | } |
739 | else |
917 | else |
740 | sv = newSVpvn ("", 0); |
918 | sv = newSVpvn ("", 0); |
741 | |
919 | |
|
|
920 | dec->cur = dec_cur; |
742 | return sv; |
921 | return sv; |
743 | |
922 | |
744 | fail: |
923 | fail: |
|
|
924 | dec->cur = dec_cur; |
745 | return 0; |
925 | return 0; |
746 | } |
926 | } |
747 | |
927 | |
748 | static SV * |
928 | static SV * |
749 | decode_num (dec_t *dec) |
929 | decode_num (dec_t *dec) |
… | |
… | |
807 | is_nv = 1; |
987 | is_nv = 1; |
808 | } |
988 | } |
809 | |
989 | |
810 | if (!is_nv) |
990 | if (!is_nv) |
811 | { |
991 | { |
812 | UV uv; |
992 | int len = dec->cur - start; |
813 | int numtype = grok_number (start, dec->cur - start, &uv); |
993 | |
814 | if (numtype & IS_NUMBER_IN_UV) |
994 | // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so |
815 | if (numtype & IS_NUMBER_NEG) |
995 | if (*start == '-') |
|
|
996 | switch (len) |
816 | { |
997 | { |
817 | if (uv < (UV)IV_MIN) |
998 | case 2: return newSViv (-( start [1] - '0' * 1)); |
818 | return newSViv (-(IV)uv); |
999 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
|
|
1000 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
|
|
1001 | case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
819 | } |
1002 | } |
|
|
1003 | else |
|
|
1004 | switch (len) |
|
|
1005 | { |
|
|
1006 | case 1: return newSViv ( start [0] - '0' * 1); |
|
|
1007 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
|
|
1008 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
|
|
1009 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
|
|
1010 | } |
|
|
1011 | |
|
|
1012 | { |
|
|
1013 | UV uv; |
|
|
1014 | int numtype = grok_number (start, len, &uv); |
|
|
1015 | if (numtype & IS_NUMBER_IN_UV) |
|
|
1016 | if (numtype & IS_NUMBER_NEG) |
|
|
1017 | { |
|
|
1018 | if (uv < (UV)IV_MIN) |
|
|
1019 | return newSViv (-(IV)uv); |
|
|
1020 | } |
820 | else |
1021 | else |
821 | return newSVuv (uv); |
1022 | return newSVuv (uv); |
822 | } |
1023 | } |
823 | |
1024 | |
|
|
1025 | len -= *start == '-' ? 1 : 0; |
|
|
1026 | |
|
|
1027 | // does not fit into IV or UV, try NV |
|
|
1028 | if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) |
|
|
1029 | #if defined (LDBL_DIG) |
|
|
1030 | || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len) |
|
|
1031 | #endif |
|
|
1032 | ) |
|
|
1033 | // fits into NV without loss of precision |
|
|
1034 | return newSVnv (Atof (start)); |
|
|
1035 | |
|
|
1036 | // everything else fails, convert it to a string |
|
|
1037 | return newSVpvn (start, dec->cur - start); |
|
|
1038 | } |
|
|
1039 | |
|
|
1040 | // loss of precision here |
824 | return newSVnv (Atof (start)); |
1041 | return newSVnv (Atof (start)); |
825 | |
1042 | |
826 | fail: |
1043 | fail: |
827 | return 0; |
1044 | return 0; |
828 | } |
1045 | } |
… | |
… | |
858 | |
1075 | |
859 | if (*dec->cur != ',') |
1076 | if (*dec->cur != ',') |
860 | ERR (", or ] expected while parsing array"); |
1077 | ERR (", or ] expected while parsing array"); |
861 | |
1078 | |
862 | ++dec->cur; |
1079 | ++dec->cur; |
|
|
1080 | |
|
|
1081 | decode_ws (dec); |
|
|
1082 | |
|
|
