… | |
… | |
4 | |
4 | |
5 | #include <assert.h> |
5 | #include <assert.h> |
6 | #include <string.h> |
6 | #include <string.h> |
7 | #include <stdlib.h> |
7 | #include <stdlib.h> |
8 | #include <stdio.h> |
8 | #include <stdio.h> |
|
|
9 | #include <limits.h> |
9 | #include <float.h> |
10 | #include <float.h> |
10 | |
11 | |
11 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
12 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
12 | # define snprintf _snprintf // C compilers have this in stdio.h |
13 | # define snprintf _snprintf // C compilers have this in stdio.h |
13 | #endif |
14 | #endif |
… | |
… | |
15 | // some old perls do not have this, try to make it work, no |
16 | // 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 | // guarentees, though. if it breaks, you get to keep the pieces. |
17 | #ifndef UTF8_MAXBYTES |
18 | #ifndef UTF8_MAXBYTES |
18 | # define UTF8_MAXBYTES 13 |
19 | # define UTF8_MAXBYTES 13 |
19 | #endif |
20 | #endif |
|
|
21 | |
|
|
22 | #define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2) |
20 | |
23 | |
21 | #define F_ASCII 0x00000001UL |
24 | #define F_ASCII 0x00000001UL |
22 | #define F_LATIN1 0x00000002UL |
25 | #define F_LATIN1 0x00000002UL |
23 | #define F_UTF8 0x00000004UL |
26 | #define F_UTF8 0x00000004UL |
24 | #define F_INDENT 0x00000008UL |
27 | #define F_INDENT 0x00000008UL |
… | |
… | |
28 | #define F_ALLOW_NONREF 0x00000100UL |
31 | #define F_ALLOW_NONREF 0x00000100UL |
29 | #define F_SHRINK 0x00000200UL |
32 | #define F_SHRINK 0x00000200UL |
30 | #define F_ALLOW_BLESSED 0x00000400UL |
33 | #define F_ALLOW_BLESSED 0x00000400UL |
31 | #define F_CONV_BLESSED 0x00000800UL |
34 | #define F_CONV_BLESSED 0x00000800UL |
32 | #define F_RELAXED 0x00001000UL |
35 | #define F_RELAXED 0x00001000UL |
33 | |
36 | #define F_ALLOW_UNKNOWN 0x00002000UL |
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 |
37 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
39 | |
38 | |
40 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
|
|
41 | #define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE )) |
|
|
42 | |
|
|
43 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
39 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
44 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
|
|
45 | |
40 | |
46 | #define INIT_SIZE 32 // initial scalar size to be allocated |
41 | #define INIT_SIZE 32 // initial scalar size to be allocated |
47 | #define INDENT_STEP 3 // spaces per indentation level |
42 | #define INDENT_STEP 3 // spaces per indentation level |
48 | |
43 | |
49 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
44 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
50 | |
45 | |
51 | #define SB do { |
46 | #define SB do { |
52 | #define SE } while (0) |
47 | #define SE } while (0) |
53 | |
48 | |
54 | #if __GNUC__ >= 3 |
49 | #if __GNUC__ >= 3 |
55 | # define expect(expr,value) __builtin_expect ((expr),(value)) |
50 | # define expect(expr,value) __builtin_expect ((expr), (value)) |
56 | # define inline inline |
51 | # define INLINE static inline |
57 | #else |
52 | #else |
58 | # define expect(expr,value) (expr) |
53 | # define expect(expr,value) (expr) |
59 | # define inline static |
54 | # define INLINE static |
60 | #endif |
55 | #endif |
61 | |
56 | |
62 | #define expect_false(expr) expect ((expr) != 0, 0) |
57 | #define expect_false(expr) expect ((expr) != 0, 0) |
63 | #define expect_true(expr) expect ((expr) != 0, 1) |
58 | #define expect_true(expr) expect ((expr) != 0, 1) |
|
|
59 | |
|
|
60 | #define IN_RANGE_INC(type,val,beg,end) \ |
|
|
61 | ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
|
|
62 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
|
|
63 | |
|
|
64 | #define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)" |
64 | |
65 | |
65 | #ifdef USE_ITHREADS |
66 | #ifdef USE_ITHREADS |
66 | # define JSON_SLOW 1 |
67 | # define JSON_SLOW 1 |
67 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
68 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
68 | #else |
69 | #else |
… | |
… | |
71 | #endif |
72 | #endif |
72 | |
73 | |
73 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
74 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
74 | static SV *json_true, *json_false; |
75 | static SV *json_true, *json_false; |
75 | |
76 | |
|
|
77 | enum { |
|
|
78 | INCR_M_WS = 0, // initial whitespace skipping, must be 0 |
|
|
79 | INCR_M_STR, // inside string |
|
|
80 | INCR_M_BS, // inside backslash |
|
|
81 | INCR_M_JSON // outside anything, count nesting |
|
|
82 | }; |
|
|
83 | |
|
|
84 | #define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) |
|
|
85 | |
76 | typedef struct { |
86 | typedef struct { |
77 | U32 flags; |
87 | U32 flags; |
|
|
88 | U32 max_depth; |
|
|
89 | STRLEN max_size; |
|
|
90 | |
78 | SV *cb_object; |
91 | SV *cb_object; |
79 | HV *cb_sk_object; |
92 | HV *cb_sk_object; |
|
|
93 | |
|
|
94 | // for the incremental parser |
|
|
95 | SV *incr_text; // the source text so far |
|
|
96 | STRLEN incr_pos; // the current offset into the text |
|
|
97 | int incr_nest; // {[]}-nesting level |
|
|
98 | unsigned char incr_mode; |
80 | } JSON; |
99 | } JSON; |
|
|
100 | |
|
|
101 | INLINE void |
|
|
102 | json_init (JSON *json) |
|
|
103 | { |
|
|
104 | Zero (json, 1, JSON); |
|
|
105 | json->max_depth = 512; |
|
|
106 | } |
81 | |
107 | |
82 | ///////////////////////////////////////////////////////////////////////////// |
