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> |
|
|
9 | #include <float.h> |
8 | |
10 | |
|
|
11 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
|
|
12 | # define snprintf _snprintf // C compilers have this in stdio.h |
|
|
13 | #endif |
|
|
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 | |
9 | #define F_ASCII 0x00000001 |
21 | #define F_ASCII 0x00000001UL |
|
|
22 | #define F_LATIN1 0x00000002UL |
10 | #define F_UTF8 0x00000002 |
23 | #define F_UTF8 0x00000004UL |
11 | #define F_INDENT 0x00000004 |
24 | #define F_INDENT 0x00000008UL |
12 | #define F_CANONICAL 0x00000008 |
25 | #define F_CANONICAL 0x00000010UL |
13 | #define F_SPACE_BEFORE 0x00000010 |
26 | #define F_SPACE_BEFORE 0x00000020UL |
14 | #define F_SPACE_AFTER 0x00000020 |
27 | #define F_SPACE_AFTER 0x00000040UL |
15 | #define F_ALLOW_NONREF 0x00000080 |
28 | #define F_ALLOW_NONREF 0x00000100UL |
16 | #define F_SHRINK 0x00000100 |
29 | #define F_SHRINK 0x00000200UL |
|
|
30 | #define F_ALLOW_BLESSED 0x00000400UL |
|
|
31 | #define F_CONV_BLESSED 0x00000800UL |
|
|
32 | #define F_RELAXED 0x00001000UL |
|
|
33 | |
|
|
34 | #define F_MAXDEPTH 0xf8000000UL |
|
|
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 |
|
|
39 | |
|
|
40 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
|
|
41 | #define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE )) |
17 | |
42 | |
18 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
43 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
19 | #define F_DEFAULT 0 |
44 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
20 | |
45 | |
21 | #define INIT_SIZE 32 // initial scalar size to be allocated |
46 | #define INIT_SIZE 32 // initial scalar size to be allocated |
22 | #define INDENT_STEP 3 // spaces per indentation level |
47 | #define INDENT_STEP 3 // spaces per indentation level |
23 | |
48 | |
24 | #define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character |
|
|
25 | #define SHORT_STRING_LEN 256 // special-case strings of up to this size |
49 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
26 | |
50 | |
27 | #define SB do { |
51 | #define SB do { |
28 | #define SE } while (0) |
52 | #define SE } while (0) |
29 | |
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 | |
30 | 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; |
31 | |
81 | |
32 | ///////////////////////////////////////////////////////////////////////////// |
82 | ///////////////////////////////////////////////////////////////////////////// |
33 | // utility functions |
83 | // utility functions |
34 | |
84 | |
35 | static UV * |
85 | inline void |
36 | SvJSON (SV *sv) |
|
|
37 | { |
|
|
38 | if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) |
|
|
39 | croak ("object is not of type JSON::XS"); |
|
|
40 | |
|
|
41 | return &SvUVX (SvRV (sv)); |
|
|
42 | } |
|
|
43 | |
|
|
44 | static void |
|
|
45 | shrink (SV *sv) |
86 | shrink (SV *sv) |
46 | { |
87 | { |
47 | sv_utf8_downgrade (sv, 1); |
88 | sv_utf8_downgrade (sv, 1); |
48 | if (SvLEN (sv) > SvCUR (sv) + 1) |
89 | if (SvLEN (sv) > SvCUR (sv) + 1) |
49 | { |
90 | { |
… | |
… | |
58 | // 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 |
59 | // case of an error. |
100 | // case of an error. |
60 | // we special-case "safe" characters from U+80 .. U+7FF, |
101 | // we special-case "safe" characters from U+80 .. U+7FF, |
61 | // but use the very good perl function to parse anything else. |
102 | // but use the very good perl function to parse anything else. |
62 | // note that we never call this function for a ascii codepoints |
103 | // note that we never call this function for a ascii codepoints |
63 | static UV |
104 | inline UV |
64 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
105 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
65 | { |
106 | { |
66 | if (s[0] > 0xdf || s[0] < 0xc2) |
107 | if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) |
67 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
108 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
68 | else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
109 | else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
69 | { |
110 | { |
70 | *clen = 2; |
111 | *clen = 2; |
71 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
112 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
72 | } |
113 | } |
73 | else |
114 | else |
|
|
115 | { |
|
|
116 | *clen = (STRLEN)-1; |
74 | return (UV)-1; |
117 | return (UV)-1; |
|
|
118 | } |
75 | } |
119 | } |
76 | |
120 | |
77 | ///////////////////////////////////////////////////////////////////////////// |
121 | ///////////////////////////////////////////////////////////////////////////// |
78 | // encoder |
122 | // encoder |
79 | |
123 | |
… | |
… | |
81 | typedef struct |
125 | typedef struct |
82 | { |
126 | { |
83 | char *cur; // SvPVX (sv) + current output position |
127 | char *cur; // SvPVX (sv) + current output position |
84 | char *end; // SvEND (sv) |
128 | char *end; // SvEND (sv) |
85 | SV *sv; // result scalar |
129 | SV *sv; // result scalar |
86 | UV flags; // F_* |
130 | JSON json; |
87 | int indent; // indentation level |
131 | U32 indent; // indentation level |
88 | int max_depth; // max. recursion level |
132 | U32 maxdepth; // max. indentation/recursion level |
89 | } enc_t; |
133 | } enc_t; |
90 | |
134 | |
91 | static void |
135 | inline void |
92 | need (enc_t *enc, STRLEN len) |
136 | need (enc_t *enc, STRLEN len) |
93 | { |
137 | { |
94 | if (enc->cur + len >= enc->end) |
138 | if (expect_false (enc->cur + len >= enc->end)) |
95 | { |
139 | { |
96 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
140 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
97 | SvGROW (enc->sv, cur + len + 1); |
141 | SvGROW (enc->sv, cur + len + 1); |
98 | enc->cur = SvPVX (enc->sv) + cur; |
142 | enc->cur = SvPVX (enc->sv) + cur; |
99 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); |
143 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
100 | } |
144 | } |
101 | } |
145 | } |
102 | |
146 | |
103 | static void |
147 | inline void |
104 | encode_ch (enc_t *enc, char ch) |
148 | encode_ch (enc_t *enc, char ch) |
105 | { |
149 | { |
106 | need (enc, 1); |
150 | need (enc, 1); |
107 | *enc->cur++ = ch; |
151 | *enc->cur++ = ch; |
108 | } |
152 | } |
… | |
… | |
116 | |
160 | |
117 | while (str < end) |
161 | while (str < end) |
118 | { |
162 | { |
119 | unsigned char ch = *(unsigned char *)str; |
163 | unsigned char ch = *(unsigned char *)str; |
120 | |
164 | |
121 | if (ch >= 0x20 && ch < 0x80) // most common case |
165 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
122 | { |
166 | { |
123 | if (ch == '"') // but with slow exceptions |
167 | if (expect_false (ch == '"')) // but with slow exceptions |
124 | { |
168 | { |
125 | need (enc, len += 1); |
169 | need (enc, len += 1); |
126 | *enc->cur++ = '\\'; |
170 | *enc->cur++ = '\\'; |
127 | *enc->cur++ = '"'; |
171 | *enc->cur++ = '"'; |
128 | } |
172 | } |
129 | else if (ch == '\\') |
173 | else if (expect_false (ch == '\\')) |
130 | { |
174 | { |
131 | need (enc, len += 1); |
175 | need (enc, len += 1); |
132 | *enc->cur++ = '\\'; |
176 | *enc->cur++ = '\\'; |
133 | *enc->cur++ = '\\'; |
177 | *enc->cur++ = '\\'; |
134 | } |
178 | } |
… | |
… | |
152 | STRLEN clen; |
196 | STRLEN clen; |
153 | UV uch; |
197 | UV uch; |
154 | |
198 | |
155 | if (is_utf8) |
199 | if (is_utf8) |
156 | { |
200 | { |
157 | //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); |
