… | |
… | |
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 |
… | |
… | |
27 | #define F_SPACE_AFTER 0x00000040UL |
30 | #define F_SPACE_AFTER 0x00000040UL |
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_MAXDEPTH 0xf8000000UL |
|
|
33 | #define S_MAXDEPTH 27 |
|
|
34 | #define F_MAXSIZE 0x01f00000UL |
35 | #define F_RELAXED 0x00001000UL |
35 | #define S_MAXSIZE 20 |
36 | #define F_ALLOW_UNKNOWN 0x00002000UL |
36 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
37 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
37 | |
38 | |
38 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
|
|
39 | #define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE )) |
|
|
40 | |
|
|
41 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
39 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
42 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
|
|
43 | |
40 | |
44 | #define INIT_SIZE 32 // initial scalar size to be allocated |
41 | #define INIT_SIZE 32 // initial scalar size to be allocated |
45 | #define INDENT_STEP 3 // spaces per indentation level |
42 | #define INDENT_STEP 3 // spaces per indentation level |
46 | |
43 | |
47 | #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 |
48 | |
45 | |
49 | #define SB do { |
46 | #define SB do { |
50 | #define SE } while (0) |
47 | #define SE } while (0) |
51 | |
48 | |
52 | #if __GNUC__ >= 3 |
49 | #if __GNUC__ >= 3 |
53 | # define expect(expr,value) __builtin_expect ((expr),(value)) |
50 | # define expect(expr,value) __builtin_expect ((expr), (value)) |
54 | # define inline inline |
51 | # define INLINE static inline |
55 | #else |
52 | #else |
56 | # define expect(expr,value) (expr) |
53 | # define expect(expr,value) (expr) |
57 | # define inline static |
54 | # define INLINE static |
58 | #endif |
55 | #endif |
59 | |
56 | |
60 | #define expect_false(expr) expect ((expr) != 0, 0) |
57 | #define expect_false(expr) expect ((expr) != 0, 0) |
61 | #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?)" |
62 | |
65 | |
63 | #ifdef USE_ITHREADS |
66 | #ifdef USE_ITHREADS |
64 | # define JSON_SLOW 1 |
67 | # define JSON_SLOW 1 |
65 | # 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)) |
66 | #else |
69 | #else |
… | |
… | |
69 | #endif |
72 | #endif |
70 | |
73 | |
71 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
74 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
72 | static SV *json_true, *json_false; |
75 | static SV *json_true, *json_false; |
73 | |
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 | |
74 | typedef struct { |
86 | typedef struct { |
75 | U32 flags; |
87 | U32 flags; |
|
|
88 | U32 max_depth; |
|
|
89 | STRLEN max_size; |
|
|
90 | |
76 | SV *cb_object; |
91 | SV *cb_object; |
77 | 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; |
78 | } JSON; |
99 | } JSON; |
|
|
100 | |
|
|
101 | INLINE void |
|
|
102 | json_init (JSON *json) |
|
|
103 | { |
|
|
104 | Zero (json, 1, JSON); |
|
|
105 | json->max_depth = 512; |
|
|
106 | } |
79 | |
107 | |
80 | ///////////////////////////////////////////////////////////////////////////// |
108 | ///////////////////////////////////////////////////////////////////////////// |
81 | // utility functions |
109 | // utility functions |
82 | |
110 | |
83 | 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 |
84 | shrink (SV *sv) |
123 | shrink (SV *sv) |
85 | { |
124 | { |
86 | sv_utf8_downgrade (sv, 1); |
125 | sv_utf8_downgrade (sv, 1); |
|
|
126 | |
87 | if (SvLEN (sv) > SvCUR (sv) + 1) |
127 | if (SvLEN (sv) > SvCUR (sv) + 1) |
88 | { |
128 | { |
89 | #ifdef SvPV_shrink_to_cur |
129 | #ifdef SvPV_shrink_to_cur |
90 | SvPV_shrink_to_cur (sv); |
130 | SvPV_shrink_to_cur (sv); |
91 | #elif defined (SvPV_renew) |
131 | #elif defined (SvPV_renew) |
… | |
… | |
97 | // 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 |
98 | // case of an error. |
138 | // case of an error. |
99 | // we special-case "safe" characters from U+80 .. U+7FF, |
139 | // we special-case "safe" characters from U+80 .. U+7FF, |
100 | // but use the very good perl function to parse anything else. |