108 | ///////////////////////////////////////////////////////////////////////////// |
83 | // utility functions |
109 | // utility functions |
84 | |
110 | |
85 | inline void |
111 | INLINE SV * |
|
|
112 | get_bool (const char *name) |
|
|
113 | { |
|
|
114 | SV *sv = get_sv (name, 1); |
|
|
115 | |
|
|
116 | SvREADONLY_on (sv); |
|
|
117 | SvREADONLY_on (SvRV (sv)); |
|
|
118 | |
|
|
119 | return sv; |
|
|
120 | } |
|
|
121 | |
|
|
122 | INLINE void |
86 | shrink (SV *sv) |
123 | shrink (SV *sv) |
87 | { |
124 | { |
88 | sv_utf8_downgrade (sv, 1); |
125 | sv_utf8_downgrade (sv, 1); |
|
|
126 | |
89 | if (SvLEN (sv) > SvCUR (sv) + 1) |
127 | if (SvLEN (sv) > SvCUR (sv) + 1) |
90 | { |
128 | { |
91 | #ifdef SvPV_shrink_to_cur |
129 | #ifdef SvPV_shrink_to_cur |
92 | SvPV_shrink_to_cur (sv); |
130 | SvPV_shrink_to_cur (sv); |
93 | #elif defined (SvPV_renew) |
131 | #elif defined (SvPV_renew) |
… | |
… | |
99 | // decode an utf-8 character and return it, or (UV)-1 in |
137 | // decode an utf-8 character and return it, or (UV)-1 in |
100 | // case of an error. |
138 | // case of an error. |
101 | // we special-case "safe" characters from U+80 .. U+7FF, |
139 | // we special-case "safe" characters from U+80 .. U+7FF, |
102 | // but use the very good perl function to parse anything else. |
140 | // but use the very good perl function to parse anything else. |
103 | // note that we never call this function for a ascii codepoints |
141 | // note that we never call this function for a ascii codepoints |
104 | inline UV |
142 | INLINE UV |
105 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
143 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
106 | { |
144 | { |
107 | if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) |
145 | if (expect_true (len >= 2 |
108 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
146 | && IN_RANGE_INC (char, s[0], 0xc2, 0xdf) |
109 | else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
147 | && IN_RANGE_INC (char, s[1], 0x80, 0xbf))) |
110 | { |
148 | { |
111 | *clen = 2; |
149 | *clen = 2; |
112 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
150 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
113 | } |
151 | } |
114 | else |
152 | else |
115 | { |
153 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
116 | *clen = (STRLEN)-1; |
154 | } |
117 | return (UV)-1; |
155 | |
118 | } |
156 | // likewise for encoding, also never called for ascii codepoints |
|
|
157 | // this function takes advantage of this fact, although current gccs |
|
|
158 | // seem to optimise the check for >= 0x80 away anyways |
|
|
159 | INLINE unsigned char * |
|
|
160 | encode_utf8 (unsigned char *s, UV ch) |
|
|
161 | { |
|
|
162 | if (expect_false (ch < 0x000080)) |
|
|
163 | *s++ = ch; |
|
|
164 | else if (expect_true (ch < 0x000800)) |
|
|
165 | *s++ = 0xc0 | ( ch >> 6), |
|
|
166 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
167 | else if ( ch < 0x010000) |
|
|
168 | *s++ = 0xe0 | ( ch >> 12), |
|
|
169 | *s++ = 0x80 | ((ch >> 6) & 0x3f), |
|
|
170 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
171 | else if ( ch < 0x110000) |
|
|
172 | *s++ = 0xf0 | ( ch >> 18), |
|
|
173 | *s++ = 0x80 | ((ch >> 12) & 0x3f), |
|
|
174 | *s++ = 0x80 | ((ch >> 6) & 0x3f), |
|
|
175 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
176 | |
|
|
177 | return s; |
119 | } |
178 | } |
120 | |
179 | |
121 | ///////////////////////////////////////////////////////////////////////////// |
180 | ///////////////////////////////////////////////////////////////////////////// |
122 | // encoder |
181 | // encoder |
123 | |
182 | |
… | |
… | |
127 | char *cur; // SvPVX (sv) + current output position |
186 | char *cur; // SvPVX (sv) + current output position |
128 | char *end; // SvEND (sv) |
187 | char *end; // SvEND (sv) |
129 | SV *sv; // result scalar |
188 | SV *sv; // result scalar |
130 | JSON json; |
189 | JSON json; |
131 | U32 indent; // indentation level |
190 | U32 indent; // indentation level |
132 | U32 maxdepth; // max. indentation/recursion level |
191 | UV limit; // escape character values >= this value when encoding |
133 | } enc_t; |
192 | } enc_t; |
134 | |
193 | |
135 | inline void |
194 | INLINE void |
136 | need (enc_t *enc, STRLEN len) |
195 | need (enc_t *enc, STRLEN len) |
137 | { |
196 | { |
138 | if (expect_false (enc->cur + len >= enc->end)) |
197 | if (expect_false (enc->cur + len >= enc->end)) |
139 | { |
198 | { |
140 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
199 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
… | |
… | |
142 | enc->cur = SvPVX (enc->sv) + cur; |
201 | enc->cur = SvPVX (enc->sv) + cur; |
143 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
202 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
144 | } |
203 | } |
145 | } |
204 | } |
146 | |
205 | |
147 | inline void |
206 | INLINE void |
148 | encode_ch (enc_t *enc, char ch) |
207 | encode_ch (enc_t *enc, char ch) |
149 | { |
208 | { |
150 | need (enc, 1); |
209 | need (enc, 1); |
151 | *enc->cur++ = ch; |
210 | *enc->cur++ = ch; |
152 | } |
211 | } |
… | |
… | |
206 | { |
265 | { |
207 | uch = ch; |
266 | uch = ch; |
208 | clen = 1; |
267 | clen = 1; |
209 | } |
268 | } |
210 | |
269 | |
211 | if (uch > 0x10FFFFUL) |
270 | if (uch < 0x80/*0x20*/ || uch >= enc->limit) |
212 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
|
|
213 | |
|
|
214 | if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF)) |
|
|
215 | { |
271 | { |
216 | if (uch > 0xFFFFUL) |