1083 | if (*dec->cur == ']' && dec->json.flags & F_RELAXED) |
|
|
1084 | { |
|
|
1085 | ++dec->cur; |
|
|
1086 | break; |
|
|
1087 | } |
863 | } |
1088 | } |
864 | |
1089 | |
865 | DEC_DEC_DEPTH; |
1090 | DEC_DEC_DEPTH; |
866 | return newRV_noinc ((SV *)av); |
1091 | return newRV_noinc ((SV *)av); |
867 | |
1092 | |
… | |
… | |
872 | } |
1097 | } |
873 | |
1098 | |
874 | static SV * |
1099 | static SV * |
875 | decode_hv (dec_t *dec) |
1100 | decode_hv (dec_t *dec) |
876 | { |
1101 | { |
|
|
1102 | SV *sv; |
877 | HV *hv = newHV (); |
1103 | HV *hv = newHV (); |
878 | |
1104 | |
879 | DEC_INC_DEPTH; |
1105 | DEC_INC_DEPTH; |
880 | decode_ws (dec); |
1106 | decode_ws (dec); |
881 | |
1107 | |
882 | if (*dec->cur == '}') |
1108 | if (*dec->cur == '}') |
883 | ++dec->cur; |
1109 | ++dec->cur; |
884 | else |
1110 | else |
885 | for (;;) |
1111 | for (;;) |
886 | { |
1112 | { |
|
|
1113 | EXPECT_CH ('"'); |
|
|
1114 | |
|
|
1115 | // heuristic: assume that |
|
|
1116 | // a) decode_str + hv_store_ent are abysmally slow. |
|
|
1117 | // b) most hash keys are short, simple ascii text. |
|
|
1118 | // => try to "fast-match" such strings to avoid |
|
|
1119 | // the overhead of decode_str + hv_store_ent. |
|
|
1120 | { |
887 | SV *key, *value; |
1121 | SV *value; |
|
|
1122 | char *p = dec->cur; |
|
|
1123 | char *e = p + 24; // only try up to 24 bytes |
888 | |
1124 | |
889 | decode_ws (dec); EXPECT_CH ('"'); |
1125 | for (;;) |
890 | |
|
|
891 | key = decode_str (dec); |
|
|
892 | if (!key) |
|
|
893 | goto fail; |
|
|
894 | |
|
|
895 | decode_ws (dec); EXPECT_CH (':'); |
|
|
896 | |
|
|
897 | value = decode_sv (dec); |
|
|
898 | if (!value) |
|
|
899 | { |
1126 | { |
|
|
1127 | // the >= 0x80 is true on most architectures |
|
|
1128 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
|
|
1129 | { |
|
|
1130 | // slow path, back up and use decode_str |
|
|
1131 | SV *key = decode_str (dec); |
|
|
1132 | if (!key) |
|
|
1133 | goto fail; |
|
|
1134 | |
|
|
1135 | decode_ws (dec); EXPECT_CH (':'); |
|
|
1136 | |
|
|
1137 | decode_ws (dec); |
|
|
1138 | value = decode_sv (dec); |
|
|
1139 | if (!value) |
|
|
1140 | { |
|
|
1141 | SvREFCNT_dec (key); |
|
|
1142 | goto fail; |
|
|
1143 | } |
|
|
1144 | |
|
|
1145 | hv_store_ent (hv, key, value, 0); |
900 | SvREFCNT_dec (key); |
1146 | SvREFCNT_dec (key); |
|
|
1147 | |
|
|
1148 | break; |
|
|
1149 | } |
|
|
1150 | else if (*p == '"') |
|
|
1151 | { |
|
|
1152 | // fast path, got a simple key |
|
|
1153 | char *key = dec->cur; |
|
|
1154 | int len = p - key; |
|
|
1155 | dec->cur = p + 1; |
|
|
1156 | |
|
|
1157 | decode_ws (dec); EXPECT_CH (':'); |
|
|
1158 | |
|
|
1159 | decode_ws (dec); |
|
|
1160 | value = decode_sv (dec); |
|
|
1161 | if (!value) |
901 | goto fail; |
1162 | goto fail; |
|
|
1163 | |
|
|
1164 | hv_store (hv, key, len, value, 0); |
|
|
1165 | |
|
|
1166 | break; |
|
|
1167 | } |
|
|
1168 | |
|
|
1169 | ++p; |
902 | } |
1170 | } |
903 | |
1171 | } |
904 | hv_store_ent (hv, key, value, 0); |
|
|
905 | SvREFCNT_dec (key); |
|
|
906 | |
1172 | |
907 | decode_ws (dec); |
1173 | decode_ws (dec); |
908 | |
1174 | |