140 | // but use the very good perl function to parse anything else. |
101 | // note that we never call this function for a ascii codepoints |
141 | // note that we never call this function for a ascii codepoints |
102 | inline UV |
142 | INLINE UV |
103 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
143 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
104 | { |
144 | { |
105 | if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) |
145 | if (expect_true (len >= 2 |
106 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
146 | && IN_RANGE_INC (char, s[0], 0xc2, 0xdf) |
107 | else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
147 | && IN_RANGE_INC (char, s[1], 0x80, 0xbf))) |
108 | { |
148 | { |
109 | *clen = 2; |
149 | *clen = 2; |
110 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
150 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
111 | } |
151 | } |
112 | else |
152 | else |
113 | { |
153 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
114 | *clen = (STRLEN)-1; |
154 | } |
115 | return (UV)-1; |
155 | |
116 | } |
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; |
117 | } |
178 | } |
118 | |
179 | |
119 | ///////////////////////////////////////////////////////////////////////////// |
180 | ///////////////////////////////////////////////////////////////////////////// |
120 | // encoder |
181 | // encoder |
121 | |
182 | |
… | |
… | |
125 | char *cur; // SvPVX (sv) + current output position |
186 | char *cur; // SvPVX (sv) + current output position |
126 | char *end; // SvEND (sv) |
187 | char *end; // SvEND (sv) |
127 | SV *sv; // result scalar |
188 | SV *sv; // result scalar |
128 | JSON json; |
189 | JSON json; |
129 | U32 indent; // indentation level |
190 | U32 indent; // indentation level |
130 | U32 maxdepth; // max. indentation/recursion level |
191 | UV limit; // escape character values >= this value when encoding |
131 | } enc_t; |
192 | } enc_t; |
132 | |
193 | |
133 | inline void |
194 | INLINE void |
134 | need (enc_t *enc, STRLEN len) |
195 | need (enc_t *enc, STRLEN len) |
135 | { |
196 | { |
136 | if (expect_false (enc->cur + len >= enc->end)) |
197 | if (expect_false (enc->cur + len >= enc->end)) |
137 | { |
198 | { |
138 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
199 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
… | |
… | |
140 | enc->cur = SvPVX (enc->sv) + cur; |
201 | enc->cur = SvPVX (enc->sv) + cur; |
141 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
202 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
142 | } |
203 | } |
143 | } |
204 | } |
144 | |
205 | |
145 | inline void |
206 | INLINE void |
146 | encode_ch (enc_t *enc, char ch) |
207 | encode_ch (enc_t *enc, char ch) |
147 | { |
208 | { |
148 | need (enc, 1); |
209 | need (enc, 1); |
149 | *enc->cur++ = ch; |
210 | *enc->cur++ = ch; |
150 | } |
211 | } |
… | |
… | |
204 | { |
265 | { |
205 | uch = ch; |
266 | uch = ch; |
206 | clen = 1; |
267 | clen = 1; |
207 | } |
268 | } |
208 | |
269 | |
209 | if (uch > 0x10FFFFUL) |
270 | if (uch < 0x80/*0x20*/ || uch >= enc->limit) |
210 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
|
|
211 | |
|
|
212 | if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF)) |
|
|
213 | { |
271 | { |
214 | if (uch > 0xFFFFUL) |
272 | if (uch >= 0x10000UL) |
215 | { |
273 | { |
|
|
274 | if (uch >= 0x110000UL) |
|
|
275 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
|
|
276 | |
216 | need (enc, len += 11); |
277 | need (enc, len += 11); |
217 | sprintf (enc->cur, "\\u%04x\\u%04x", |
278 | sprintf (enc->cur, "\\u%04x\\u%04x", |
218 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
279 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
219 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
280 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
220 | enc->cur += 12; |
281 | enc->cur += 12; |
… | |
… | |
248 | while (--clen); |
309 | while (--clen); |
249 | } |
310 | } |
250 | else |
311 | else |
251 | { |
312 | { |
252 | 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 |