|
|
158 | uch = decode_utf8 (str, end - str, &clen); |
201 | uch = decode_utf8 (str, end - str, &clen); |
159 | if (clen == (STRLEN)-1) |
202 | if (clen == (STRLEN)-1) |
160 | 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); |
161 | } |
204 | } |
162 | else |
205 | else |
… | |
… | |
166 | } |
209 | } |
167 | |
210 | |
168 | if (uch > 0x10FFFFUL) |
211 | if (uch > 0x10FFFFUL) |
169 | 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); |
170 | |
213 | |
171 | if (uch < 0x80 || enc->flags & F_ASCII) |
214 | if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF)) |
172 | { |
215 | { |
173 | if (uch > 0xFFFFUL) |
216 | if (uch > 0xFFFFUL) |
174 | { |
217 | { |
175 | need (enc, len += 11); |
218 | need (enc, len += 11); |
176 | sprintf (enc->cur, "\\u%04x\\u%04x", |
219 | sprintf (enc->cur, "\\u%04x\\u%04x", |
… | |
… | |
190 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
233 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
191 | } |
234 | } |
192 | |
235 | |
193 | str += clen; |
236 | str += clen; |
194 | } |
237 | } |
|
|
238 | else if (enc->json.flags & F_LATIN1) |
|
|
239 | { |
|
|
240 | *enc->cur++ = uch; |
|
|
241 | str += clen; |
|
|
242 | } |
195 | else if (is_utf8) |
243 | else if (is_utf8) |
196 | { |
244 | { |
197 | need (enc, len += clen); |
245 | need (enc, len += clen); |
198 | do |
246 | do |
199 | { |
247 | { |
… | |
… | |
201 | } |
249 | } |
202 | while (--clen); |
250 | while (--clen); |
203 | } |
251 | } |
204 | else |
252 | else |
205 | { |
253 | { |
206 | need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed |
254 | need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
207 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
255 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
208 | ++str; |
256 | ++str; |
209 | } |
257 | } |
210 | } |
258 | } |
211 | } |
259 | } |
… | |
… | |
213 | |
261 | |
214 | --len; |
262 | --len; |
215 | } |
263 | } |
216 | } |
264 | } |
217 | |
265 | |
218 | static void |
266 | inline void |
219 | encode_indent (enc_t *enc) |
267 | encode_indent (enc_t *enc) |
220 | { |
268 | { |
221 | if (enc->flags & F_INDENT) |
269 | if (enc->json.flags & F_INDENT) |
222 | { |
270 | { |
223 | int spaces = enc->indent * INDENT_STEP; |
271 | int spaces = enc->indent * INDENT_STEP; |
224 | |
272 | |
225 | need (enc, spaces); |
273 | need (enc, spaces); |
226 | memset (enc->cur, ' ', spaces); |
274 | memset (enc->cur, ' ', spaces); |
227 | enc->cur += spaces; |
275 | enc->cur += spaces; |
228 | } |
276 | } |
229 | } |
277 | } |
230 | |
278 | |
231 | static void |
279 | inline void |
232 | encode_space (enc_t *enc) |
280 | encode_space (enc_t *enc) |
233 | { |
281 | { |
234 | need (enc, 1); |
282 | need (enc, 1); |
235 | encode_ch (enc, ' '); |
283 | encode_ch (enc, ' '); |
236 | } |
284 | } |
237 | |
285 | |
238 | static void |
286 | inline void |
239 | encode_nl (enc_t *enc) |
287 | encode_nl (enc_t *enc) |
240 | { |
288 | { |
241 | if (enc->flags & F_INDENT) |
289 | if (enc->json.flags & F_INDENT) |
242 | { |
290 | { |
243 | need (enc, 1); |
291 | need (enc, 1); |
244 | encode_ch (enc, '\n'); |
292 | encode_ch (enc, '\n'); |
245 | } |
293 | } |
246 | } |
294 | } |
247 | |
295 | |
248 | static void |
296 | inline void |
249 | encode_comma (enc_t *enc) |
297 | encode_comma (enc_t *enc) |
250 | { |
298 | { |
251 | encode_ch (enc, ','); |
299 | encode_ch (enc, ','); |
252 | |
300 | |
253 | if (enc->flags & F_INDENT) |
301 | if (enc->json.flags & F_INDENT) |
254 | encode_nl (enc); |
302 | encode_nl (enc); |
255 | else if (enc->flags & F_SPACE_AFTER) |
303 | else if (enc->json.flags & F_SPACE_AFTER) |
256 | encode_space (enc); |
304 | encode_space (enc); |
257 | } |
305 | } |
258 | |
306 | |
259 | static void encode_sv (enc_t *enc, SV *sv); |
307 | static void encode_sv (enc_t *enc, SV *sv); |
260 | |
308 | |
261 | static void |
309 | static void |
262 | encode_av (enc_t *enc, AV *av) |
310 | encode_av (enc_t *enc, AV *av) |
263 | { |
311 | { |
264 | int i, len = av_len (av); |
312 | int i, len = av_len (av); |
265 | |
313 | |
266 | encode_ch (enc, '['); encode_nl (enc); |
314 | if (enc->indent >= enc->maxdepth) |
267 | ++enc->indent; |
315 | croak ("data structure too deep (hit recursion limit)"); |
268 | |
316 | |
|
|
317 | encode_ch (enc, '['); |
|
|
318 | |
|
|
319 | if (len >= 0) |
|
|
320 | { |
|
|
321 | encode_nl (enc); ++enc->indent; |
|
|
322 | |
269 | for (i = 0; i <= len; ++i) |
323 | for (i = 0; i <= len; ++i) |
270 | { |
324 | { |
|
|
325 | SV **svp = av_fetch (av, i, 0); |
|
|
326 | |
271 | encode_indent (enc); |
327 | encode_indent (enc); |
272 | encode_sv (enc, *av_fetch (av, i, 0)); |
|
|
273 | |
328 | |
|
|
329 | if (svp) |
|
|
330 | encode_sv (enc, *svp); |
|
|
331 | else |
|
|
332 | encode_str (enc, "null", 4, 0); |
|
|
333 | |
274 | if (i < len) |
334 | if (i < len) |
275 | encode_comma (enc); |
335 | encode_comma (enc); |
276 | } |
336 | } |
277 | |
337 | |
|
|
338 | encode_nl (enc); --enc->indent; encode_indent (enc); |
|
|
339 | } |
|
|
340 | |
278 | encode_nl (enc); |
341 | encode_ch (enc, ']'); |
279 | |
|
|
280 | --enc->indent; |
|
|
281 | encode_indent (enc); encode_ch (enc, ']'); |
|
|
282 | } |
342 | } |
283 | |
343 | |
284 | static void |
344 | static void |
285 | encode_he (enc_t *enc, HE *he) |
345 | encode_hk (enc_t *enc, HE *he) |
286 | { |
346 | { |
287 | encode_ch (enc, '"'); |
347 | encode_ch (enc, '"'); |
288 | |
348 | |
289 | if (HeKLEN (he) == HEf_SVKEY) |
349 | if (HeKLEN (he) == HEf_SVKEY) |
290 | { |
350 | { |
… | |
… | |
300 | else |
360 | else |
301 | encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
361 | encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
302 | |
362 | |
303 | encode_ch (enc, '"'); |
363 | encode_ch (enc, '"'); |
304 | |
364 | |
305 | if (enc->flags & F_SPACE_BEFORE) encode_space (enc); |
365 | if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); |
306 | encode_ch (enc, ':'); |
366 | encode_ch (enc, ':'); |
307 | if (enc->flags & F_SPACE_AFTER ) encode_space (enc); |
367 | if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); |
308 | encode_sv (enc, HeVAL (he)); |
|
|
309 | } |
368 | } |
310 | |
369 | |
311 | // compare hash entries, used when all keys are bytestrings |
370 | // compare hash entries, used when all keys are bytestrings |
312 | static int |
371 | static int |
313 | he_cmp_fast (const void *a_, const void *b_) |
372 | he_cmp_fast (const void *a_, const void *b_) |
… | |
… | |
318 | HE *b = *(HE **)b_; |
377 | HE *b = *(HE **)b_; |
319 | |
378 | |
320 | STRLEN la = HeKLEN (a); |
379 | STRLEN la = HeKLEN (a); |
321 | STRLEN lb = HeKLEN (b); |
380 | STRLEN lb = HeKLEN (b); |
322 | |
381 | |
323 | if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) |
382 | if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la))) |
324 | cmp = la - lb; |
383 | cmp = lb - la; |
325 | |
384 | |
326 | return cmp; |
385 | return cmp; |
327 | } |
386 | } |
328 | |
387 | |
329 | // compare hash entries, used when some keys are sv's or utf-x |
388 | // compare hash entries, used when some keys are sv's or utf-x |
330 | static int |
389 | static int |
331 | he_cmp_slow (const void *a, const void *b) |
390 | he_cmp_slow (const void *a, const void *b) |
332 | { |
391 | { |
333 | return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); |
392 | return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a)); |
334 | } |
393 | } |
335 | |
394 | |
336 | static void |
395 | static void |
337 | encode_hv (enc_t *enc, HV *hv) |