909 | if (*dec->cur == '}') |
1175 | if (*dec->cur == '}') |
910 | { |
1176 | { |
… | |
… | |
914 | |
1180 | |
915 | if (*dec->cur != ',') |
1181 | if (*dec->cur != ',') |
916 | ERR (", or } expected while parsing object/hash"); |
1182 | ERR (", or } expected while parsing object/hash"); |
917 | |
1183 | |
918 | ++dec->cur; |
1184 | ++dec->cur; |
|
|
1185 | |
|
|
1186 | decode_ws (dec); |
|
|
1187 | |
|
|
1188 | if (*dec->cur == '}' && dec->json.flags & F_RELAXED) |
|
|
1189 | { |
|
|
1190 | ++dec->cur; |
|
|
1191 | break; |
|
|
1192 | } |
919 | } |
1193 | } |
920 | |
1194 | |
921 | DEC_DEC_DEPTH; |
1195 | DEC_DEC_DEPTH; |
922 | return newRV_noinc ((SV *)hv); |
1196 | sv = newRV_noinc ((SV *)hv); |
|
|
1197 | |
|
|
1198 | // check filter callbacks |
|
|
1199 | if (dec->json.flags & F_HOOK) |
|
|
1200 | { |
|
|
1201 | if (dec->json.cb_sk_object && HvKEYS (hv) == 1) |
|
|
1202 | { |
|
|
1203 | HE *cb, *he; |
|
|
1204 | |
|
|
1205 | hv_iterinit (hv); |
|
|
1206 | he = hv_iternext (hv); |
|
|
1207 | hv_iterinit (hv); |
|
|
1208 | |
|
|
1209 | // the next line creates a mortal sv each time its called. |
|
|
1210 | // might want to optimise this for common cases. |
|
|
1211 | cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); |
|
|
1212 | |
|
|
1213 | if (cb) |
|
|
1214 | { |
|
|
1215 | dSP; |
|
|
1216 | int count; |
|
|
1217 | |
|
|
1218 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
1219 | XPUSHs (HeVAL (he)); |
|
|
1220 | |
|
|
1221 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
|
|
1222 | |
|
|
1223 | if (count == 1) |
|
|
1224 | { |
|
|
1225 | sv = newSVsv (POPs); |
|
|
1226 | FREETMPS; LEAVE; |
|
|
1227 | return sv; |
|
|
1228 | } |
|
|
1229 | |
|
|
1230 | FREETMPS; LEAVE; |
|
|
1231 | } |
|
|
1232 | } |
|
|
1233 | |
|
|
1234 | if (dec->json.cb_object) |
|
|
1235 | { |
|
|
1236 | dSP; |
|
|
1237 | int count; |
|
|
1238 | |
|
|
1239 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
1240 | XPUSHs (sv_2mortal (sv)); |
|
|
1241 | |
|
|
1242 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
|
|
1243 | |
|
|
1244 | if (count == 1) |
|
|
1245 | { |
|
|
1246 | sv = newSVsv (POPs); |
|
|
1247 | FREETMPS; LEAVE; |
|
|
1248 | return sv; |
|
|
1249 | } |
|
|
1250 | |
|
|
1251 | SvREFCNT_inc (sv); |
|
|
1252 | FREETMPS; LEAVE; |
|
|
1253 | } |
|
|
1254 | } |
|
|
1255 | |
|
|
1256 | return sv; |
923 | |
1257 | |
924 | fail: |
1258 | fail: |
925 | SvREFCNT_dec (hv); |
1259 | SvREFCNT_dec (hv); |
926 | DEC_DEC_DEPTH; |
1260 | DEC_DEC_DEPTH; |
927 | return 0; |
1261 | return 0; |
928 | } |
1262 | } |
929 | |
1263 | |
930 | static SV * |
1264 | static SV * |
931 | decode_sv (dec_t *dec) |
1265 | decode_sv (dec_t *dec) |
932 | { |
1266 | { |
933 | decode_ws (dec); |
1267 | // the beauty of JSON: you need exactly one character lookahead |
|
|
1268 | // to parse anything. |
934 | switch (*dec->cur) |
1269 | switch (*dec->cur) |
935 | { |
1270 | { |
936 | case '"': ++dec->cur; return decode_str (dec); |
1271 | case '"': ++dec->cur; return decode_str (dec); |
937 | case '[': ++dec->cur; return decode_av (dec); |
1272 | case '[': ++dec->cur; return decode_av (dec); |