253 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
314 | enc->cur = encode_utf8 (enc->cur, uch); |
254 | ++str; |
315 | ++str; |
255 | } |
316 | } |
256 | } |
317 | } |
257 | } |
318 | } |
258 | } |
319 | } |
259 | |
320 | |
260 | --len; |
321 | --len; |
261 | } |
322 | } |
262 | } |
323 | } |
263 | |
324 | |
264 | inline void |
325 | INLINE void |
265 | encode_indent (enc_t *enc) |
326 | encode_indent (enc_t *enc) |
266 | { |
327 | { |
267 | if (enc->json.flags & F_INDENT) |
328 | if (enc->json.flags & F_INDENT) |
268 | { |
329 | { |
269 | int spaces = enc->indent * INDENT_STEP; |
330 | int spaces = enc->indent * INDENT_STEP; |
… | |
… | |
272 | memset (enc->cur, ' ', spaces); |
333 | memset (enc->cur, ' ', spaces); |
273 | enc->cur += spaces; |
334 | enc->cur += spaces; |
274 | } |
335 | } |
275 | } |
336 | } |
276 | |
337 | |
277 | inline void |
338 | INLINE void |
278 | encode_space (enc_t *enc) |
339 | encode_space (enc_t *enc) |
279 | { |
340 | { |
280 | need (enc, 1); |
341 | need (enc, 1); |
281 | encode_ch (enc, ' '); |
342 | encode_ch (enc, ' '); |
282 | } |
343 | } |
283 | |
344 | |
284 | inline void |
345 | INLINE void |
285 | encode_nl (enc_t *enc) |
346 | encode_nl (enc_t *enc) |
286 | { |
347 | { |
287 | if (enc->json.flags & F_INDENT) |
348 | if (enc->json.flags & F_INDENT) |
288 | { |
349 | { |
289 | need (enc, 1); |
350 | need (enc, 1); |
290 | encode_ch (enc, '\n'); |
351 | encode_ch (enc, '\n'); |
291 | } |
352 | } |
292 | } |
353 | } |
293 | |
354 | |
294 | inline void |
355 | INLINE void |
295 | encode_comma (enc_t *enc) |
356 | encode_comma (enc_t *enc) |
296 | { |
357 | { |
297 | encode_ch (enc, ','); |
358 | encode_ch (enc, ','); |
298 | |
359 | |
299 | if (enc->json.flags & F_INDENT) |
360 | if (enc->json.flags & F_INDENT) |
… | |
… | |
307 | static void |
368 | static void |
308 | encode_av (enc_t *enc, AV *av) |
369 | encode_av (enc_t *enc, AV *av) |
309 | { |
370 | { |
310 | int i, len = av_len (av); |
371 | int i, len = av_len (av); |
311 | |
372 | |
312 | if (enc->indent >= enc->maxdepth) |
373 | if (enc->indent >= enc->json.max_depth) |
313 | croak ("data structure too deep (hit recursion limit)"); |
374 | croak (ERR_NESTING_EXCEEDED); |
314 | |
375 | |
315 | encode_ch (enc, '['); encode_nl (enc); |
376 | encode_ch (enc, '['); |
316 | ++enc->indent; |
377 | |
|
|
378 | if (len >= 0) |
|
|
379 | { |
|
|
380 | encode_nl (enc); ++enc->indent; |
317 | |
381 | |
318 | for (i = 0; i <= len; ++i) |
382 | for (i = 0; i <= len; ++i) |
319 | { |
383 | { |
320 | SV **svp = av_fetch (av, i, 0); |
384 | SV **svp = av_fetch (av, i, 0); |
321 | |
385 | |
322 | encode_indent (enc); |
386 | encode_indent (enc); |
323 | |
387 | |
324 | if (svp) |
388 | if (svp) |
325 | encode_sv (enc, *svp); |
389 | encode_sv (enc, *svp); |
326 | else |
390 | else |
327 | encode_str (enc, "null", 4, 0); |
391 | encode_str (enc, "null", 4, 0); |
328 | |
392 | |
329 | if (i < len) |
393 | if (i < len) |
330 | encode_comma (enc); |
394 | encode_comma (enc); |
331 | } |
395 | } |
332 | |
396 | |
|
|
397 | encode_nl (enc); --enc->indent; encode_indent (enc); |
|
|
398 | } |
|
|
399 | |
333 | encode_nl (enc); |
400 | encode_ch (enc, ']'); |
334 | |
|
|
335 | --enc->indent; |
|
|
336 | encode_indent (enc); encode_ch (enc, ']'); |
|
|
337 | } |
401 | } |
338 | |
402 | |
339 | static void |
403 | static void |
340 | encode_hk (enc_t *enc, HE *he) |
404 | encode_hk (enc_t *enc, HE *he) |
341 | { |
405 | { |
… | |
… | |
389 | |
453 | |
390 | static void |
454 | static void |
391 | encode_hv (enc_t *enc, HV *hv) |
455 | encode_hv (enc_t *enc, HV *hv) |
392 | { |
456 | { |
393 | HE *he; |
457 | HE *he; |
394 | int count; |
|
|
395 | |
458 | |
396 | if (enc->indent >= enc->maxdepth) |
459 | if (enc->indent >= enc->json.max_depth) |
397 | croak ("data structure too deep (hit recursion limit)"); |
460 | croak (ERR_NESTING_EXCEEDED); |
398 | |
461 | |
399 | encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; |
462 | encode_ch (enc, '{'); |
400 | |
463 | |
401 | // for canonical output we have to sort by keys first |