272 | if (uch >= 0x10000UL) |
217 | { |
273 | { |
|
|
274 | if (uch >= 0x110000UL) |
|
|
275 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
|
|
276 | |
218 | need (enc, len += 11); |
277 | need (enc, len += 11); |
219 | sprintf (enc->cur, "\\u%04x\\u%04x", |
278 | sprintf (enc->cur, "\\u%04x\\u%04x", |
220 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
279 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
221 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
280 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
222 | enc->cur += 12; |
281 | enc->cur += 12; |
… | |
… | |
250 | while (--clen); |
309 | while (--clen); |
251 | } |
310 | } |
252 | else |
311 | else |
253 | { |
312 | { |
254 | need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
313 | need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
255 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
314 | enc->cur = encode_utf8 (enc->cur, uch); |
256 | ++str; |
315 | ++str; |
257 | } |
316 | } |
258 | } |
317 | } |
259 | } |
318 | } |
260 | } |
319 | } |
261 | |
320 | |
262 | --len; |
321 | --len; |
263 | } |
322 | } |
264 | } |
323 | } |
265 | |
324 | |
266 | inline void |
325 | INLINE void |
267 | encode_indent (enc_t *enc) |
326 | encode_indent (enc_t *enc) |
268 | { |
327 | { |
269 | if (enc->json.flags & F_INDENT) |
328 | if (enc->json.flags & F_INDENT) |
270 | { |
329 | { |
271 | int spaces = enc->indent * INDENT_STEP; |
330 | int spaces = enc->indent * INDENT_STEP; |
… | |
… | |
274 | memset (enc->cur, ' ', spaces); |
333 | memset (enc->cur, ' ', spaces); |
275 | enc->cur += spaces; |
334 | enc->cur += spaces; |
276 | } |
335 | } |
277 | } |
336 | } |
278 | |
337 | |
279 | inline void |
338 | INLINE void |
280 | encode_space (enc_t *enc) |
339 | encode_space (enc_t *enc) |
281 | { |
340 | { |
282 | need (enc, 1); |
341 | need (enc, 1); |
283 | encode_ch (enc, ' '); |
342 | encode_ch (enc, ' '); |
284 | } |
343 | } |
285 | |
344 | |
286 | inline void |
345 | INLINE void |
287 | encode_nl (enc_t *enc) |
346 | encode_nl (enc_t *enc) |
288 | { |
347 | { |
289 | if (enc->json.flags & F_INDENT) |
348 | if (enc->json.flags & F_INDENT) |
290 | { |
349 | { |
291 | need (enc, 1); |
350 | need (enc, 1); |
292 | encode_ch (enc, '\n'); |
351 | encode_ch (enc, '\n'); |
293 | } |
352 | } |
294 | } |
353 | } |
295 | |
354 | |
296 | inline void |
355 | INLINE void |
297 | encode_comma (enc_t *enc) |
356 | encode_comma (enc_t *enc) |
298 | { |
357 | { |
299 | encode_ch (enc, ','); |
358 | encode_ch (enc, ','); |
300 | |
359 | |
301 | if (enc->json.flags & F_INDENT) |
360 | if (enc->json.flags & F_INDENT) |
… | |
… | |
309 | static void |
368 | static void |
310 | encode_av (enc_t *enc, AV *av) |
369 | encode_av (enc_t *enc, AV *av) |
311 | { |
370 | { |
312 | int i, len = av_len (av); |
371 | int i, len = av_len (av); |
313 | |
372 | |
314 | if (enc->indent >= enc->maxdepth) |
373 | if (enc->indent >= enc->json.max_depth) |
315 | croak ("data structure too deep (hit recursion limit)"); |
374 | croak (ERR_NESTING_EXCEEDED); |
316 | |
375 | |
317 | encode_ch (enc, '['); |
376 | encode_ch (enc, '['); |
318 | |
377 | |
319 | if (len >= 0) |
378 | if (len >= 0) |
320 | { |
379 | { |
… | |
… | |
394 | |
453 | |
395 | static void |
454 | static void |
396 | encode_hv (enc_t *enc, HV *hv) |
455 | encode_hv (enc_t *enc, HV *hv) |
397 | { |
456 | { |
398 | HE *he; |
457 | HE *he; |
399 | int count; |
|
|
400 | |
458 | |
401 | if (enc->indent >= enc->maxdepth) |
459 | if (enc->indent >= enc->json.max_depth) |
402 | croak ("data structure too deep (hit recursion limit)"); |
460 | croak (ERR_NESTING_EXCEEDED); |
403 | |
461 | |
404 | encode_ch (enc, '{'); |
462 | encode_ch (enc, '{'); |
405 | |
463 | |
406 | // for canonical output we have to sort by keys first |
464 | // for canonical output we have to sort by keys first |
407 | // actually, this is mostly due to the stupid so-called |
465 | // actually, this is mostly due to the stupid so-called |
… | |
… | |
588 | |
646 | |
589 | if (len == 1 && *pv == '1') |
647 | if (len == 1 && *pv == '1') |
590 | encode_str (enc, "true", 4, 0); |
648 | encode_str (enc, "true", 4, 0); |
591 | else if (len == 1 && *pv == '0') |
649 | else if (len == 1 && *pv == '0') |
592 | encode_str (enc, "false", 5, 0); |
650 | encode_str (enc, "false", 5, 0); |
|
|
651 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
652 | encode_str (enc, "null", 4, 0); |
593 | else |
653 | else |
594 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
654 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
595 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
655 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
596 | } |
656 | } |
|
|
657 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
658 | encode_str (enc, "null", 4, 0); |
597 | else |
659 | else |
598 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
660 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
599 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
661 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
600 | } |
662 | } |
601 | |
663 | |
… | |
… | |
619 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
681 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
620 | enc->cur += strlen (enc->cur); |
682 | enc->cur += strlen (enc->cur); |
621 | } |
683 | } |
622 | else if (SvIOKp (sv)) |
684 | else if (SvIOKp (sv)) |
623 | { |
685 | { |
624 | // we assume we can always read an IV as a UV |