938 | case '{': ++dec->cur; return decode_hv (dec); |
1273 | case '{': ++dec->cur; return decode_hv (dec); |
… | |
… | |
944 | |
1279 | |
945 | case 't': |
1280 | case 't': |
946 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1281 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
947 | { |
1282 | { |
948 | dec->cur += 4; |
1283 | dec->cur += 4; |
949 | return newSViv (1); |
1284 | #if JSON_SLOW |
|
|
1285 | json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); |
|
|
1286 | #endif |
|
|
1287 | return SvREFCNT_inc (json_true); |
950 | } |
1288 | } |
951 | else |
1289 | else |
952 | ERR ("'true' expected"); |
1290 | ERR ("'true' expected"); |
953 | |
1291 | |
954 | break; |
1292 | break; |
955 | |
1293 | |
956 | case 'f': |
1294 | case 'f': |
957 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1295 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
958 | { |
1296 | { |
959 | dec->cur += 5; |
1297 | dec->cur += 5; |
960 | return newSViv (0); |
1298 | #if JSON_SLOW |
|
|
1299 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
|
|
1300 | #endif |
|
|
1301 | return SvREFCNT_inc (json_false); |
961 | } |
1302 | } |
962 | else |
1303 | else |
963 | ERR ("'false' expected"); |
1304 | ERR ("'false' expected"); |
964 | |
1305 | |
965 | break; |
1306 | break; |
… | |
… | |
983 | fail: |
1324 | fail: |
984 | return 0; |
1325 | return 0; |
985 | } |
1326 | } |
986 | |
1327 | |
987 | static SV * |
1328 | static SV * |
988 | decode_json (SV *string, U32 flags, UV *offset_return) |
1329 | decode_json (SV *string, JSON *json, UV *offset_return) |
989 | { |
1330 | { |
990 | dec_t dec; |
1331 | dec_t dec; |
991 | UV offset; |
1332 | UV offset; |
992 | SV *sv; |
1333 | SV *sv; |
993 | |
1334 | |
994 | SvGETMAGIC (string); |
1335 | SvGETMAGIC (string); |
995 | SvUPGRADE (string, SVt_PV); |
1336 | SvUPGRADE (string, SVt_PV); |
996 | |
1337 | |
|
|
1338 | if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) |
|
|
1339 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
|
|
1340 | (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); |
|
|
1341 | |
997 | if (flags & F_UTF8) |
1342 | if (json->flags & F_UTF8) |
998 | sv_utf8_downgrade (string, 0); |
1343 | sv_utf8_downgrade (string, 0); |
999 | else |
1344 | else |
1000 | sv_utf8_upgrade (string); |
1345 | sv_utf8_upgrade (string); |
1001 | |
1346 | |
1002 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1347 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1003 | |
1348 | |
1004 | dec.flags = flags; |
1349 | dec.json = *json; |
1005 | dec.cur = SvPVX (string); |
1350 | dec.cur = SvPVX (string); |
1006 | dec.end = SvEND (string); |
1351 | dec.end = SvEND (string); |
1007 | dec.err = 0; |
1352 | dec.err = 0; |
1008 | dec.depth = 0; |
1353 | dec.depth = 0; |
1009 | dec.maxdepth = DEC_DEPTH (dec.flags); |
1354 | dec.maxdepth = DEC_DEPTH (dec.json.flags); |
|
|
1355 | |
|
|
1356 | if (dec.json.cb_object || dec.json.cb_sk_object) |
|
|
1357 | dec.json.flags |= F_HOOK; |
1010 | |
1358 | |
1011 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1359 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