396 | encode_hv (enc_t *enc, HV *hv) |
338 | { |
397 | { |
|
|
398 | HE *he; |
339 | int count, i; |
399 | int count; |
340 | |
400 | |
341 | encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; |
401 | if (enc->indent >= enc->maxdepth) |
|
|
402 | croak ("data structure too deep (hit recursion limit)"); |
342 | |
403 | |
343 | if ((count = hv_iterinit (hv))) |
404 | encode_ch (enc, '{'); |
344 | { |
405 | |
345 | // for canonical output we have to sort by keys first |
406 | // for canonical output we have to sort by keys first |
346 | // actually, this is mostly due to the stupid so-called |
407 | // actually, this is mostly due to the stupid so-called |
347 | // security workaround added somewhere in 5.8.x. |
408 | // security workaround added somewhere in 5.8.x. |
348 | // that randomises hash orderings |
409 | // that randomises hash orderings |
349 | if (enc->flags & F_CANONICAL) |
410 | if (enc->json.flags & F_CANONICAL) |
|
|
411 | { |
|
|
412 | int count = hv_iterinit (hv); |
|
|
413 | |
|
|
414 | if (SvMAGICAL (hv)) |
350 | { |
415 | { |
|
|
416 | // need to count by iterating. could improve by dynamically building the vector below |
|
|
417 | // but I don't care for the speed of this special case. |
|
|
418 | // note also that we will run into undefined behaviour when the two iterations |
|
|
419 | // do not result in the same count, something I might care for in some later release. |
|
|
420 | |
|
|
421 | count = 0; |
|
|
422 | while (hv_iternext (hv)) |
|
|
423 | ++count; |
|
|
424 | |
|
|
425 | hv_iterinit (hv); |
|
|
426 | } |
|
|
427 | |
|
|
428 | if (count) |
|
|
429 | { |
|
|
430 | int i, fast = 1; |
|
|
431 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
|
|
432 | HE **hes = _alloca (count * sizeof (HE)); |
|
|
433 | #else |
351 | HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode |
434 | HE *hes [count]; // if your compiler dies here, you need to enable C99 mode |
352 | int fast = 1; |
435 | #endif |
353 | |
436 | |
354 | i = 0; |
437 | i = 0; |
355 | while ((he = hv_iternext (hv))) |
438 | while ((he = hv_iternext (hv))) |
356 | { |
439 | { |
357 | hes [i++] = he; |
440 | hes [i++] = he; |
… | |
… | |
379 | |
462 | |
380 | FREETMPS; |
463 | FREETMPS; |
381 | LEAVE; |
464 | LEAVE; |
382 | } |
465 | } |
383 | |
466 | |
384 | for (i = 0; i < count; ++i) |
467 | encode_nl (enc); ++enc->indent; |
|
|
468 | |
|
|
469 | while (count--) |
385 | { |
470 | { |
386 | encode_indent (enc); |
471 | encode_indent (enc); |
|
|
472 | he = hes [count]; |
387 | encode_he (enc, hes [i]); |
473 | encode_hk (enc, he); |
|
|
474 | encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
388 | |
475 | |
389 | if (i < count - 1) |
476 | if (count) |
390 | encode_comma (enc); |
477 | encode_comma (enc); |
391 | } |
478 | } |
392 | |
479 | |
|
|
480 | encode_nl (enc); --enc->indent; encode_indent (enc); |
|
|
481 | } |
|
|
482 | } |
|
|
483 | else |
|
|
484 | { |
|
|
485 | if (hv_iterinit (hv) || SvMAGICAL (hv)) |
|
|
486 | if ((he = hv_iternext (hv))) |
|
|
487 | { |
|
|
488 | encode_nl (enc); ++enc->indent; |
|
|
489 | |
|
|
490 | for (;;) |
|
|
491 | { |
393 | encode_nl (enc); |
492 | encode_indent (enc); |
|
|
493 | encode_hk (enc, he); |
|
|
494 | encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
|
|
495 | |
|
|
496 | if (!(he = hv_iternext (hv))) |
|
|
497 | break; |
|
|
498 | |
|
|
499 | encode_comma (enc); |
|
|
500 | } |
|
|
501 | |
|
|
502 | encode_nl (enc); --enc->indent; encode_indent (enc); |
|
|
503 | } |
|
|
504 | } |
|
|
505 | |
|
|
506 | encode_ch (enc, '}'); |
|
|
507 | } |
|
|
508 | |
|
|
509 | // encode objects, arrays and special \0=false and \1=true values. |
|
|
510 | static void |
|
|
511 | encode_rv (enc_t *enc, SV *sv) |
|
|
512 | { |
|
|
513 | svtype svt; |
|
|
514 | |
|
|
515 | SvGETMAGIC (sv); |
|
|
516 | svt = SvTYPE (sv); |
|
|
517 | |
|
|
518 | if (expect_false (SvOBJECT (sv))) |
|
|
519 | { |
|
|
520 | HV *stash = !JSON_SLOW || json_boolean_stash |
|
|
521 | ? json_boolean_stash |
|
|
522 | : gv_stashpv ("JSON::XS::Boolean", 1); |
|
|
523 | |
|
|
524 | if (SvSTASH (sv) == stash) |
|
|
525 | { |
|
|
526 | if (SvIV (sv)) |
|
|
527 | encode_str (enc, "true", 4, 0); |
|
|
528 | else |
|
|
529 | encode_str (enc, "false", 5, 0); |
394 | } |
530 | } |
395 | else |
531 | else |
396 | { |
532 | { |
397 | SV *sv; |
533 | #if 0 |
398 | HE *he = hv_iternext (hv); |
534 | if (0 && sv_derived_from (rv, "JSON::Literal")) |
399 | |
|
|
400 | for (;;) |
|
|
401 | { |
535 | { |
402 | encode_indent (enc); |
536 | // not yet |
403 | encode_he (enc, he); |
|
|
404 | |
|
|
405 | if (!(he = hv_iternext (hv))) |
|
|
406 | break; |
|
|
407 | |
|
|
408 | encode_comma (enc); |
|
|
409 | } |
537 | } |
|
|
538 | #endif |
|
|
539 | if (enc->json.flags & F_CONV_BLESSED) |
|
|
540 | { |
|
|
541 | // we re-bless the reference to get overload and other niceties right |
|
|
542 | GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0); |
410 | |
543 | |
|
|
544 | if (to_json) |
|
|
545 | { |
|
|
546 | dSP; |
|
|
547 | |
|
|
548 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
549 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); |
|
|
550 | |
|
|
551 | // calling with G_SCALAR ensures that we always get a 1 return value |
|
|
552 | PUTBACK; |
|
|
553 | call_sv ((SV *)GvCV (to_json), G_SCALAR); |
|
|
554 | SPAGAIN; |
|
|
555 | |
|
|
556 | // catch this surprisingly common error |
|
|
557 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
558 | croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
559 | |
|
|
560 | sv = POPs; |
|
|
561 | PUTBACK; |
|
|
562 | |
411 | encode_nl (enc); |
563 | encode_sv (enc, sv); |
|
|
564 | |
|
|
565 | FREETMPS; LEAVE; |
|
|
566 | } |
|
|
567 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
568 | encode_str (enc, "null", 4, 0); |
|
|
569 | else |
|
|
570 | croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it", |
|
|
571 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
572 | } |
|
|
573 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
574 | encode_str (enc, "null", 4, 0); |
|
|
575 | else |
|
|
576 | croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", |
|
|
577 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
412 | } |
578 | } |
413 | } |
579 | } |
|
|
580 | else if (svt == SVt_PVHV) |
|
|
581 | encode_hv (enc, (HV *)sv); |
|
|
582 | else if (svt == SVt_PVAV) |
|
|
583 | encode_av (enc, (AV *)sv); |
|
|
584 | else if (svt < SVt_PVAV) |
|
|
585 | { |
|
|
586 | STRLEN len = 0; |
|
|
587 | char *pv = svt ? SvPV (sv, len) : 0; |
414 | |
588 | |
415 | --enc->indent; encode_indent (enc); encode_ch (enc, '}'); |
589 | if (len == 1 && *pv == '1') |
|
|
590 | encode_str (enc, "true", 4, 0); |
|
|
591 | else if (len == 1 && *pv == '0') |
|
|
592 | encode_str (enc, "false", 5, 0); |
|
|
593 | else |
|
|
594 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
|
|
595 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
596 | } |
|
|
597 | else |
|
|
598 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
|
|
599 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
416 | } |
600 | } |
417 | |
601 | |
418 | static void |
602 | static void |
419 | encode_sv (enc_t *enc, SV *sv) |
603 | encode_sv (enc_t *enc, SV *sv) |
420 | { |
604 | { |
… | |
… | |
428 | encode_str (enc, str, len, SvUTF8 (sv)); |
612 | encode_str (enc, str, len, SvUTF8 (sv)); |
429 | encode_ch (enc, '"'); |
613 | encode_ch (enc, '"'); |
430 | } |
614 | } |
431 | else if (SvNOKp (sv)) |