686 | // we assume we can always read an IV as a UV and vice versa |
625 | if (SvUV (sv) & ~(UV)0x7fff) |
687 | // we assume two's complement |
626 | { |
688 | // we assume no aliasing issues in the union |
627 | // large integer, use the (rather slow) snprintf way. |
689 | if (SvIsUV (sv) ? SvUVX (sv) <= 59000 |
628 | need (enc, sizeof (UV) * 3); |
690 | : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000) |
629 | enc->cur += |
|
|
630 | SvIsUV(sv) |
|
|
631 | ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv)) |
|
|
632 | : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv)); |
|
|
633 | } |
|
|
634 | else |
|
|
635 | { |
691 | { |
636 | // optimise the "small number case" |
692 | // optimise the "small number case" |
637 | // code will likely be branchless and use only a single multiplication |
693 | // code will likely be branchless and use only a single multiplication |
|
|
694 | // works for numbers up to 59074 |
638 | I32 i = SvIV (sv); |
695 | I32 i = SvIVX (sv); |
639 | U32 u; |
696 | U32 u; |
640 | char digit, nz = 0; |
697 | char digit, nz = 0; |
641 | |
698 | |
642 | need (enc, 6); |
699 | need (enc, 6); |
643 | |
700 | |
… | |
… | |
649 | |
706 | |
650 | // now output digit by digit, each time masking out the integer part |
707 | // 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, |
708 | // and multiplying by 5 while moving the decimal point one to the right, |
652 | // resulting in a net multiplication by 10. |
709 | // resulting in a net multiplication by 10. |
653 | // we always write the digit to memory but conditionally increment |
710 | // we always write the digit to memory but conditionally increment |
654 | // the pointer, to ease the usage of conditional move instructions. |
711 | // the pointer, to enable the use of conditional move instructions. |
655 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; |
712 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5; |
656 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; |
713 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5; |
657 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; |
714 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5; |
658 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; |
715 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5; |
659 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
716 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
660 | } |
717 | } |
|
|
718 | else |
|
|
719 | { |
|
|
720 | // large integer, use the (rather slow) snprintf way. |
|
|
721 | need (enc, IVUV_MAXCHARS); |
|
|
722 | enc->cur += |
|
|
723 | SvIsUV(sv) |
|
|
724 | ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) |
|
|
725 | : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); |
|
|
726 | } |
661 | } |
727 | } |
662 | else if (SvROK (sv)) |
728 | else if (SvROK (sv)) |
663 | encode_rv (enc, SvRV (sv)); |
729 | encode_rv (enc, SvRV (sv)); |
664 | else if (!SvOK (sv)) |
730 | else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) |
665 | encode_str (enc, "null", 4, 0); |
731 | encode_str (enc, "null", 4, 0); |
666 | else |
732 | else |
667 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
733 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
668 | SvPV_nolen (sv), SvFLAGS (sv)); |
734 | SvPV_nolen (sv), SvFLAGS (sv)); |
669 | } |
735 | } |
… | |
… | |
679 | enc.json = *json; |
745 | enc.json = *json; |
680 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
746 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
681 | enc.cur = SvPVX (enc.sv); |
747 | enc.cur = SvPVX (enc.sv); |
682 | enc.end = SvEND (enc.sv); |
748 | enc.end = SvEND (enc.sv); |
683 | enc.indent = 0; |
749 | enc.indent = 0; |
684 | enc.maxdepth = DEC_DEPTH (enc.json.flags); |
750 | enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
|
|
751 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
|
|
752 | : 0x110000UL; |
685 | |
753 | |
686 | SvPOK_only (enc.sv); |
754 | SvPOK_only (enc.sv); |
687 | encode_sv (&enc, scalar); |
755 | encode_sv (&enc, scalar); |
688 | |
756 | |
689 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
757 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
… | |
… | |
710 | JSON json; |
778 | JSON json; |
711 | U32 depth; // recursion depth |
779 | U32 depth; // recursion depth |
712 | U32 maxdepth; // recursion depth limit |
780 | U32 maxdepth; // recursion depth limit |
713 | } dec_t; |
781 | } dec_t; |
714 | |
782 | |
715 | inline void |
783 | INLINE void |
716 | decode_comment (dec_t *dec) |
784 | decode_comment (dec_t *dec) |
717 | { |
785 | { |
718 | // only '#'-style comments allowed a.t.m. |
786 | // only '#'-style comments allowed a.t.m. |
719 | |
787 | |
720 | while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) |
788 | while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) |
721 | ++dec->cur; |
789 | ++dec->cur; |
722 | } |
790 | } |
723 | |
791 | |
724 | inline void |
792 | INLINE void |
725 | decode_ws (dec_t *dec) |
793 | decode_ws (dec_t *dec) |
726 | { |
794 | { |
727 | for (;;) |
795 | for (;;) |
728 | { |
796 | { |
729 | char ch = *dec->cur; |
797 | char ch = *dec->cur; |
… | |
… | |
753 | if (*dec->cur != ch) \ |
821 | if (*dec->cur != ch) \ |
754 | ERR (# ch " expected"); \ |
822 | ERR (# ch " expected"); \ |
755 | ++dec->cur; \ |
823 | ++dec->cur; \ |
756 | SE |
824 | SE |
757 | |
825 | |
758 | #define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") |
826 | #define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED) |