|
|
1360 | |
|
|
1361 | decode_ws (&dec); |
1012 | sv = decode_sv (&dec); |
1362 | sv = decode_sv (&dec); |
1013 | |
1363 | |
1014 | if (offset_return || !sv) |
1364 | if (!(offset_return || !sv)) |
1015 | { |
|
|
1016 | offset = dec.flags & F_UTF8 |
|
|
1017 | ? dec.cur - SvPVX (string) |
|
|
1018 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
1019 | |
|
|
1020 | if (offset_return) |
|
|
1021 | *offset_return = offset; |
|
|
1022 | } |
|
|
1023 | else |
|
|
1024 | { |
1365 | { |
1025 | // check for trailing garbage |
1366 | // check for trailing garbage |
1026 | decode_ws (&dec); |
1367 | decode_ws (&dec); |
1027 | |
1368 | |
1028 | if (*dec.cur) |
1369 | if (*dec.cur) |
1029 | { |
1370 | { |
1030 | dec.err = "garbage after JSON object"; |
1371 | dec.err = "garbage after JSON object"; |
1031 | SvREFCNT_dec (sv); |
1372 | SvREFCNT_dec (sv); |
1032 | sv = 0; |
1373 | sv = 0; |
1033 | } |
1374 | } |
|
|
1375 | } |
|
|
1376 | |
|
|
1377 | if (offset_return || !sv) |
|
|
1378 | { |
|
|
1379 | offset = dec.json.flags & F_UTF8 |
|
|
1380 | ? dec.cur - SvPVX (string) |
|
|
1381 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
1382 | |
|
|
1383 | if (offset_return) |
|
|
1384 | *offset_return = offset; |
1034 | } |
1385 | } |
1035 | |
1386 | |
1036 | if (!sv) |
1387 | if (!sv) |
1037 | { |
1388 | { |
1038 | SV *uni = sv_newmortal (); |
1389 | SV *uni = sv_newmortal (); |
… | |
… | |
1052 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1403 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1053 | } |
1404 | } |
1054 | |
1405 | |
1055 | sv = sv_2mortal (sv); |
1406 | sv = sv_2mortal (sv); |
1056 | |
1407 | |
1057 | if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1408 | if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1058 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1409 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1059 | |
1410 | |
1060 | return sv; |
1411 | return sv; |
1061 | } |
1412 | } |
1062 | |
1413 | |
… | |
… | |
1066 | MODULE = JSON::XS PACKAGE = JSON::XS |
1417 | MODULE = JSON::XS PACKAGE = JSON::XS |
1067 | |
1418 | |
1068 | BOOT: |
1419 | BOOT: |
1069 | { |
1420 | { |
1070 | int i; |
1421 | int i; |
1071 | |
|
|
1072 | memset (decode_hexdigit, 0xff, 256); |
|
|
1073 | |
1422 | |
1074 | for (i = 0; i < 256; ++i) |
1423 | for (i = 0; i < 256; ++i) |
1075 | decode_hexdigit [i] = |
1424 | decode_hexdigit [i] = |
1076 | i >= '0' && i <= '9' ? i - '0' |
1425 | i >= '0' && i <= '9' ? i - '0' |
1077 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1426 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1078 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1427 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1079 | : -1; |
1428 | : -1; |
1080 | |
1429 | |
1081 | json_stash = gv_stashpv ("JSON::XS", 1); |
1430 | json_stash = gv_stashpv ("JSON::XS" , 1); |
|
|
1431 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
|
|
1432 | |
|
|
1433 | json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); |
|
|
1434 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1082 | } |
1435 | } |
1083 | |