615 | else if (SvNOKp (sv)) |
432 | { |
616 | { |
|
|
617 | // trust that perl will do the right thing w.r.t. JSON syntax. |
433 | need (enc, NV_DIG + 32); |
618 | need (enc, NV_DIG + 32); |
434 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
619 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
435 | enc->cur += strlen (enc->cur); |
620 | enc->cur += strlen (enc->cur); |
436 | } |
621 | } |
437 | else if (SvIOKp (sv)) |
622 | else if (SvIOKp (sv)) |
438 | { |
623 | { |
439 | need (enc, 64); |
624 | // we assume we can always read an IV as a UV |
|
|
625 | if (SvUV (sv) & ~(UV)0x7fff) |
|
|
626 | { |
|
|
627 | // large integer, use the (rather slow) snprintf way. |
|
|
628 | need (enc, sizeof (UV) * 3); |
440 | enc->cur += |
629 | enc->cur += |
441 | SvIsUV(sv) |
630 | SvIsUV(sv) |
442 | ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) |
631 | ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv)) |
443 | : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); |
632 | : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv)); |
|
|
633 | } |
|
|
634 | else |
|
|
635 | { |
|
|
636 | // optimise the "small number case" |
|
|
637 | // code will likely be branchless and use only a single multiplication |
|
|
638 | I32 i = SvIV (sv); |
|
|
639 | U32 u; |
|
|
640 | char digit, nz = 0; |
|
|
641 | |
|
|
642 | need (enc, 6); |
|
|
643 | |
|
|
644 | *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0; |
|
|
645 | u = i < 0 ? -i : i; |
|
|
646 | |
|
|
647 | // convert to 4.28 fixed-point representation |
|
|
648 | u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits |
|
|
649 | |
|
|
650 | // now output digit by digit, each time masking out the integer part |
|
|
651 | // and multiplying by 5 while moving the decimal point one to the right, |
|
|
652 | // resulting in a net multiplication by 10. |
|
|
653 | // we always write the digit to memory but conditionally increment |
|
|
654 | // the pointer, to ease the usage of conditional move instructions. |
|
|
655 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; |
|
|
656 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; |
|
|
657 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; |
|
|
658 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; |
|
|
659 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
|
|
660 | } |
444 | } |
661 | } |
445 | else if (SvROK (sv)) |
662 | else if (SvROK (sv)) |
446 | { |
663 | encode_rv (enc, SvRV (sv)); |
447 | SV *rv = SvRV (sv); |
|
|
448 | |
|
|
449 | if (enc->indent >= enc->max_depth) |
|
|
450 | croak ("data structure too deep (hit recursion limit)"); |
|
|
451 | |
|
|
452 | switch (SvTYPE (rv)) |
|
|
453 | { |
|
|
454 | case SVt_PVAV: encode_av (enc, (AV *)rv); break; |
|
|
455 | case SVt_PVHV: encode_hv (enc, (HV *)rv); break; |
|
|
456 | |
|
|
457 | default: |
|
|
458 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
|
|
459 | SvPV_nolen (sv)); |
|
|
460 | } |
|
|
461 | } |
|
|
462 | else if (!SvOK (sv)) |
664 | else if (!SvOK (sv)) |
463 | encode_str (enc, "null", 4, 0); |
665 | encode_str (enc, "null", 4, 0); |
464 | else |
666 | else |
465 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
667 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
466 | SvPV_nolen (sv), SvFLAGS (sv)); |
668 | SvPV_nolen (sv), SvFLAGS (sv)); |
467 | } |
669 | } |
468 | |
670 | |
469 | static SV * |
671 | static SV * |
470 | encode_json (SV *scalar, UV flags) |
672 | encode_json (SV *scalar, JSON *json) |
471 | { |
673 | { |
|
|
674 | enc_t enc; |
|
|
675 | |
472 | if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
676 | if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
473 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
677 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
474 | |
678 | |
475 | enc_t enc; |
679 | enc.json = *json; |
476 | enc.flags = flags; |
|
|
477 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
680 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
478 | enc.cur = SvPVX (enc.sv); |
681 | enc.cur = SvPVX (enc.sv); |
479 | enc.end = SvEND (enc.sv); |
682 | enc.end = SvEND (enc.sv); |
480 | enc.indent = 0; |
683 | enc.indent = 0; |
481 | enc.max_depth = 0x7fffffffUL; |
684 | enc.maxdepth = DEC_DEPTH (enc.json.flags); |
482 | |
685 | |
483 | SvPOK_only (enc.sv); |
686 | SvPOK_only (enc.sv); |
484 | encode_sv (&enc, scalar); |
687 | encode_sv (&enc, scalar); |
485 | |
688 | |
|
|
689 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
|
|
690 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
|
|
691 | |
486 | if (!(flags & (F_ASCII | F_UTF8))) |
692 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
487 | SvUTF8_on (enc.sv); |
693 | SvUTF8_on (enc.sv); |
488 | |
694 | |
489 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
|
|
490 | |
|
|
491 | if (enc.flags & F_SHRINK) |
695 | if (enc.json.flags & F_SHRINK) |
492 | shrink (enc.sv); |
696 | shrink (enc.sv); |
493 | |
697 | |
494 | return enc.sv; |
698 | return enc.sv; |
495 | } |
699 | } |
496 | |
700 | |
… | |
… | |
501 | typedef struct |
705 | typedef struct |
502 | { |
706 | { |
503 | char *cur; // current parser pointer |
707 | char *cur; // current parser pointer |
504 | char *end; // end of input string |
708 | char *end; // end of input string |
505 | const char *err; // parse error, if != 0 |
709 | const char *err; // parse error, if != 0 |
506 | UV flags; // F_* |
710 | JSON json; |
|
|
711 | U32 depth; // recursion depth |
|
|
712 | U32 maxdepth; // recursion depth limit |
507 | } dec_t; |
713 | } dec_t; |
508 | |
714 | |
509 | static void |
715 | inline void |
|
|
716 | decode_comment (dec_t *dec) |
|
|
717 | { |
|
|
718 | // only '#'-style comments allowed a.t.m. |
|
|
719 | |
|
|
720 | while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) |
|
|
721 | ++dec->cur; |
|
|
722 | } |
|
|
723 | |
|
|
724 | inline void |
510 | decode_ws (dec_t *dec) |
725 | decode_ws (dec_t *dec) |
511 | { |
726 | { |
512 | for (;;) |
727 | for (;;) |
513 | { |
728 | { |
514 | char ch = *dec->cur; |
729 | char ch = *dec->cur; |
515 | |
730 | |
516 | if (ch > 0x20 |
731 | if (ch > 0x20) |
|
|
732 | { |
|
|
733 | if (expect_false (ch == '#')) |
|
|
734 | { |
|
|
735 | if (dec->json.flags & F_RELAXED) |
|
|
736 | decode_comment (dec); |
|
|
737 | else |
|
|
738 | break; |
|
|
739 | } |
|
|
740 | else |
|
|
741 | break; |
|
|
742 | } |
517 | || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) |
743 | else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09) |
518 | break; |
744 | break; // parse error, but let higher level handle it, gives better error messages |
519 | |
745 | |
520 | ++dec->cur; |
746 | ++dec->cur; |
521 | } |
747 | } |
522 | } |
748 | } |
523 | |
749 | |
524 | #define ERR(reason) SB dec->err = reason; goto fail; SE |
750 | #define ERR(reason) SB dec->err = reason; goto fail; SE |
|
|
751 | |
525 | #define EXPECT_CH(ch) SB \ |
752 | #define EXPECT_CH(ch) SB \ |
526 | if (*dec->cur != ch) \ |
753 | if (*dec->cur != ch) \ |
527 | ERR (# ch " expected"); \ |
754 | ERR (# ch " expected"); \ |
528 | ++dec->cur; \ |
755 | ++dec->cur; \ |
529 | SE |
756 | SE |
530 | |
757 | |
|
|
758 | #define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") |
|
|
759 | #define DEC_DEC_DEPTH --dec->depth |
|
|
760 | |
531 | static SV *decode_sv (dec_t *dec); |
761 | static SV *decode_sv (dec_t *dec); |
532 | |
762 | |
533 | static signed char decode_hexdigit[256]; |
763 | static signed char decode_hexdigit[256]; |
534 | |