759 | #define DEC_DEC_DEPTH --dec->depth |
827 | #define DEC_DEC_DEPTH --dec->depth |
760 | |
828 | |
761 | static SV *decode_sv (dec_t *dec); |
829 | static SV *decode_sv (dec_t *dec); |
762 | |
830 | |
763 | static signed char decode_hexdigit[256]; |
831 | static signed char decode_hexdigit[256]; |
… | |
… | |
855 | |
923 | |
856 | if (hi >= 0x80) |
924 | if (hi >= 0x80) |
857 | { |
925 | { |
858 | utf8 = 1; |
926 | utf8 = 1; |
859 | |
927 | |
860 | cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); |
928 | cur = encode_utf8 (cur, hi); |
861 | } |
929 | } |
862 | else |
930 | else |
863 | *cur++ = hi; |
931 | *cur++ = hi; |
864 | } |
932 | } |
865 | break; |
933 | break; |
… | |
… | |
867 | default: |
935 | default: |
868 | --dec_cur; |
936 | --dec_cur; |
869 | ERR ("illegal backslash escape sequence in string"); |
937 | ERR ("illegal backslash escape sequence in string"); |
870 | } |
938 | } |
871 | } |
939 | } |
872 | else if (expect_true (ch >= 0x20 && ch <= 0x7f)) |
940 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
873 | *cur++ = ch; |
941 | *cur++ = ch; |
874 | else if (ch >= 0x80) |
942 | else if (ch >= 0x80) |
875 | { |
943 | { |
876 | STRLEN clen; |
944 | STRLEN clen; |
877 | UV uch; |
945 | UV uch; |
… | |
… | |
1000 | |
1068 | |
1001 | if (!is_nv) |
1069 | if (!is_nv) |
1002 | { |
1070 | { |
1003 | int len = dec->cur - start; |
1071 | int len = dec->cur - start; |
1004 | |
1072 | |
1005 | // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so |
1073 | // special case the rather common 1..5-digit-int case |
1006 | if (*start == '-') |
1074 | if (*start == '-') |
1007 | switch (len) |
1075 | switch (len) |
1008 | { |
1076 | { |
1009 | case 2: return newSViv (-( start [1] - '0' * 1)); |
1077 | case 2: return newSViv (-( start [1] - '0' * 1)); |
1010 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
1078 | 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)); |
1079 | 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)); |
1080 | case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
|
|
1081 | case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
1013 | } |
1082 | } |
1014 | else |
1083 | else |
1015 | switch (len) |
1084 | switch (len) |
1016 | { |
1085 | { |
1017 | case 1: return newSViv ( start [0] - '0' * 1); |
1086 | case 1: return newSViv ( start [0] - '0' * 1); |
1018 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1087 | 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); |
1088 | 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); |
1089 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
|
|
1090 | case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
1021 | } |
1091 | } |
1022 | |
1092 | |
1023 | { |
1093 | { |
1024 | UV uv; |
1094 | UV uv; |
1025 | int numtype = grok_number (start, len, &uv); |
1095 | int numtype = grok_number (start, len, &uv); |
… | |
… | |
1133 | char *p = dec->cur; |
1203 | char *p = dec->cur; |
1134 | char *e = p + 24; // only try up to 24 bytes |
1204 | char *e = p + 24; // only try up to 24 bytes |
1135 | |
1205 | |
1136 | for (;;) |
1206 | for (;;) |
1137 | { |
1207 | { |
1138 | // the >= 0x80 is true on most architectures |
1208 | // the >= 0x80 is false on most architectures |
1139 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1209 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1140 | { |
1210 | { |
1141 | // slow path, back up and use decode_str |
1211 | // slow path, back up and use decode_str |
1142 | SV *key = decode_str (dec); |
1212 | SV *key = decode_str (dec); |
1143 | if (!key) |
1213 | if (!key) |
… | |
… | |
1274 | |
1344 | |
1275 | static SV * |
1345 | static SV * |
1276 | decode_sv (dec_t *dec) |
1346 | decode_sv (dec_t *dec) |
1277 | { |
1347 | { |
1278 | // the beauty of JSON: you need exactly one character lookahead |
1348 | // the beauty of JSON: you need exactly one character lookahead |
1279 | // to parse anything. |
1349 | // to parse everything. |
1280 | switch (*dec->cur) |
1350 | switch (*dec->cur) |
1281 | { |
1351 | { |
1282 | case '"': ++dec->cur; return decode_str (dec); |
1352 | case '"': ++dec->cur; return decode_str (dec); |
1283 | case '[': ++dec->cur; return decode_av (dec); |
1353 | case '[': ++dec->cur; return decode_av (dec); |
1284 | case '{': ++dec->cur; return decode_hv (dec); |
1354 | case '{': ++dec->cur; return decode_hv (dec); |
1285 | |
1355 | |
1286 | case '-': |
1356 | case '-': |
1287 | case '0': case '1': case '2': case '3': case '4': |
1357 | case '0': case '1': case '2': case '3': case '4': |
1288 | case '5': case '6': case '7': case '8': case '9': |
1358 | case '5': case '6': case '7': case '8': case '9': |
1289 | return decode_num (dec); |
1359 | return decode_num (dec); |
… | |
… | |
1291 | case 't': |
1361 | case 't': |
1292 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1362 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1293 | { |
1363 | { |
1294 | dec->cur += 4; |
1364 | dec->cur += 4; |
1295 | #if JSON_SLOW |
1365 | #if JSON_SLOW |
1296 | json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); |
1366 | json_true = get_bool ("JSON::XS::true"); |
1297 | #endif |
1367 | #endif |
1298 | return SvREFCNT_inc (json_true); |
1368 | return newSVsv (json_true); |
1299 | } |
1369 | } |
1300 | else |
1370 | else |
1301 | ERR ("'true' expected"); |
1371 | ERR ("'true' expected"); |
1302 | |
1372 | |
1303 | break; |
1373 | break; |
… | |
… | |
1305 | case 'f': |
1375 | case 'f': |