1436 | |
1084 | PROTOTYPES: DISABLE |
1437 | PROTOTYPES: DISABLE |
1085 | |
1438 | |
1086 | SV *new (char *dummy) |
1439 | void CLONE (...) |
1087 | CODE: |
1440 | CODE: |
1088 | RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); |
1441 | json_stash = 0; |
1089 | OUTPUT: |
1442 | json_boolean_stash = 0; |
1090 | RETVAL |
|
|
1091 | |
1443 | |
|
|
1444 | void new (char *klass) |
|
|
1445 | PPCODE: |
|
|
1446 | { |
|
|
1447 | SV *pv = NEWSV (0, sizeof (JSON)); |
|
|
1448 | SvPOK_only (pv); |
|
|
1449 | Zero (SvPVX (pv), 1, JSON); |
|
|
1450 | ((JSON *)SvPVX (pv))->flags = F_DEFAULT; |
|
|
1451 | XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH))); |
|
|
1452 | } |
|
|
1453 | |
1092 | SV *ascii (SV *self, int enable = 1) |
1454 | void ascii (JSON *self, int enable = 1) |
1093 | ALIAS: |
1455 | ALIAS: |
1094 | ascii = F_ASCII |
1456 | ascii = F_ASCII |
1095 | latin1 = F_LATIN1 |
1457 | latin1 = F_LATIN1 |
1096 | utf8 = F_UTF8 |
1458 | utf8 = F_UTF8 |
1097 | indent = F_INDENT |
1459 | indent = F_INDENT |
1098 | canonical = F_CANONICAL |
1460 | canonical = F_CANONICAL |
1099 | space_before = F_SPACE_BEFORE |
1461 | space_before = F_SPACE_BEFORE |
1100 | space_after = F_SPACE_AFTER |
1462 | space_after = F_SPACE_AFTER |
1101 | pretty = F_PRETTY |
1463 | pretty = F_PRETTY |
1102 | allow_nonref = F_ALLOW_NONREF |
1464 | allow_nonref = F_ALLOW_NONREF |
1103 | shrink = F_SHRINK |
1465 | shrink = F_SHRINK |
|
|
1466 | allow_blessed = F_ALLOW_BLESSED |
|
|
1467 | convert_blessed = F_CONV_BLESSED |
|
|
1468 | relaxed = F_RELAXED |
1104 | CODE: |
1469 | PPCODE: |
1105 | { |
1470 | { |
1106 | UV *uv = SvJSON (self); |
|
|
1107 | if (enable) |
1471 | if (enable) |
1108 | *uv |= ix; |
1472 | self->flags |= ix; |
1109 | else |
1473 | else |
1110 | *uv &= ~ix; |
1474 | self->flags &= ~ix; |
1111 | |
1475 | |
1112 | RETVAL = newSVsv (self); |
1476 | XPUSHs (ST (0)); |
1113 | } |
1477 | } |
1114 | OUTPUT: |
|
|
1115 | RETVAL |
|
|
1116 | |
1478 | |
1117 | SV *max_depth (SV *self, UV max_depth = 0x80000000UL) |
1479 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
1118 | CODE: |
1480 | PPCODE: |
1119 | { |
1481 | { |
1120 | UV *uv = SvJSON (self); |
|
|
1121 | UV log2 = 0; |
1482 | UV log2 = 0; |
1122 | |
1483 | |
1123 | if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
1484 | if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
1124 | |
1485 | |
1125 | while ((1UL << log2) < max_depth) |
1486 | while ((1UL << log2) < max_depth) |
1126 | ++log2; |
1487 | ++log2; |
1127 | |
1488 | |
1128 | *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
1489 | self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
1129 | |
1490 | |
1130 | RETVAL = newSVsv (self); |
1491 | XPUSHs (ST (0)); |
1131 | } |
1492 | } |
1132 | OUTPUT: |
|
|
1133 | RETVAL |
|
|
1134 | |
1493 | |
1135 | void encode (SV *self, SV *scalar) |
1494 | void max_size (JSON *self, UV max_size = 0) |
1136 | PPCODE: |
1495 | PPCODE: |
1137 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
1496 | { |
|
|
1497 | UV log2 = 0; |
1138 | |
1498 | |
1139 | void decode (SV *self, SV *jsonstr) |