464 | // for canonical output we have to sort by keys first |
402 | // actually, this is mostly due to the stupid so-called |
465 | // actually, this is mostly due to the stupid so-called |
403 | // security workaround added somewhere in 5.8.x. |
466 | // security workaround added somewhere in 5.8.x. |
404 | // that randomises hash orderings |
467 | // that randomises hash orderings |
… | |
… | |
457 | |
520 | |
458 | FREETMPS; |
521 | FREETMPS; |
459 | LEAVE; |
522 | LEAVE; |
460 | } |
523 | } |
461 | |
524 | |
|
|
525 | encode_nl (enc); ++enc->indent; |
|
|
526 | |
462 | while (count--) |
527 | while (count--) |
463 | { |
528 | { |
464 | encode_indent (enc); |
529 | encode_indent (enc); |
465 | he = hes [count]; |
530 | he = hes [count]; |
466 | encode_hk (enc, he); |
531 | encode_hk (enc, he); |
467 | encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
532 | encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
468 | |
533 | |
469 | if (count) |
534 | if (count) |
470 | encode_comma (enc); |
535 | encode_comma (enc); |
471 | } |
536 | } |
|
|
537 | |
|
|
538 | encode_nl (enc); --enc->indent; encode_indent (enc); |
472 | } |
539 | } |
473 | } |
540 | } |
474 | else |
541 | else |
475 | { |
542 | { |
476 | if (hv_iterinit (hv) || SvMAGICAL (hv)) |
543 | if (hv_iterinit (hv) || SvMAGICAL (hv)) |
477 | if ((he = hv_iternext (hv))) |
544 | if ((he = hv_iternext (hv))) |
|
|
545 | { |
|
|
546 | encode_nl (enc); ++enc->indent; |
|
|
547 | |
478 | for (;;) |
548 | for (;;) |
479 | { |
549 | { |
480 | encode_indent (enc); |
550 | encode_indent (enc); |
481 | encode_hk (enc, he); |
551 | encode_hk (enc, he); |
482 | encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
552 | encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
483 | |
553 | |
484 | if (!(he = hv_iternext (hv))) |
554 | if (!(he = hv_iternext (hv))) |
485 | break; |
555 | break; |
486 | |
556 | |
487 | encode_comma (enc); |
557 | encode_comma (enc); |
488 | } |
558 | } |
489 | } |
|
|
490 | |
559 | |
|
|
560 | encode_nl (enc); --enc->indent; encode_indent (enc); |
|
|
561 | } |
|
|
562 | } |
|
|
563 | |
491 | encode_nl (enc); |
564 | encode_ch (enc, '}'); |
492 | |
|
|
493 | --enc->indent; encode_indent (enc); encode_ch (enc, '}'); |
|
|
494 | } |
565 | } |
495 | |
566 | |
496 | // encode objects, arrays and special \0=false and \1=true values. |
567 | // encode objects, arrays and special \0=false and \1=true values. |
497 | static void |
568 | static void |
498 | encode_rv (enc_t *enc, SV *sv) |
569 | encode_rv (enc_t *enc, SV *sv) |
… | |
… | |
575 | |
646 | |
576 | if (len == 1 && *pv == '1') |
647 | if (len == 1 && *pv == '1') |
577 | encode_str (enc, "true", 4, 0); |
648 | encode_str (enc, "true", 4, 0); |
578 | else if (len == 1 && *pv == '0') |
649 | else if (len == 1 && *pv == '0') |
579 | 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); |
580 | else |
653 | else |
581 | 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", |
582 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
655 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
583 | } |
656 | } |
|
|
657 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
658 | encode_str (enc, "null", 4, 0); |
584 | else |
659 | else |
585 | 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", |
586 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
661 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
587 | } |
662 | } |
588 | |
663 | |
… | |
… | |
606 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
681 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
607 | enc->cur += strlen (enc->cur); |
682 | enc->cur += strlen (enc->cur); |
608 | } |
683 | } |
609 | else if (SvIOKp (sv)) |
684 | else if (SvIOKp (sv)) |
610 | { |
685 | { |
611 | // 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 |
612 | if (SvUV (sv) & ~(UV)0x7fff) |
687 | // we assume two's complement |
613 | { |
688 | // we assume no aliasing issues in the union |
614 | // large integer, use the (rather slow) snprintf way. |
689 | if (SvIsUV (sv) ? SvUVX (sv) <= 59000 |
615 | need (enc, sizeof (UV) * 3); |