1306 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1376 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1307 | { |
1377 | { |
1308 | dec->cur += 5; |
1378 | dec->cur += 5; |
1309 | #if JSON_SLOW |
1379 | #if JSON_SLOW |
1310 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1380 | json_false = get_bool ("JSON::XS::false"); |
1311 | #endif |
1381 | #endif |
1312 | return SvREFCNT_inc (json_false); |
1382 | return newSVsv (json_false); |
1313 | } |
1383 | } |
1314 | else |
1384 | else |
1315 | ERR ("'false' expected"); |
1385 | ERR ("'false' expected"); |
1316 | |
1386 | |
1317 | break; |
1387 | break; |
… | |
… | |
1335 | fail: |
1405 | fail: |
1336 | return 0; |
1406 | return 0; |
1337 | } |
1407 | } |
1338 | |
1408 | |
1339 | static SV * |
1409 | static SV * |
1340 | decode_json (SV *string, JSON *json, UV *offset_return) |
1410 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1341 | { |
1411 | { |
1342 | dec_t dec; |
1412 | dec_t dec; |
1343 | UV offset; |
1413 | STRLEN offset; |
1344 | SV *sv; |
1414 | SV *sv; |
1345 | |
1415 | |
1346 | SvGETMAGIC (string); |
1416 | SvGETMAGIC (string); |
1347 | SvUPGRADE (string, SVt_PV); |
1417 | SvUPGRADE (string, SVt_PV); |
1348 | |
1418 | |
1349 | if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) |
1419 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
|
|
1420 | * assertion with -DDEBUGGING, although SvCUR is documented to |
|
|
1421 | * return the xpv_cur field which certainly exists after upgrading. |
|
|
1422 | * according to nicholas clark, calling SvPOK fixes this. |
|
|
1423 | * But it doesn't fix it, so try another workaround, call SvPV_nolen |
|
|
1424 | * and hope for the best. |
|
|
1425 | */ |
|
|
1426 | #ifdef DEBUGGING |
|
|
1427 | SvPV_nolen (string); |
|
|
1428 | #endif |
|
|
1429 | |
|
|
1430 | if (SvCUR (string) > json->max_size && json->max_size) |
1350 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1431 | 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)); |
1432 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
1352 | |
1433 | |
1353 | if (json->flags & F_UTF8) |
1434 | if (json->flags & F_UTF8) |
1354 | sv_utf8_downgrade (string, 0); |
1435 | sv_utf8_downgrade (string, 0); |
1355 | else |
1436 | else |
1356 | sv_utf8_upgrade (string); |
1437 | sv_utf8_upgrade (string); |
1357 | |
1438 | |
1358 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1439 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1359 | |
1440 | |
1360 | dec.json = *json; |
1441 | dec.json = *json; |
1361 | dec.cur = SvPVX (string); |
1442 | dec.cur = SvPVX (string); |
1362 | dec.end = SvEND (string); |
1443 | dec.end = SvEND (string); |
1363 | dec.err = 0; |
1444 | dec.err = 0; |
1364 | dec.depth = 0; |
1445 | dec.depth = 0; |
1365 | dec.maxdepth = DEC_DEPTH (dec.json.flags); |
|
|
1366 | |
1446 | |
1367 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1447 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1368 | dec.json.flags |= F_HOOK; |
1448 | dec.json.flags |= F_HOOK; |
1369 | |
1449 | |
1370 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1450 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
… | |
… | |
1421 | |
1501 | |
1422 | return sv; |
1502 | return sv; |
1423 | } |
1503 | } |
1424 | |
1504 | |
1425 | ///////////////////////////////////////////////////////////////////////////// |
1505 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1506 | // incremental parser |
|
|
1507 | |
|
|
1508 | static void |
|
|
1509 | incr_parse (JSON *self) |
|
|
1510 | { |
|
|
1511 | const char *p = SvPVX (self->incr_text) + self->incr_pos; |
|
|
1512 | |
|
|
1513 | for (;;) |
|
|
1514 | { |
|
|
1515 | //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D |
|
|
1516 | switch (self->incr_mode) |
|
|
1517 | { |
|
|
1518 | // only used for intiial whitespace skipping |
|
|
1519 | case INCR_M_WS: |
|
|
1520 | for (;;) |
|
|
1521 | { |
|
|
1522 | if (*p > 0x20) |
|
|
1523 | { |
|
|
1524 | self->incr_mode = INCR_M_JSON; |
|
|
1525 | goto incr_m_json; |
|
|
1526 | } |
|
|
1527 | else if (!*p) |
|
|
1528 | goto interrupt; |
|
|
1529 | |
|
|
1530 | ++p; |
|
|
1531 | } |
|
|
1532 | |
|
|
1533 | // skip a single char inside a string (for \\-processing) |
|
|
1534 | case INCR_M_BS: |
|
|
1535 | if (!*p) |
|
|
1536 | goto interrupt; |
|
|
1537 | |
|
|
1538 | ++p; |
|
|
1539 | self->incr_mode = INCR_M_STR; |
|
|
1540 | goto incr_m_str; |
|
|
1541 | |
|
|
1542 | // inside a string |
|
|
1543 | case INCR_M_STR: |
|
|
1544 | incr_m_str: |
|
|
1545 | for (;;) |
|
|
1546 | { |
|
|
1547 | if (*p == '"') |
|
|
1548 | { |
|
|
1549 | ++p; |
|
|
1550 | self->incr_mode = INCR_M_JSON; |
|
|
1551 | |
|
|
1552 | if (!self->incr_nest) |
|
|
1553 | goto interrupt; |
|
|
1554 | |
|
|
1555 | goto incr_m_json; |
|
|
1556 | } |
|
|
1557 | else if (*p == '\\') |
|
|
1558 | { |
|
|
1559 | ++p; // "virtually" consumes character after \ |
|
|
1560 | |
|
|
1561 | if (!*p) // if at end of string we have to switch modes |
|
|
1562 | { |
|
|
1563 | self->incr_mode = INCR_M_BS; |
|
|
1564 | goto interrupt; |
|
|
1565 | } |
|
|
1566 | } |
|
|
1567 | else if (!*p) |
|
|
1568 | goto interrupt; |
|
|
1569 | |
|
|
1570 | ++p; |
|
|
1571 | } |
|
|
1572 | |
|
|
1573 | // after initial ws, outside string |
|
|
1574 | case INCR_M_JSON: |
|
|
1575 | incr_m_json: |
|
|
1576 | for (;;) |
|
|
1577 | { |
|
|
1578 | switch (*p++) |
|
|
1579 | { |
|
|
1580 | case 0: |
|
|
1581 | --p; |
|
|
1582 | goto interrupt; |
|
|
1583 | |
|
|
1584 | case 0x09: |
|
|