764 | |
535 | static UV |
765 | static UV |
536 | decode_4hex (dec_t *dec) |
766 | decode_4hex (dec_t *dec) |
537 | { |
767 | { |
538 | signed char d1, d2, d3, d4; |
768 | signed char d1, d2, d3, d4; |
539 | unsigned char *cur = (unsigned char *)dec->cur; |
769 | unsigned char *cur = (unsigned char *)dec->cur; |
540 | |
770 | |
541 | d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); |
771 | d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected"); |
542 | d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); |
772 | d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected"); |
543 | d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); |
773 | d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected"); |
544 | d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); |
774 | d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected"); |
545 | |
775 | |
546 | dec->cur += 4; |
776 | dec->cur += 4; |
547 | |
777 | |
548 | return ((UV)d1) << 12 |
778 | return ((UV)d1) << 12 |
549 | | ((UV)d2) << 8 |
779 | | ((UV)d2) << 8 |
… | |
… | |
557 | static SV * |
787 | static SV * |
558 | decode_str (dec_t *dec) |
788 | decode_str (dec_t *dec) |
559 | { |
789 | { |
560 | SV *sv = 0; |
790 | SV *sv = 0; |
561 | int utf8 = 0; |
791 | int utf8 = 0; |
|
|
792 | char *dec_cur = dec->cur; |
562 | |
793 | |
563 | do |
794 | do |
564 | { |
795 | { |
565 | char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; |
796 | char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; |
566 | char *cur = buf; |
797 | char *cur = buf; |
567 | |
798 | |
568 | do |
799 | do |
569 | { |
800 | { |
570 | unsigned char ch = *(unsigned char *)dec->cur++; |
801 | unsigned char ch = *(unsigned char *)dec_cur++; |
571 | |
802 | |
572 | if (ch == '"') |
803 | if (expect_false (ch == '"')) |
573 | { |
804 | { |
574 | --dec->cur; |
805 | --dec_cur; |
575 | break; |
806 | break; |
576 | } |
807 | } |
577 | else if (ch == '\\') |
808 | else if (expect_false (ch == '\\')) |
578 | { |
809 | { |
579 | switch (*dec->cur) |
810 | switch (*dec_cur) |
580 | { |
811 | { |
581 | case '\\': |
812 | case '\\': |
582 | case '/': |
813 | case '/': |
583 | case '"': *cur++ = *dec->cur++; break; |
814 | case '"': *cur++ = *dec_cur++; break; |
584 | |
815 | |
585 | case 'b': ++dec->cur; *cur++ = '\010'; break; |
816 | case 'b': ++dec_cur; *cur++ = '\010'; break; |
586 | case 't': ++dec->cur; *cur++ = '\011'; break; |
817 | case 't': ++dec_cur; *cur++ = '\011'; break; |
587 | case 'n': ++dec->cur; *cur++ = '\012'; break; |
818 | case 'n': ++dec_cur; *cur++ = '\012'; break; |
588 | case 'f': ++dec->cur; *cur++ = '\014'; break; |
819 | case 'f': ++dec_cur; *cur++ = '\014'; break; |
589 | case 'r': ++dec->cur; *cur++ = '\015'; break; |
820 | case 'r': ++dec_cur; *cur++ = '\015'; break; |
590 | |
821 | |
591 | case 'u': |
822 | case 'u': |
592 | { |
823 | { |
593 | UV lo, hi; |
824 | UV lo, hi; |
594 | ++dec->cur; |
825 | ++dec_cur; |
595 | |
826 | |
|
|
827 | dec->cur = dec_cur; |
596 | hi = decode_4hex (dec); |
828 | hi = decode_4hex (dec); |
|
|
829 | dec_cur = dec->cur; |
597 | if (hi == (UV)-1) |
830 | if (hi == (UV)-1) |
598 | goto fail; |
831 | goto fail; |
599 | |
832 | |
600 | // possibly a surrogate pair |
833 | // possibly a surrogate pair |
601 | if (hi >= 0xd800) |
834 | if (hi >= 0xd800) |
602 | if (hi < 0xdc00) |
835 | if (hi < 0xdc00) |
603 | { |
836 | { |
604 | if (dec->cur [0] != '\\' || dec->cur [1] != 'u') |
837 | if (dec_cur [0] != '\\' || dec_cur [1] != 'u') |
605 | ERR ("missing low surrogate character in surrogate pair"); |
838 | ERR ("missing low surrogate character in surrogate pair"); |
606 | |
839 | |
607 | dec->cur += 2; |
840 | dec_cur += 2; |
608 | |
841 | |
|
|
842 | dec->cur = dec_cur; |
609 | lo = decode_4hex (dec); |
843 | lo = decode_4hex (dec); |
|
|
844 | dec_cur = dec->cur; |
610 | if (lo == (UV)-1) |
845 | if (lo == (UV)-1) |
611 | goto fail; |
846 | goto fail; |
612 | |
847 | |
613 | if (lo < 0xdc00 || lo >= 0xe000) |
848 | if (lo < 0xdc00 || lo >= 0xe000) |
614 | ERR ("surrogate pair expected"); |
849 | ERR ("surrogate pair expected"); |
… | |
… | |
628 | *cur++ = hi; |
863 | *cur++ = hi; |
629 | } |
864 | } |
630 | break; |
865 | break; |
631 | |
866 | |
632 | default: |
867 | default: |
633 | --dec->cur; |
868 | --dec_cur; |
634 | ERR ("illegal backslash escape sequence in string"); |
869 | ERR ("illegal backslash escape sequence in string"); |
635 | } |
870 | } |
636 | } |
871 | } |
637 | else if (ch >= 0x20 && ch <= 0x7f) |
872 | else if (expect_true (ch >= 0x20 && ch <= 0x7f)) |
638 | *cur++ = ch; |
873 | *cur++ = ch; |
639 | else if (ch >= 0x80) |
874 | else if (ch >= 0x80) |
640 | { |
875 | { |
641 | --dec->cur; |
|
|
642 | |
|
|
643 | STRLEN clen; |
876 | STRLEN clen; |
|
|
877 | UV uch; |
|
|
878 | |
|
|
879 | --dec_cur; |
|
|
880 | |
644 | UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); |
881 | uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
645 | if (clen == (STRLEN)-1) |
882 | if (clen == (STRLEN)-1) |
646 | ERR ("malformed UTF-8 character in JSON string"); |
883 | ERR ("malformed UTF-8 character in JSON string"); |
647 | |
884 | |
648 | do |
885 | do |
649 | { |
|
|
650 | *cur++ = *dec->cur++; |
886 | *cur++ = *dec_cur++; |
651 | } |
|
|
652 | while (--clen); |
887 | while (--clen); |
653 | |
888 | |
654 | utf8 = 1; |
889 | utf8 = 1; |
655 | } |
890 | } |
656 | else if (!ch) |
|
|
657 | ERR ("unexpected end of string while parsing json string"); |
|
|
658 | else |
891 | else |
|
|
892 | { |
|
|
893 | --dec_cur; |
|
|
894 | |
|
|
895 | if (!ch) |
|
|
896 | ERR ("unexpected end of string while parsing JSON string"); |
|
|
897 | else |
659 | ERR ("invalid character encountered"); |
898 | ERR ("invalid character encountered while parsing JSON string"); |
660 | |
899 | } |
661 | } |
900 | } |
662 | while (cur < buf + SHORT_STRING_LEN); |
901 | while (cur < buf + SHORT_STRING_LEN); |
663 | |
902 | |
|
|
903 | { |
664 | STRLEN len = cur - buf; |
904 | STRLEN len = cur - buf; |
665 | |
905 | |
666 | if (sv) |
906 | if (sv) |
667 | { |
907 | { |
668 | SvGROW (sv, SvCUR (sv) + len + 1); |
908 | SvGROW (sv, SvCUR (sv) + len + 1); |
669 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
909 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
670 | SvCUR_set (sv, SvCUR (sv) + len); |
910 | SvCUR_set (sv, SvCUR (sv) + len); |
671 | } |
911 | } |
672 | else |
912 | else |
673 | sv = newSVpvn (buf, len); |
913 | sv = newSVpvn (buf, len); |
674 | } |
914 | } |
|
|
915 | } |
675 | while (*dec->cur != '"'); |
916 | while (*dec_cur != '"'); |
676 | |
917 | |
677 | ++dec->cur; |
918 | ++dec_cur; |
678 | |
919 | |
679 | if (sv) |
920 | if (sv) |
680 | { |
921 | { |
681 | SvPOK_only (sv); |
922 | SvPOK_only (sv); |
682 | *SvEND (sv) = 0; |
923 | *SvEND (sv) = 0; |
… | |
… | |
685 | SvUTF8_on (sv); |
926 | SvUTF8_on (sv); |
686 | } |
927 | } |
687 | else |
928 | else |
688 | sv = newSVpvn ("", 0); |
929 | sv = newSVpvn ("", 0); |
689 | |
930 | |
|
|
931 | dec->cur = dec_cur; |
690 | return sv; |
932 | return sv; |
691 | |
933 | |
692 | fail: |
934 | fail: |
|
|
935 | dec->cur = dec_cur; |
693 | return 0; |
936 | return 0; |
694 | } |
937 | } |
695 | |
938 | |
696 | static SV * |
939 | static SV * |
697 | decode_num (dec_t *dec) |
940 | decode_num (dec_t *dec) |
… | |
… | |
755 | is_nv = 1; |
998 | is_nv = 1; |
756 | } |
999 | } |
757 | |
1000 | |
758 | if (!is_nv) |
1001 | if (!is_nv) |
759 | { |
1002 | { |
760 | UV uv; |