690 | : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000) |
616 | enc->cur += |
|
|
617 | SvIsUV(sv) |
|
|
618 | ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv)) |
|
|
619 | : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv)); |
|
|
620 | } |
|
|
621 | else |
|
|
622 | { |
691 | { |
623 | // optimise the "small number case" |
692 | // optimise the "small number case" |
624 | // 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 |
625 | I32 i = SvIV (sv); |
695 | I32 i = SvIVX (sv); |
626 | U32 u; |
696 | U32 u; |
627 | char digit, nz = 0; |
697 | char digit, nz = 0; |
628 | |
698 | |
629 | need (enc, 6); |
699 | need (enc, 6); |
630 | |
700 | |
… | |
… | |
636 | |
706 | |
637 | // 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 |
638 | // 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, |
639 | // resulting in a net multiplication by 10. |
709 | // resulting in a net multiplication by 10. |
640 | // we always write the digit to memory but conditionally increment |
710 | // we always write the digit to memory but conditionally increment |
641 | // the pointer, to ease the usage of conditional move instructions. |
711 | // the pointer, to enable the use of conditional move instructions. |
642 | 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; |
643 | 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; |
644 | 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; |
645 | 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; |
646 | 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' |
647 | } |
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 | } |
648 | } |
727 | } |
649 | else if (SvROK (sv)) |
728 | else if (SvROK (sv)) |
650 | encode_rv (enc, SvRV (sv)); |
729 | encode_rv (enc, SvRV (sv)); |
651 | else if (!SvOK (sv)) |
730 | else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) |
652 | encode_str (enc, "null", 4, 0); |
731 | encode_str (enc, "null", 4, 0); |
653 | else |
732 | else |
654 | 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", |
655 | SvPV_nolen (sv), SvFLAGS (sv)); |
734 | SvPV_nolen (sv), SvFLAGS (sv)); |
656 | } |
735 | } |
… | |
… | |
666 | enc.json = *json; |
745 | enc.json = *json; |
667 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
746 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
668 | enc.cur = SvPVX (enc.sv); |
747 | enc.cur = SvPVX (enc.sv); |
669 | enc.end = SvEND (enc.sv); |
748 | enc.end = SvEND (enc.sv); |
670 | enc.indent = 0; |
749 | enc.indent = 0; |
671 | 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; |
672 | |
753 | |
673 | SvPOK_only (enc.sv); |
754 | SvPOK_only (enc.sv); |
674 | encode_sv (&enc, scalar); |
755 | encode_sv (&enc, scalar); |
675 | |
756 | |
676 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
757 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
… | |
… | |
697 | JSON json; |
778 | JSON json; |
698 | U32 depth; // recursion depth |
779 | U32 depth; // recursion depth |
699 | U32 maxdepth; // recursion depth limit |
780 | U32 maxdepth; // recursion depth limit |
700 | } dec_t; |
781 | } dec_t; |
701 | |
782 | |
702 | inline void |
783 | INLINE void |
|
|
784 | decode_comment (dec_t *dec) |
|
|
785 | { |
|
|
786 | // only '#'-style comments allowed a.t.m. |
|
|
787 | |
|
|
788 | while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) |
|
|
789 | ++dec->cur; |
|
|
790 | } |
|
|
791 | |
|
|
792 | INLINE void |
703 | decode_ws (dec_t *dec) |
793 | decode_ws (dec_t *dec) |
704 | { |
794 | { |
705 | for (;;) |
795 | for (;;) |
706 | { |
796 | { |
707 | char ch = *dec->cur; |
797 | char ch = *dec->cur; |
708 | |
798 | |
709 | if (ch > 0x20 |
799 | if (ch > 0x20) |
|
|
800 | { |
|
|
801 | if (expect_false (ch == '#')) |
|
|
802 | { |
|
|
803 | if (dec->json.flags & F_RELAXED) |
|
|
804 | decode_comment (dec); |
|
|
805 | else |
|
|
806 | break; |
|
|
807 | } |
|
|
808 | else |
|
|
809 | break; |
|
|
810 | } |
710 | || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) |
811 | else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09) |
711 | break; |