1585 | case 0x0a: |
|
|
1586 | case 0x0d: |
|
|
1587 | case 0x20: |
|
|
1588 | if (!self->incr_nest) |
|
|
1589 | { |
|
|
1590 | --p; // do not eat the whitespace, let the next round do it |
|
|
1591 | goto interrupt; |
|
|
1592 | } |
|
|
1593 | break; |
|
|
1594 | |
|
|
1595 | case '"': |
|
|
1596 | self->incr_mode = INCR_M_STR; |
|
|
1597 | goto incr_m_str; |
|
|
1598 | |
|
|
1599 | case '[': |
|
|
1600 | case '{': |
|
|
1601 | if (++self->incr_nest > self->max_depth) |
|
|
1602 | croak (ERR_NESTING_EXCEEDED); |
|
|
1603 | break; |
|
|
1604 | |
|
|
1605 | case ']': |
|
|
1606 | case '}': |
|
|
1607 | if (--self->incr_nest <= 0) |
|
|
1608 | goto interrupt; |
|
|
1609 | } |
|
|
1610 | } |
|
|
1611 | } |
|
|
1612 | |
|
|
1613 | modechange: |
|
|
1614 | ; |
|
|
1615 | } |
|
|
1616 | |
|
|
1617 | interrupt: |
|
|
1618 | self->incr_pos = p - SvPVX (self->incr_text); |
|
|
1619 | //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D |
|
|
1620 | } |
|
|
1621 | |
|
|
1622 | ///////////////////////////////////////////////////////////////////////////// |
1426 | // XS interface functions |
1623 | // XS interface functions |
1427 | |
1624 | |
1428 | MODULE = JSON::XS PACKAGE = JSON::XS |
1625 | MODULE = JSON::XS PACKAGE = JSON::XS |
1429 | |
1626 | |
1430 | BOOT: |
1627 | BOOT: |
… | |
… | |
1439 | : -1; |
1636 | : -1; |
1440 | |
1637 | |
1441 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1638 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1442 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
1639 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
1443 | |
1640 | |
1444 | json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); |
1641 | json_true = get_bool ("JSON::XS::true"); |
1445 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1642 | json_false = get_bool ("JSON::XS::false"); |
1446 | } |
1643 | } |
1447 | |
1644 | |
1448 | PROTOTYPES: DISABLE |
1645 | PROTOTYPES: DISABLE |
1449 | |
1646 | |
1450 | void CLONE (...) |
1647 | void CLONE (...) |
… | |
… | |
1455 | void new (char *klass) |
1652 | void new (char *klass) |
1456 | PPCODE: |
1653 | PPCODE: |
1457 | { |
1654 | { |
1458 | SV *pv = NEWSV (0, sizeof (JSON)); |
1655 | SV *pv = NEWSV (0, sizeof (JSON)); |
1459 | SvPOK_only (pv); |
1656 | SvPOK_only (pv); |
1460 | Zero (SvPVX (pv), 1, JSON); |
1657 | json_init ((JSON *)SvPVX (pv)); |
1461 | ((JSON *)SvPVX (pv))->flags = F_DEFAULT; |
1658 | XPUSHs (sv_2mortal (sv_bless ( |
1462 | XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH))); |
1659 | newRV_noinc (pv), |
|
|
1660 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
|
|
1661 | ))); |
1463 | } |
1662 | } |
1464 | |
1663 | |
1465 | void ascii (JSON *self, int enable = 1) |
1664 | void ascii (JSON *self, int enable = 1) |
1466 | ALIAS: |
1665 | ALIAS: |
1467 | ascii = F_ASCII |
1666 | ascii = F_ASCII |
… | |
… | |
1475 | allow_nonref = F_ALLOW_NONREF |
1674 | allow_nonref = F_ALLOW_NONREF |
1476 | shrink = F_SHRINK |
1675 | shrink = F_SHRINK |
1477 | allow_blessed = F_ALLOW_BLESSED |
1676 | allow_blessed = F_ALLOW_BLESSED |
1478 | convert_blessed = F_CONV_BLESSED |
1677 | convert_blessed = F_CONV_BLESSED |
1479 | relaxed = F_RELAXED |
1678 | relaxed = F_RELAXED |
|
|
1679 | allow_unknown = F_ALLOW_UNKNOWN |
1480 | PPCODE: |
1680 | PPCODE: |
1481 | { |
1681 | { |
1482 | if (enable) |
1682 | if (enable) |
1483 | self->flags |= ix; |
1683 | self->flags |= ix; |
1484 | else |
1684 | else |
… | |
… | |
1494 | get_utf8 = F_UTF8 |
1694 | get_utf8 = F_UTF8 |
1495 | get_indent = F_INDENT |
1695 | get_indent = F_INDENT |
1496 | get_canonical = F_CANONICAL |
1696 | get_canonical = F_CANONICAL |
1497 | get_space_before = F_SPACE_BEFORE |
1697 | get_space_before = F_SPACE_BEFORE |
1498 | get_space_after = F_SPACE_AFTER |
1698 | get_space_after = F_SPACE_AFTER |
1499 | get_pretty = F_PRETTY |
|
|
1500 | get_allow_nonref = F_ALLOW_NONREF |
1699 | get_allow_nonref = F_ALLOW_NONREF |
1501 | get_shrink = F_SHRINK |
1700 | get_shrink = F_SHRINK |
1502 | get_allow_blessed = F_ALLOW_BLESSED |
1701 | get_allow_blessed = F_ALLOW_BLESSED |
1503 | get_convert_blessed = F_CONV_BLESSED |
1702 | get_convert_blessed = F_CONV_BLESSED |
1504 | get_relaxed = F_RELAXED |
1703 | get_relaxed = F_RELAXED |
|
|
1704 | get_allow_unknown = F_ALLOW_UNKNOWN |
1505 | PPCODE: |
1705 | PPCODE: |
1506 | XPUSHs (boolSV (self->flags & ix)); |
1706 | XPUSHs (boolSV (self->flags & ix)); |
1507 | |
1707 | |
1508 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
1708 | void max_depth (JSON *self, U32 max_depth = 0x80000000UL) |
1509 | PPCODE: |
1709 | PPCODE: |
1510 | { |
1710 | self->max_depth = max_depth; |
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)); |
1711 | XPUSHs (ST (0)); |
1521 | } |
|
|
1522 | |
1712 | |
1523 | U32 get_max_depth (JSON *self) |
1713 | U32 get_max_depth (JSON *self) |
1524 | CODE: |
1714 | CODE: |
1525 | RETVAL = DEC_DEPTH (self->flags); |
1715 | RETVAL = self->max_depth; |
1526 | OUTPUT: |
1716 | OUTPUT: |
1527 | RETVAL |
1717 | RETVAL |
1528 | |
1718 | |
1529 | void max_size (JSON *self, UV max_size = 0) |
1719 | void max_size (JSON *self, U32 max_size = 0) |
1530 | PPCODE: |
1720 | PPCODE: |
1531 | { |
1721 | self->max_size = max_size; |
1532 | UV log2 = 0; |
|
|
1533 | |
|
|
1534 | if (max_size > 0x80000000UL) max_size = 0x80000000UL; |
|
|
1535 | if (max_size == 1) max_size = 2; |
|
|
1536 | |
|
|
1537 | while ((1UL << log2) < max_size) |
|
|
1538 | ++log2; |
|
|
1539 | |
|
|