1003 | int len = dec->cur - start; |
761 | int numtype = grok_number (start, dec->cur - start, &uv); |
1004 | |
762 | if (numtype & IS_NUMBER_IN_UV) |
1005 | // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so |
763 | if (numtype & IS_NUMBER_NEG) |
1006 | if (*start == '-') |
|
|
1007 | switch (len) |
764 | { |
1008 | { |
765 | if (uv < (UV)IV_MIN) |
1009 | case 2: return newSViv (-( start [1] - '0' * 1)); |
766 | return newSViv (-(IV)uv); |
1010 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
|
|
1011 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
|
|
1012 | case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
767 | } |
1013 | } |
|
|
1014 | else |
|
|
1015 | switch (len) |
|
|
1016 | { |
|
|
1017 | case 1: return newSViv ( start [0] - '0' * 1); |
|
|
1018 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
|
|
1019 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
|
|
1020 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
|
|
1021 | } |
|
|
1022 | |
|
|
1023 | { |
|
|
1024 | UV uv; |
|
|
1025 | int numtype = grok_number (start, len, &uv); |
|
|
1026 | if (numtype & IS_NUMBER_IN_UV) |
|
|
1027 | if (numtype & IS_NUMBER_NEG) |
|
|
1028 | { |
|
|
1029 | if (uv < (UV)IV_MIN) |
|
|
1030 | return newSViv (-(IV)uv); |
|
|
1031 | } |
768 | else |
1032 | else |
769 | return newSVuv (uv); |
1033 | return newSVuv (uv); |
770 | } |
1034 | } |
771 | |
1035 | |
|
|
1036 | len -= *start == '-' ? 1 : 0; |
|
|
1037 | |
|
|
1038 | // does not fit into IV or UV, try NV |
|
|
1039 | if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) |
|
|
1040 | #if defined (LDBL_DIG) |
|
|
1041 | || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len) |
|
|
1042 | #endif |
|
|
1043 | ) |
|
|
1044 | // fits into NV without loss of precision |
|
|
1045 | return newSVnv (Atof (start)); |
|
|
1046 | |
|
|
1047 | // everything else fails, convert it to a string |
|
|
1048 | return newSVpvn (start, dec->cur - start); |
|
|
1049 | } |
|
|
1050 | |
|
|
1051 | // loss of precision here |
772 | return newSVnv (Atof (start)); |
1052 | return newSVnv (Atof (start)); |
773 | |
1053 | |
774 | fail: |
1054 | fail: |
775 | return 0; |
1055 | return 0; |
776 | } |
1056 | } |
… | |
… | |
778 | static SV * |
1058 | static SV * |
779 | decode_av (dec_t *dec) |
1059 | decode_av (dec_t *dec) |
780 | { |
1060 | { |
781 | AV *av = newAV (); |
1061 | AV *av = newAV (); |
782 | |
1062 | |
|
|
1063 | DEC_INC_DEPTH; |
783 | decode_ws (dec); |
1064 | decode_ws (dec); |
|
|
1065 | |
784 | if (*dec->cur == ']') |
1066 | if (*dec->cur == ']') |
785 | ++dec->cur; |
1067 | ++dec->cur; |
786 | else |
1068 | else |
787 | for (;;) |
1069 | for (;;) |
788 | { |
1070 | { |
… | |
… | |
804 | |
1086 | |
805 | if (*dec->cur != ',') |
1087 | if (*dec->cur != ',') |
806 | ERR (", or ] expected while parsing array"); |
1088 | ERR (", or ] expected while parsing array"); |
807 | |
1089 | |
808 | ++dec->cur; |
1090 | ++dec->cur; |
|
|
1091 | |
|
|
1092 | decode_ws (dec); |
|
|
1093 | |
|
|
1094 | if (*dec->cur == ']' && dec->json.flags & F_RELAXED) |
|
|
1095 | { |
|
|
1096 | ++dec->cur; |
|
|
1097 | break; |
|
|
1098 | } |
809 | } |
1099 | } |
810 | |
1100 | |
|
|
1101 | DEC_DEC_DEPTH; |
811 | return newRV_noinc ((SV *)av); |
1102 | return newRV_noinc ((SV *)av); |
812 | |
1103 | |
813 | fail: |
1104 | fail: |
814 | SvREFCNT_dec (av); |
1105 | SvREFCNT_dec (av); |
|
|
1106 | DEC_DEC_DEPTH; |
815 | return 0; |
1107 | return 0; |
816 | } |
1108 | } |
817 | |
1109 | |
818 | static SV * |
1110 | static SV * |
819 | decode_hv (dec_t *dec) |
1111 | decode_hv (dec_t *dec) |
820 | { |
1112 | { |
|
|
1113 | SV *sv; |
821 | HV *hv = newHV (); |
1114 | HV *hv = newHV (); |
822 | |
1115 | |
|
|
1116 | DEC_INC_DEPTH; |
823 | decode_ws (dec); |
1117 | decode_ws (dec); |
|
|
1118 | |
824 | if (*dec->cur == '}') |
1119 | if (*dec->cur == '}') |
825 | ++dec->cur; |
1120 | ++dec->cur; |
826 | else |
1121 | else |
827 | for (;;) |
1122 | for (;;) |
828 | { |
1123 | { |
|
|
1124 | EXPECT_CH ('"'); |
|
|
1125 | |
|
|
1126 | // heuristic: assume that |
|
|
1127 | // a) decode_str + hv_store_ent are abysmally slow. |
|
|
1128 | // b) most hash keys are short, simple ascii text. |
|
|
1129 | // => try to "fast-match" such strings to avoid |
|
|
1130 | // the overhead of decode_str + hv_store_ent. |
|
|
1131 | { |
829 | SV *key, *value; |
1132 | SV *value; |
|
|
1133 | char *p = dec->cur; |
|
|
1134 | char *e = p + 24; // only try up to 24 bytes |
830 | |
1135 | |
831 | decode_ws (dec); EXPECT_CH ('"'); |
1136 | for (;;) |
832 | |
|
|
833 | key = decode_str (dec); |
|
|
834 | if (!key) |
|
|
835 | goto fail; |
|
|
836 | |
|
|
837 | decode_ws (dec); EXPECT_CH (':'); |
|
|
838 | |
|
|
839 | value = decode_sv (dec); |
|
|
840 | if (!value) |
|
|
841 | { |
1137 | { |
|
|
1138 | // the >= 0x80 is true on most architectures |
|
|
1139 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
|
|
1140 | { |
|
|
1141 | // slow path, back up and use decode_str |
|
|
1142 | SV *key = decode_str (dec); |
|
|
1143 | if (!key) |
|
|
1144 | goto fail; |
|
|
1145 | |
|
|
1146 | decode_ws (dec); EXPECT_CH (':'); |
|
|
1147 | |
|
|
1148 | decode_ws (dec); |
|
|
1149 | value = decode_sv (dec); |
|
|
1150 | if (!value) |
|
|
1151 | { |
|
|
1152 | SvREFCNT_dec (key); |
|
|
1153 | goto fail; |
|
|
1154 | } |
|
|
1155 | |
|
|
1156 | hv_store_ent (hv, key, value, 0); |
842 | SvREFCNT_dec (key); |
1157 | SvREFCNT_dec (key); |
|
|
1158 | |
|
|
1159 | break; |
|
|
1160 | } |
|
|
1161 | else if (*p == '"') |
|
|
1162 | { |
|
|
1163 | // fast path, got a simple key |
|
|
1164 | char *key = dec->cur; |
|
|
1165 | int len = p - key; |
|
|
1166 | dec->cur = p + 1; |
|
|
1167 | |
|
|
1168 | decode_ws (dec); EXPECT_CH (':'); |
|
|
1169 | |
|
|
1170 | decode_ws (dec); |
|
|
1171 | value = decode_sv (dec); |
|
|
1172 | if (!value) |
843 | goto fail; |
1173 | goto fail; |
|
|
1174 | |
|
|
1175 | hv_store (hv, key, len, value, 0); |
|
|
1176 | |
|
|
1177 | break; |
|
|
1178 | } |
|
|
1179 | |
|
|
1180 | ++p; |
844 | } |
1181 | } |
845 | |
1182 | } |
846 | //TODO: optimise |
|
|
847 | hv_store_ent (hv, key, value, 0); |
|
|
848 | |
1183 | |
849 | decode_ws (dec); |
1184 | decode_ws (dec); |
850 | |
1185 | |
851 | if (*dec->cur == '}') |
1186 | if (*dec->cur == '}') |
852 | { |
1187 | { |
… | |
… | |
856 | |
1191 | |
857 | if (*dec->cur != ',') |
1192 | if (*dec->cur != ',') |
858 | ERR (", or } expected while parsing object/hash"); |
1193 | ERR (", or } expected while parsing object/hash"); |
859 | |
1194 | |
860 | ++dec->cur; |
1195 | ++dec->cur; |
|
|
1196 | |
|
|
1197 | decode_ws (dec); |
|
|
1198 | |
|
|
1199 | if (*dec->cur == '}' && dec->json.flags & F_RELAXED) |
|
|
1200 | { |
|
|
1201 | ++dec->cur; |
|
|
1202 | break; |
|
|
1203 | } |
861 | } |
1204 | } |
862 | |
1205 | |
|
|
1206 | DEC_DEC_DEPTH; |
863 | return newRV_noinc ((SV *)hv); |
1207 | sv = newRV_noinc ((SV *)hv); |
|
|
1208 | |
|
|
1209 | // check filter callbacks |
|
|
1210 | if (dec->json.flags & F_HOOK) |
|
|
1211 | { |
|
|
1212 | if (dec->json.cb_sk_object && HvKEYS (hv) == 1) |
|
|
1213 | { |
|
|
1214 | HE *cb, *he; |
|
|
1215 | |
|
|
1216 | hv_iterinit (hv); |
|
|
1217 | he = hv_iternext (hv); |
|
|
1218 | hv_iterinit (hv); |
|
|
1219 | |
|
|
1220 | // the next line creates a mortal sv each time its called. |
|
|
1221 | // might want to optimise this for common cases. |
|
|
1222 | cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); |
|
|
1223 | |
|
|
1224 | if (cb) |
|
|
1225 | { |