812 | break; // parse error, but let higher level handle it, gives better error messages |
712 | |
813 | |
713 | ++dec->cur; |
814 | ++dec->cur; |
714 | } |
815 | } |
715 | } |
816 | } |
716 | |
817 | |
… | |
… | |
720 | if (*dec->cur != ch) \ |
821 | if (*dec->cur != ch) \ |
721 | ERR (# ch " expected"); \ |
822 | ERR (# ch " expected"); \ |
722 | ++dec->cur; \ |
823 | ++dec->cur; \ |
723 | SE |
824 | SE |
724 | |
825 | |
725 | #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) |
726 | #define DEC_DEC_DEPTH --dec->depth |
827 | #define DEC_DEC_DEPTH --dec->depth |
727 | |
828 | |
728 | static SV *decode_sv (dec_t *dec); |
829 | static SV *decode_sv (dec_t *dec); |
729 | |
830 | |
730 | static signed char decode_hexdigit[256]; |
831 | static signed char decode_hexdigit[256]; |
… | |
… | |
822 | |
923 | |
823 | if (hi >= 0x80) |
924 | if (hi >= 0x80) |
824 | { |
925 | { |
825 | utf8 = 1; |
926 | utf8 = 1; |
826 | |
927 | |
827 | cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); |
928 | cur = encode_utf8 (cur, hi); |
828 | } |
929 | } |
829 | else |
930 | else |
830 | *cur++ = hi; |
931 | *cur++ = hi; |
831 | } |
932 | } |
832 | break; |
933 | break; |
… | |
… | |
834 | default: |
935 | default: |
835 | --dec_cur; |
936 | --dec_cur; |
836 | ERR ("illegal backslash escape sequence in string"); |
937 | ERR ("illegal backslash escape sequence in string"); |
837 | } |
938 | } |
838 | } |
939 | } |
839 | else if (expect_true (ch >= 0x20 && ch <= 0x7f)) |
940 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
840 | *cur++ = ch; |
941 | *cur++ = ch; |
841 | else if (ch >= 0x80) |
942 | else if (ch >= 0x80) |
842 | { |
943 | { |
843 | STRLEN clen; |
944 | STRLEN clen; |
844 | UV uch; |
945 | UV uch; |
… | |
… | |
967 | |
1068 | |
968 | if (!is_nv) |
1069 | if (!is_nv) |
969 | { |
1070 | { |
970 | int len = dec->cur - start; |
1071 | int len = dec->cur - start; |
971 | |
1072 | |
972 | // 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 |
973 | if (*start == '-') |
1074 | if (*start == '-') |
974 | switch (len) |
1075 | switch (len) |
975 | { |
1076 | { |
976 | case 2: return newSViv (-( start [1] - '0' * 1)); |
1077 | case 2: return newSViv (-( start [1] - '0' * 1)); |
977 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
1078 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
978 | 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)); |
979 | 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)); |
980 | } |
1082 | } |
981 | else |
1083 | else |
982 | switch (len) |
1084 | switch (len) |
983 | { |
1085 | { |
984 | case 1: return newSViv ( start [0] - '0' * 1); |
1086 | case 1: return newSViv ( start [0] - '0' * 1); |
985 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1087 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
986 | 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); |
987 | 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); |
988 | } |
1091 | } |
989 | |
1092 | |
990 | { |
1093 | { |
991 | UV uv; |
1094 | UV uv; |
992 | int numtype = grok_number (start, len, &uv); |
1095 | int numtype = grok_number (start, len, &uv); |
… | |
… | |
1053 | |
1156 | |
1054 | if (*dec->cur != ',') |
1157 | if (*dec->cur != ',') |
1055 | ERR (", or ] expected while parsing array"); |
1158 | ERR (", or ] expected while parsing array"); |
1056 | |
1159 | |
1057 | ++dec->cur; |
1160 | ++dec->cur; |
|
|
1161 | |
|
|
1162 | decode_ws (dec); |
|
|
1163 | |
|
|
1164 | if (*dec->cur == ']' && dec->json.flags & F_RELAXED) |
|
|
1165 | { |
|
|
1166 | ++dec->cur; |
|
|
1167 | break; |
|
|
1168 | } |
1058 | } |
1169 | } |
1059 | |
1170 | |
1060 | DEC_DEC_DEPTH; |
1171 | DEC_DEC_DEPTH; |
1061 | return newRV_noinc ((SV *)av); |
1172 | return newRV_noinc ((SV *)av); |
1062 | |
1173 | |
… | |
… | |
1078 | if (*dec->cur == '}') |
1189 | if (*dec->cur == '}') |
1079 | ++dec->cur; |
1190 | ++dec->cur; |
1080 | else |
1191 | else |
1081 | for (;;) |
1192 | for (;;) |
1082 | { |
1193 | { |
1083 | decode_ws (dec); EXPECT_CH ('"'); |
1194 | EXPECT_CH ('"'); |
1084 | |
1195 | |
1085 | // heuristic: assume that |