1540 | self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); |
|
|
1541 | |
|
|
1542 | XPUSHs (ST (0)); |
1722 | XPUSHs (ST (0)); |
1543 | } |
|
|
1544 | |
1723 | |
1545 | int get_max_size (JSON *self) |
1724 | int get_max_size (JSON *self) |
1546 | CODE: |
1725 | CODE: |
1547 | RETVAL = DEC_SIZE (self->flags); |
1726 | RETVAL = self->max_size; |
1548 | OUTPUT: |
1727 | OUTPUT: |
1549 | RETVAL |
1728 | RETVAL |
1550 | |
1729 | |
1551 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1730 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1552 | PPCODE: |
1731 | PPCODE: |
… | |
… | |
1588 | XPUSHs (decode_json (jsonstr, self, 0)); |
1767 | XPUSHs (decode_json (jsonstr, self, 0)); |
1589 | |
1768 | |
1590 | void decode_prefix (JSON *self, SV *jsonstr) |
1769 | void decode_prefix (JSON *self, SV *jsonstr) |
1591 | PPCODE: |
1770 | PPCODE: |
1592 | { |
1771 | { |
1593 | UV offset; |
1772 | STRLEN offset; |
1594 | EXTEND (SP, 2); |
1773 | EXTEND (SP, 2); |
1595 | PUSHs (decode_json (jsonstr, self, &offset)); |
1774 | PUSHs (decode_json (jsonstr, self, &offset)); |
1596 | PUSHs (sv_2mortal (newSVuv (offset))); |
1775 | PUSHs (sv_2mortal (newSVuv (offset))); |
|
|
1776 | } |
|
|
1777 | |
|
|
1778 | void incr_parse (JSON *self, SV *jsonstr = 0) |
|
|
1779 | PPCODE: |
|
|
1780 | { |
|
|
1781 | if (!self->incr_text) |
|
|
1782 | self->incr_text = newSVpvn ("", 0); |
|
|
1783 | |
|
|
1784 | // append data, if any |
|
|
1785 | if (jsonstr) |
|
|
1786 | { |
|
|
1787 | if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) |
|
|
1788 | { |
|
|
1789 | /* utf-8-ness differs, need to upgrade */ |
|
|
1790 | sv_utf8_upgrade (self->incr_text); |
|
|
1791 | |
|
|
1792 | if (self->incr_pos) |
|
|
1793 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
|
|
1794 | - (U8 *)SvPVX (self->incr_text); |
|
|
1795 | } |
|
|
1796 | |
|
|
1797 | { |
|
|
1798 | STRLEN len; |
|
|
1799 | const char *str = SvPV (jsonstr, len); |
|
|
1800 | SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); |
|
|
1801 | Move (str, SvEND (self->incr_text), len, char); |
|
|
1802 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
|
|
1803 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
|
|
1804 | } |
|
|
1805 | } |
|
|
1806 | |
|
|
1807 | if (GIMME_V != G_VOID) |
|
|
1808 | do |
|
|
1809 | { |
|
|
1810 | STRLEN offset; |
|
|
1811 | |
|
|
1812 | if (!INCR_DONE (self)) |
|
|
1813 | { |
|
|
1814 | incr_parse (self); |
|
|
1815 | |
|
|
1816 | if (self->incr_pos > self->max_size && self->max_size) |
|
|
1817 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
|
|
1818 | (unsigned long)self->incr_pos, (unsigned long)self->max_size); |
|
|
1819 | |
|
|
1820 | if (!INCR_DONE (self)) |
|
|
1821 | break; |
|
|
1822 | } |
|
|
1823 | |
|
|
1824 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
|
|
1825 | |
|
|
1826 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset); |
|
|
1827 | self->incr_pos -= offset; |
|
|
1828 | self->incr_nest = 0; |
|
|
1829 | self->incr_mode = 0; |
|
|
1830 | } |
|
|
1831 | while (GIMME_V == G_ARRAY); |
|
|
1832 | } |
|
|
1833 | |
|
|
1834 | SV *incr_text (JSON *self) |
|
|
1835 | ATTRS: lvalue |
|
|
1836 | CODE: |
|
|
1837 | { |
|
|
1838 | if (self->incr_pos) |
|
|
1839 | croak ("incr_text can not be called when the incremental parser already started parsing"); |
|
|
1840 | |
|
|
1841 | RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef; |
|
|
1842 | } |
|
|
1843 | OUTPUT: |
|
|
1844 | RETVAL |
|
|
1845 | |
|
|
1846 | void incr_skip (JSON *self) |
|
|
1847 | CODE: |
|
|
1848 | { |
|
|
1849 | if (self->incr_pos) |
|
|
1850 | { |
|
|
1851 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos); |
|
|
1852 | self->incr_pos = 0; |
|
|
1853 | self->incr_nest = 0; |
|
|
1854 | self->incr_mode = 0; |
|
|
1855 | } |
|
|
1856 | } |
|
|
1857 | |
|
|
1858 | void incr_reset (JSON *self) |
|
|
1859 | CODE: |
|
|
1860 | { |
|
|
1861 | SvREFCNT_dec (self->incr_text); |
|
|
1862 | self->incr_text = 0; |
|
|
1863 | self->incr_pos = 0; |
|
|
1864 | self->incr_nest = 0; |
|
|
1865 | self->incr_mode = 0; |
1597 | } |
1866 | } |
1598 | |
1867 | |
1599 | void DESTROY (JSON *self) |
1868 | void DESTROY (JSON *self) |
1600 | CODE: |
1869 | CODE: |
1601 | SvREFCNT_dec (self->cb_sk_object); |
1870 | SvREFCNT_dec (self->cb_sk_object); |
1602 | SvREFCNT_dec (self->cb_object); |
1871 | SvREFCNT_dec (self->cb_object); |
|
|
1872 | SvREFCNT_dec (self->incr_text); |
1603 | |
1873 | |
1604 | PROTOTYPES: ENABLE |
1874 | PROTOTYPES: ENABLE |
1605 | |
1875 | |
1606 | void to_json (SV *scalar) |
1876 | void encode_json (SV *scalar) |
|
|
1877 | ALIAS: |
|
|
1878 | to_json_ = 0 |
|
|
1879 | encode_json = F_UTF8 |
1607 | PPCODE: |
1880 | PPCODE: |
1608 | { |
1881 | { |
1609 | JSON json = { F_DEFAULT | F_UTF8 }; |
1882 | JSON json; |
|
|
1883 | json_init (&json); |
|
|
1884 | json.flags |= ix; |
1610 | XPUSHs (encode_json (scalar, &json)); |
1885 | XPUSHs (encode_json (scalar, &json)); |
1611 | } |
1886 | } |
1612 | |
1887 | |
1613 | void from_json (SV *jsonstr) |
1888 | void decode_json (SV *jsonstr) |
|
|
1889 | ALIAS: |
|
|
1890 | from_json_ = 0 |
|
|
1891 | decode_json = F_UTF8 |
1614 | PPCODE: |
1892 | PPCODE: |
1615 | { |
1893 | { |
1616 | JSON json = { F_DEFAULT | F_UTF8 }; |
1894 | JSON json; |
|
|
1895 | json_init (&json); |
|
|
1896 | json.flags |= ix; |
1617 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1897 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1618 | } |
1898 | } |
1619 | |
1899 | |
|
|
1900 | |