|
|
1226 | dSP; |
|
|
1227 | int count; |
|
|
1228 | |
|
|
1229 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
1230 | XPUSHs (HeVAL (he)); |
|
|
1231 | |
|
|
1232 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
|
|
1233 | |
|
|
1234 | if (count == 1) |
|
|
1235 | { |
|
|
1236 | sv = newSVsv (POPs); |
|
|
1237 | FREETMPS; LEAVE; |
|
|
1238 | return sv; |
|
|
1239 | } |
|
|
1240 | |
|
|
1241 | FREETMPS; LEAVE; |
|
|
1242 | } |
|
|
1243 | } |
|
|
1244 | |
|
|
1245 | if (dec->json.cb_object) |
|
|
1246 | { |
|
|
1247 | dSP; |
|
|
1248 | int count; |
|
|
1249 | |
|
|
1250 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
1251 | XPUSHs (sv_2mortal (sv)); |
|
|
1252 | |
|
|
1253 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
|
|
1254 | |
|
|
1255 | if (count == 1) |
|
|
1256 | { |
|
|
1257 | sv = newSVsv (POPs); |
|
|
1258 | FREETMPS; LEAVE; |
|
|
1259 | return sv; |
|
|
1260 | } |
|
|
1261 | |
|
|
1262 | SvREFCNT_inc (sv); |
|
|
1263 | FREETMPS; LEAVE; |
|
|
1264 | } |
|
|
1265 | } |
|
|
1266 | |
|
|
1267 | return sv; |
864 | |
1268 | |
865 | fail: |
1269 | fail: |
866 | SvREFCNT_dec (hv); |
1270 | SvREFCNT_dec (hv); |
|
|
1271 | DEC_DEC_DEPTH; |
867 | return 0; |
1272 | return 0; |
868 | } |
1273 | } |
869 | |
1274 | |
870 | static SV * |
1275 | static SV * |
871 | decode_sv (dec_t *dec) |
1276 | decode_sv (dec_t *dec) |
872 | { |
1277 | { |
873 | decode_ws (dec); |
1278 | // the beauty of JSON: you need exactly one character lookahead |
|
|
1279 | // to parse anything. |
874 | switch (*dec->cur) |
1280 | switch (*dec->cur) |
875 | { |
1281 | { |
876 | case '"': ++dec->cur; return decode_str (dec); |
1282 | case '"': ++dec->cur; return decode_str (dec); |
877 | case '[': ++dec->cur; return decode_av (dec); |
1283 | case '[': ++dec->cur; return decode_av (dec); |
878 | case '{': ++dec->cur; return decode_hv (dec); |
1284 | case '{': ++dec->cur; return decode_hv (dec); |
… | |
… | |
884 | |
1290 | |
885 | case 't': |
1291 | case 't': |
886 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1292 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
887 | { |
1293 | { |
888 | dec->cur += 4; |
1294 | dec->cur += 4; |
889 | return newSViv (1); |
1295 | #if JSON_SLOW |
|
|
1296 | json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); |
|
|
1297 | #endif |
|
|
1298 | return SvREFCNT_inc (json_true); |
890 | } |
1299 | } |
891 | else |
1300 | else |
892 | ERR ("'true' expected"); |
1301 | ERR ("'true' expected"); |
893 | |
1302 | |
894 | break; |
1303 | break; |
895 | |
1304 | |
896 | case 'f': |
1305 | case 'f': |
897 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1306 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
898 | { |
1307 | { |
899 | dec->cur += 5; |
1308 | dec->cur += 5; |
900 | return newSViv (0); |
1309 | #if JSON_SLOW |
|
|
1310 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
|
|
1311 | #endif |
|
|
1312 | return SvREFCNT_inc (json_false); |
901 | } |
1313 | } |
902 | else |
1314 | else |
903 | ERR ("'false' expected"); |
1315 | ERR ("'false' expected"); |
904 | |
1316 | |
905 | break; |
1317 | break; |
… | |
… | |
914 | ERR ("'null' expected"); |
1326 | ERR ("'null' expected"); |
915 | |
1327 | |
916 | break; |
1328 | break; |
917 | |
1329 | |
918 | default: |
1330 | default: |
919 | ERR ("malformed json string, neither array, object, number, string or atom"); |
1331 | ERR ("malformed JSON string, neither array, object, number, string or atom"); |
920 | break; |
1332 | break; |
921 | } |
1333 | } |
922 | |
1334 | |
923 | fail: |
1335 | fail: |
924 | return 0; |
1336 | return 0; |
925 | } |
1337 | } |
926 | |
1338 | |
927 | static SV * |
1339 | static SV * |
928 | decode_json (SV *string, UV flags) |
1340 | decode_json (SV *string, JSON *json, UV *offset_return) |
929 | { |
1341 | { |
|
|
1342 | dec_t dec; |
|
|
1343 | UV offset; |
930 | SV *sv; |
1344 | SV *sv; |
931 | |
1345 | |
|
|
1346 | SvGETMAGIC (string); |
|
|
1347 | SvUPGRADE (string, SVt_PV); |
|
|
1348 | |
|
|
1349 | if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) |
|
|
1350 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
|
|
1351 | (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); |
|
|
1352 | |
932 | if (flags & F_UTF8) |
1353 | if (json->flags & F_UTF8) |
933 | sv_utf8_downgrade (string, 0); |
1354 | sv_utf8_downgrade (string, 0); |
934 | else |
1355 | else |
935 | sv_utf8_upgrade (string); |
1356 | sv_utf8_upgrade (string); |
936 | |
1357 | |
937 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1358 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
938 | |
1359 | |
939 | dec_t dec; |
1360 | dec.json = *json; |
940 | dec.flags = flags; |
|
|
941 | dec.cur = SvPVX (string); |
1361 | dec.cur = SvPVX (string); |
942 | dec.end = SvEND (string); |
1362 | dec.end = SvEND (string); |
943 | dec.err = 0; |
1363 | dec.err = 0; |
|
|
1364 | dec.depth = 0; |
|
|
1365 | dec.maxdepth = DEC_DEPTH (dec.json.flags); |
944 | |
1366 | |
|
|
1367 | if (dec.json.cb_object || dec.json.cb_sk_object) |
|
|
1368 | dec.json.flags |= F_HOOK; |
|
|
1369 | |
|
|
1370 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
|
|
1371 | |
|
|
1372 | decode_ws (&dec); |
945 | sv = decode_sv (&dec); |
1373 | sv = decode_sv (&dec); |
946 | |
1374 | |
|
|
1375 | if (!(offset_return || !sv)) |
|
|
1376 | { |
|
|
1377 | // check for trailing garbage |
|
|
1378 | decode_ws (&dec); |
|
|
1379 | |
|
|
1380 | if (*dec.cur) |
|
|
1381 | { |
|
|
1382 | dec.err = "garbage after JSON object"; |
|
|
1383 | SvREFCNT_dec (sv); |
|
|
1384 | sv = 0; |
|
|
1385 | } |
|
|
1386 | } |
|
|
1387 | |
|
|
1388 | if (offset_return || !sv) |
|
|
1389 | { |
|
|
1390 | offset = dec.json.flags & F_UTF8 |
|
|
1391 | ? dec.cur - SvPVX (string) |
|
|
1392 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
1393 | |
|
|
1394 | if (offset_return) |
|
|
1395 | *offset_return = offset; |
|
|
1396 | } |
|
|
1397 | |
947 | if (!sv) |
1398 | if (!sv) |
948 | { |
1399 | { |
949 | IV offset = dec.flags & F_UTF8 |
|
|
950 | ? dec.cur - SvPVX (string) |
|
|
951 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
952 | SV *uni = sv_newmortal (); |
1400 | SV *uni = sv_newmortal (); |
953 | |
1401 | |
954 | // horrible hack to silence warning inside pv_uni_display |
1402 | // horrible hack to silence warning inside pv_uni_display |
955 | COP cop = *PL_curcop; |
1403 | COP cop = *PL_curcop; |
956 | cop.cop_warnings = pWARN_NONE; |
1404 | cop.cop_warnings = pWARN_NONE; |
… | |
… | |
958 | SAVEVPTR (PL_curcop); |
1406 | SAVEVPTR (PL_curcop); |
959 | PL_curcop = &cop; |
1407 | PL_curcop = &cop; |
960 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1408 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
961 | LEAVE; |
1409 | LEAVE; |
962 | |
1410 | |
963 | croak ("%s, at character offset %d (%s)", |
1411 | croak ("%s, at character offset %d [\"%s\"]", |
964 | dec.err, |
1412 | dec.err, |
965 | (int)offset, |
1413 | (int)offset, |
966 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1414 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
967 | } |
1415 | } |
968 | |
1416 | |
969 | sv = sv_2mortal (sv); |
1417 | sv = sv_2mortal (sv); |
970 | |
1418 | |
971 | if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1419 | if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
972 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1420 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
973 | |
1421 | |
974 | return sv; |
1422 | return sv; |
975 | } |
1423 | } |
976 | |
1424 | |