1196 | // heuristic: assume that |
1086 | // a) decode_str + hv_store_ent are abysmally slow. |
1197 | // a) decode_str + hv_store_ent are abysmally slow. |
1087 | // b) most hash keys are short, simple ascii text. |
1198 | // b) most hash keys are short, simple ascii text. |
1088 | // => try to "fast-match" such strings to avoid |
1199 | // => try to "fast-match" such strings to avoid |
… | |
… | |
1092 | char *p = dec->cur; |
1203 | char *p = dec->cur; |
1093 | char *e = p + 24; // only try up to 24 bytes |
1204 | char *e = p + 24; // only try up to 24 bytes |
1094 | |
1205 | |
1095 | for (;;) |
1206 | for (;;) |
1096 | { |
1207 | { |
1097 | // the >= 0x80 is true on most architectures |
1208 | // the >= 0x80 is false on most architectures |
1098 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1209 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1099 | { |
1210 | { |
1100 | // slow path, back up and use decode_str |
1211 | // slow path, back up and use decode_str |
1101 | SV *key = decode_str (dec); |
1212 | SV *key = decode_str (dec); |
1102 | if (!key) |
1213 | if (!key) |
1103 | goto fail; |
1214 | goto fail; |
1104 | |
1215 | |
1105 | decode_ws (dec); EXPECT_CH (':'); |
1216 | decode_ws (dec); EXPECT_CH (':'); |
1106 | |
1217 | |
|
|
1218 | decode_ws (dec); |
1107 | value = decode_sv (dec); |
1219 | value = decode_sv (dec); |
1108 | if (!value) |
1220 | if (!value) |
1109 | { |
1221 | { |
1110 | SvREFCNT_dec (key); |
1222 | SvREFCNT_dec (key); |
1111 | goto fail; |
1223 | goto fail; |
… | |
… | |
1123 | int len = p - key; |
1235 | int len = p - key; |
1124 | dec->cur = p + 1; |
1236 | dec->cur = p + 1; |
1125 | |
1237 | |
1126 | decode_ws (dec); EXPECT_CH (':'); |
1238 | decode_ws (dec); EXPECT_CH (':'); |
1127 | |
1239 | |
|
|
1240 | decode_ws (dec); |
1128 | value = decode_sv (dec); |
1241 | value = decode_sv (dec); |
1129 | if (!value) |
1242 | if (!value) |
1130 | goto fail; |
1243 | goto fail; |
1131 | |
1244 | |
1132 | hv_store (hv, key, len, value, 0); |
1245 | hv_store (hv, key, len, value, 0); |
… | |
… | |
1148 | |
1261 | |
1149 | if (*dec->cur != ',') |
1262 | if (*dec->cur != ',') |
1150 | ERR (", or } expected while parsing object/hash"); |
1263 | ERR (", or } expected while parsing object/hash"); |
1151 | |
1264 | |
1152 | ++dec->cur; |
1265 | ++dec->cur; |
|
|
1266 | |
|
|
1267 | decode_ws (dec); |
|
|
1268 | |
|
|
1269 | if (*dec->cur == '}' && dec->json.flags & F_RELAXED) |
|
|
1270 | { |
|
|
1271 | ++dec->cur; |
|
|
1272 | break; |
|
|
1273 | } |
1153 | } |
1274 | } |
1154 | |
1275 | |
1155 | DEC_DEC_DEPTH; |
1276 | DEC_DEC_DEPTH; |
1156 | sv = newRV_noinc ((SV *)hv); |
1277 | sv = newRV_noinc ((SV *)hv); |
1157 | |
1278 | |
… | |
… | |
1222 | } |
1343 | } |
1223 | |
1344 | |
1224 | static SV * |
1345 | static SV * |
1225 | decode_sv (dec_t *dec) |
1346 | decode_sv (dec_t *dec) |
1226 | { |
1347 | { |
1227 | decode_ws (dec); |
|
|
1228 | |
|
|
1229 | // the beauty of JSON: you need exactly one character lookahead |
1348 | // the beauty of JSON: you need exactly one character lookahead |
1230 | // to parse anything. |
1349 | // to parse everything. |
1231 | switch (*dec->cur) |
1350 | switch (*dec->cur) |
1232 | { |
1351 | { |
1233 | case '"': ++dec->cur; return decode_str (dec); |
1352 | case '"': ++dec->cur; return decode_str (dec); |
1234 | case '[': ++dec->cur; return decode_av (dec); |
1353 | case '[': ++dec->cur; return decode_av (dec); |
1235 | case '{': ++dec->cur; return decode_hv (dec); |
1354 | case '{': ++dec->cur; return decode_hv (dec); |
1236 | |
1355 | |
1237 | case '-': |
1356 | case '-': |
1238 | case '0': case '1': case '2': case '3': case '4': |
1357 | case '0': case '1': case '2': case '3': case '4': |
1239 | case '5': case '6': case '7': case '8': case '9': |
1358 | case '5': case '6': case '7': case '8': case '9': |
1240 | return decode_num (dec); |
1359 | return decode_num (dec); |
… | |
… | |
1242 | case 't': |
1361 | case 't': |
1243 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1362 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1244 | { |