1499 | if (max_size > 0x80000000UL) max_size = 0x80000000UL; |
|
|
1500 | if (max_size == 1) max_size = 2; |
|
|
1501 | |
|
|
1502 | while ((1UL << log2) < max_size) |
|
|
1503 | ++log2; |
|
|
1504 | |
|
|
1505 | self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); |
|
|
1506 | |
|
|
1507 | XPUSHs (ST (0)); |
|
|
1508 | } |
|
|
1509 | |
|
|
1510 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1140 | PPCODE: |
1511 | PPCODE: |
|
|
1512 | { |
|
|
1513 | SvREFCNT_dec (self->cb_object); |
|
|
1514 | self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; |
|
|
1515 | |
|
|
1516 | XPUSHs (ST (0)); |
|
|
1517 | } |
|
|
1518 | |
|
|
1519 | void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) |
|
|
1520 | PPCODE: |
|
|
1521 | { |
|
|
1522 | if (!self->cb_sk_object) |
|
|
1523 | self->cb_sk_object = newHV (); |
|
|
1524 | |
|
|
1525 | if (SvOK (cb)) |
|
|
1526 | hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
|
|
1527 | else |
|
|
1528 | { |
|
|
1529 | hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0); |
|
|
1530 | |
|
|
1531 | if (!HvKEYS (self->cb_sk_object)) |
|
|
1532 | { |
|
|
1533 | SvREFCNT_dec (self->cb_sk_object); |
|
|
1534 | self->cb_sk_object = 0; |
|
|
1535 | } |
|
|
1536 | } |
|
|
1537 | |
|
|
1538 | XPUSHs (ST (0)); |
|
|
1539 | } |
|
|
1540 | |
|
|
1541 | void encode (JSON *self, SV *scalar) |
|
|
1542 | PPCODE: |
|
|
1543 | XPUSHs (encode_json (scalar, self)); |
|
|
1544 | |
|
|
1545 | void decode (JSON *self, SV *jsonstr) |
|
|
1546 | PPCODE: |
1141 | XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); |
1547 | XPUSHs (decode_json (jsonstr, self, 0)); |
1142 | |
1548 | |
1143 | void decode_prefix (SV *self, SV *jsonstr) |
1549 | void decode_prefix (JSON *self, SV *jsonstr) |
1144 | PPCODE: |
1550 | PPCODE: |
1145 | { |
1551 | { |
1146 | UV offset; |
1552 | UV offset; |
1147 | EXTEND (SP, 2); |
1553 | EXTEND (SP, 2); |
1148 | PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); |
1554 | PUSHs (decode_json (jsonstr, self, &offset)); |
1149 | PUSHs (sv_2mortal (newSVuv (offset))); |
1555 | PUSHs (sv_2mortal (newSVuv (offset))); |
1150 | } |
1556 | } |
1151 | |
1557 | |
|
|
1558 | void DESTROY (JSON *self) |
|
|
1559 | CODE: |
|
|
1560 | SvREFCNT_dec (self->cb_sk_object); |
|
|
1561 | SvREFCNT_dec (self->cb_object); |
|
|
1562 | |
1152 | PROTOTYPES: ENABLE |
1563 | PROTOTYPES: ENABLE |
1153 | |
1564 | |
1154 | void to_json (SV *scalar) |
1565 | void to_json (SV *scalar) |
1155 | ALIAS: |
|
|
1156 | objToJson = 0 |
|
|
1157 | PPCODE: |
1566 | PPCODE: |
|
|
1567 | { |
|
|
1568 | JSON json = { F_DEFAULT | F_UTF8 }; |
1158 | XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); |
1569 | XPUSHs (encode_json (scalar, &json)); |
|
|
1570 | } |
1159 | |
1571 | |
1160 | void from_json (SV *jsonstr) |
1572 | void from_json (SV *jsonstr) |
1161 | ALIAS: |
|
|
1162 | jsonToObj = 0 |
|
|
1163 | PPCODE: |
1573 | PPCODE: |
|
|
1574 | { |
|
|
1575 | JSON json = { F_DEFAULT | F_UTF8 }; |
1164 | XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0)); |
1576 | XPUSHs (decode_json (jsonstr, &json, 0)); |
|
|
1577 | } |
1165 | |
1578 | |