… | |
… | |
981 | |
1429 | |
982 | BOOT: |
1430 | BOOT: |
983 | { |
1431 | { |
984 | int i; |
1432 | int i; |
985 | |
1433 | |
986 | memset (decode_hexdigit, 0xff, 256); |
|
|
987 | for (i = 10; i--; ) |
1434 | for (i = 0; i < 256; ++i) |
988 | decode_hexdigit ['0' + i] = i; |
1435 | decode_hexdigit [i] = |
|
|
1436 | i >= '0' && i <= '9' ? i - '0' |
|
|
1437 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
|
|
1438 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
|
|
1439 | : -1; |
989 | |
1440 | |
990 | for (i = 7; i--; ) |
|
|
991 | { |
|
|
992 | decode_hexdigit ['a' + i] = 10 + i; |
|
|
993 | decode_hexdigit ['A' + i] = 10 + i; |
|
|
994 | } |
|
|
995 | |
|
|
996 | json_stash = gv_stashpv ("JSON::XS", 1); |
1441 | json_stash = gv_stashpv ("JSON::XS" , 1); |
|
|
1442 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
|
|
1443 | |
|
|
1444 | json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); |
|
|
1445 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
997 | } |
1446 | } |
998 | |
1447 | |
999 | PROTOTYPES: DISABLE |
1448 | PROTOTYPES: DISABLE |
1000 | |
1449 | |
1001 | SV *new (char *dummy) |
1450 | void CLONE (...) |
1002 | CODE: |
1451 | CODE: |
1003 | RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); |
1452 | json_stash = 0; |
|
|
1453 | json_boolean_stash = 0; |
|
|
1454 | |
|
|
1455 | void new (char *klass) |
|
|
1456 | PPCODE: |
|
|
1457 | { |
|
|
1458 | SV *pv = NEWSV (0, sizeof (JSON)); |
|
|
1459 | SvPOK_only (pv); |
|
|
1460 | Zero (SvPVX (pv), 1, JSON); |
|
|
1461 | ((JSON *)SvPVX (pv))->flags = F_DEFAULT; |
|
|
1462 | XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH))); |
|
|
1463 | } |
|
|
1464 | |
|
|
1465 | void ascii (JSON *self, int enable = 1) |
|
|
1466 | ALIAS: |
|
|
1467 | ascii = F_ASCII |
|
|
1468 | latin1 = F_LATIN1 |
|
|
1469 | utf8 = F_UTF8 |
|
|
1470 | indent = F_INDENT |
|
|
1471 | canonical = F_CANONICAL |
|
|
1472 | space_before = F_SPACE_BEFORE |
|
|
1473 | space_after = F_SPACE_AFTER |
|
|
1474 | pretty = F_PRETTY |
|
|
1475 | allow_nonref = F_ALLOW_NONREF |
|
|
1476 | shrink = F_SHRINK |
|
|
1477 | allow_blessed = F_ALLOW_BLESSED |
|
|
1478 | convert_blessed = F_CONV_BLESSED |
|
|
1479 | relaxed = F_RELAXED |
|
|
1480 | PPCODE: |
|
|
1481 | { |
|
|
1482 | if (enable) |
|
|
1483 | self->flags |= ix; |
|
|
1484 | else |
|
|
1485 | self->flags &= ~ix; |
|
|
1486 | |
|
|
1487 | XPUSHs (ST (0)); |
|
|
1488 | } |
|
|
1489 | |
|
|
1490 | void get_ascii (JSON *self) |
|
|
1491 | ALIAS: |
|
|
1492 | get_ascii = F_ASCII |
|
|
1493 | get_latin1 = F_LATIN1 |
|
|
1494 | get_utf8 = F_UTF8 |
|
|
1495 | get_indent = F_INDENT |
|
|
1496 | get_canonical = F_CANONICAL |
|
|
1497 | get_space_before = F_SPACE_BEFORE |
|
|
1498 | get_space_after = F_SPACE_AFTER |
|
|
1499 | get_pretty = F_PRETTY |
|
|
1500 | get_allow_nonref = F_ALLOW_NONREF |
|
|
1501 | get_shrink = F_SHRINK |
|
|
1502 | get_allow_blessed = F_ALLOW_BLESSED |
|
|
1503 | get_convert_blessed = F_CONV_BLESSED |
|
|
1504 | get_relaxed = F_RELAXED |
|
|
1505 | PPCODE: |
|
|
1506 | XPUSHs (boolSV (self->flags & ix)); |
|
|
1507 | |
|
|
1508 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
|
|
1509 | PPCODE: |
|
|
1510 | { |
|
|
1511 | UV log2 = 0; |
|
|
1512 | |
|
|
1513 | if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
|
|
1514 | |
|
|
1515 | while ((1UL << log2) < max_depth) |
|
|
1516 | ++log2; |
|
|
1517 | |
|
|
1518 | self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
|
|
1519 | |
|
|
1520 | XPUSHs (ST (0)); |
|
|
1521 | } |
|
|
1522 | |
|
|
1523 | U32 get_max_depth (JSON *self) |
|
|
1524 | CODE: |
|
|
1525 | RETVAL = DEC_DEPTH (self->flags); |
1004 | OUTPUT: |
1526 | OUTPUT: |
1005 | RETVAL |
1527 | RETVAL |
1006 | |
1528 | |
1007 | SV *ascii (SV *self, int enable = 1) |
1529 | void max_size (JSON *self, UV max_size = 0) |
1008 | ALIAS: |
1530 | PPCODE: |
1009 | ascii = F_ASCII |
1531 | { |
1010 | utf8 = F_UTF8 |
1532 | UV log2 = 0; |
1011 | indent = F_INDENT |
1533 | |
1012 | canonical = F_CANONICAL |
1534 | if (max_size > 0x80000000UL) max_size = 0x80000000UL; |
1013 | space_before = F_SPACE_BEFORE |
1535 | if (max_size == 1) max_size = 2; |
1014 | space_after = F_SPACE_AFTER |
1536 | |
1015 | pretty = F_PRETTY |
1537 | while ((1UL << log2) < max_size) |
1016 | allow_nonref = F_ALLOW_NONREF |
1538 | ++log2; |
1017 | shrink = F_SHRINK |
1539 | |
|
|
1540 | self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); |
|
|
1541 | |
|
|
1542 | XPUSHs (ST (0)); |
|
|
1543 | } |
|
|
1544 | |
|
|
1545 | int get_max_size (JSON *self) |
1018 | CODE: |
1546 | CODE: |
1019 | { |
1547 | RETVAL = DEC_SIZE (self->flags); |
1020 | UV *uv = SvJSON (self); |
|
|
1021 | if (enable) |
|
|
1022 | *uv |= ix; |
|
|
1023 | else |
|
|
1024 | *uv &= ~ix; |
|
|
1025 | |
|
|
1026 | RETVAL = newSVsv (self); |
|
|
1027 | } |
|
|
1028 | OUTPUT: |
1548 | OUTPUT: |
1029 | RETVAL |
1549 | RETVAL |
1030 | |
1550 | |
1031 | void encode (SV *self, SV *scalar) |
1551 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1032 | PPCODE: |
1552 | PPCODE: |
1033 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
1553 | { |
|
|
1554 | SvREFCNT_dec (self->cb_object); |
|
|
1555 | self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; |
1034 | |
1556 | |
1035 | void decode (SV *self, SV *jsonstr) |
1557 | XPUSHs (ST (0)); |
|
|
1558 | } |
|
|
1559 | |
|
|
1560 | void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) |
1036 | PPCODE: |
1561 | PPCODE: |
|
|
1562 | { |
|
|
1563 | if (!self->cb_sk_object) |
|
|
1564 | self->cb_sk_object = newHV (); |
|
|
1565 | |
|
|
1566 | if (SvOK (cb)) |
|
|
1567 | hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
|
|
1568 | else |
|
|
1569 | { |
|
|
1570 | hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0); |
|
|
1571 | |
|
|
1572 | if (!HvKEYS (self->cb_sk_object)) |
|
|
1573 | { |
|
|
1574 | SvREFCNT_dec (self->cb_sk_object); |
|
|
1575 | self->cb_sk_object = 0; |
|
|
1576 | } |
|
|
1577 | } |
|
|
1578 | |
|
|
1579 | XPUSHs (ST (0)); |
|
|
1580 | } |
|
|
1581 | |
|
|
1582 | void encode (JSON *self, SV *scalar) |
|
|
1583 | PPCODE: |
|
|
1584 | XPUSHs (encode_json (scalar, self)); |
|
|
1585 | |
|
|
1586 | void decode (JSON *self, SV *jsonstr) |
|
|
1587 | PPCODE: |
1037 | XPUSHs (decode_json (jsonstr, *SvJSON (self))); |
1588 | XPUSHs (decode_json (jsonstr, self, 0)); |
|
|
1589 | |
|
|
1590 | void decode_prefix (JSON *self, SV *jsonstr) |
|
|
1591 | PPCODE: |
|
|
1592 | { |
|
|
1593 | UV offset; |
|
|
1594 | EXTEND (SP, 2); |
|
|
1595 | PUSHs (decode_json (jsonstr, self, &offset)); |
|
|
1596 | PUSHs (sv_2mortal (newSVuv (offset))); |
|
|
1597 | } |
|
|
1598 | |
|
|
1599 | void DESTROY (JSON *self) |
|
|
1600 | CODE: |
|
|
1601 | SvREFCNT_dec (self->cb_sk_object); |
|
|
1602 | SvREFCNT_dec (self->cb_object); |
1038 | |
1603 | |
1039 | PROTOTYPES: ENABLE |
1604 | PROTOTYPES: ENABLE |
1040 | |
1605 | |
1041 | void to_json (SV *scalar) |
1606 | void to_json (SV *scalar) |
1042 | PPCODE: |
1607 | PPCODE: |
|
|
1608 | { |
|
|
1609 | JSON json = { F_DEFAULT | F_UTF8 }; |
1043 | XPUSHs (encode_json (scalar, F_UTF8)); |
1610 | XPUSHs (encode_json (scalar, &json)); |
|
|
1611 | } |
1044 | |
1612 | |
1045 | void from_json (SV *jsonstr) |
1613 | void from_json (SV *jsonstr) |
1046 | PPCODE: |
1614 | PPCODE: |
|
|
1615 | { |
|
|
1616 | JSON json = { F_DEFAULT | F_UTF8 }; |
1047 | XPUSHs (decode_json (jsonstr, F_UTF8)); |
1617 | XPUSHs (decode_json (jsonstr, &json, 0)); |
|
|
1618 | } |
1048 | |
1619 | |