1363 | { |
1245 | dec->cur += 4; |
1364 | dec->cur += 4; |
1246 | #if JSON_SLOW |
1365 | #if JSON_SLOW |
1247 | json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); |
1366 | json_true = get_bool ("JSON::XS::true"); |
1248 | #endif |
1367 | #endif |
1249 | return SvREFCNT_inc (json_true); |
1368 | return newSVsv (json_true); |
1250 | } |
1369 | } |
1251 | else |
1370 | else |
1252 | ERR ("'true' expected"); |
1371 | ERR ("'true' expected"); |
1253 | |
1372 | |
1254 | break; |
1373 | break; |
… | |
… | |
1256 | case 'f': |
1375 | case 'f': |
1257 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1376 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1258 | { |
1377 | { |
1259 | dec->cur += 5; |
1378 | dec->cur += 5; |
1260 | #if JSON_SLOW |
1379 | #if JSON_SLOW |
1261 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1380 | json_false = get_bool ("JSON::XS::false"); |
1262 | #endif |
1381 | #endif |
1263 | return SvREFCNT_inc (json_false); |
1382 | return newSVsv (json_false); |
1264 | } |
1383 | } |
1265 | else |
1384 | else |
1266 | ERR ("'false' expected"); |
1385 | ERR ("'false' expected"); |
1267 | |
1386 | |
1268 | break; |
1387 | break; |
… | |
… | |
1286 | fail: |
1405 | fail: |
1287 | return 0; |
1406 | return 0; |
1288 | } |
1407 | } |
1289 | |
1408 | |
1290 | static SV * |
1409 | static SV * |
1291 | decode_json (SV *string, JSON *json, UV *offset_return) |
1410 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1292 | { |
1411 | { |
1293 | dec_t dec; |
1412 | dec_t dec; |
1294 | UV offset; |
1413 | STRLEN offset; |
1295 | SV *sv; |
1414 | SV *sv; |
1296 | |
1415 | |
1297 | SvGETMAGIC (string); |
1416 | SvGETMAGIC (string); |
1298 | SvUPGRADE (string, SVt_PV); |
1417 | SvUPGRADE (string, SVt_PV); |
1299 | |
1418 | |
1300 | 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 | */ |
|
|
1424 | SvPOK (string); |
|
|
1425 | |
|
|
1426 | if (SvCUR (string) > json->max_size && json->max_size) |
1301 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1427 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1302 | (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); |
1428 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
1303 | |
1429 | |
1304 | if (json->flags & F_UTF8) |
1430 | if (json->flags & F_UTF8) |
1305 | sv_utf8_downgrade (string, 0); |
1431 | sv_utf8_downgrade (string, 0); |
1306 | else |
1432 | else |
1307 | sv_utf8_upgrade (string); |
1433 | sv_utf8_upgrade (string); |
1308 | |
1434 | |
1309 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1435 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1310 | |
1436 | |
1311 | dec.json = *json; |
1437 | dec.json = *json; |
1312 | dec.cur = SvPVX (string); |
1438 | dec.cur = SvPVX (string); |
1313 | dec.end = SvEND (string); |
1439 | dec.end = SvEND (string); |
1314 | dec.err = 0; |
1440 | dec.err = 0; |
1315 | dec.depth = 0; |
1441 | dec.depth = 0; |
1316 | dec.maxdepth = DEC_DEPTH (dec.json.flags); |
|
|
1317 | |
1442 | |
1318 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1443 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1319 | dec.json.flags |= F_HOOK; |
1444 | dec.json.flags |= F_HOOK; |
1320 | |
1445 | |
1321 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1446 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
|
|
1447 | |
|
|
1448 | decode_ws (&dec); |
1322 | sv = decode_sv (&dec); |
1449 | sv = decode_sv (&dec); |
1323 | |
1450 | |
1324 | if (!(offset_return || !sv)) |
1451 | if (!(offset_return || !sv)) |
1325 | { |
1452 | { |
1326 | // check for trailing garbage |
1453 | // check for trailing garbage |
… | |
… | |
1370 | |
1497 | |
1371 | return sv; |
1498 | return sv; |
1372 | } |
1499 | } |
1373 | |
1500 | |
1374 | ///////////////////////////////////////////////////////////////////////////// |
1501 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1502 | // incremental parser |
|
|
1503 | |
|
|
1504 | static void |
|
|
1505 | incr_parse (JSON *self) |
|
|
1506 | { |
|
|
1507 | const char *p = SvPVX (self->incr_text) + self->incr_pos; |
|
|
1508 | |
|
|
1509 | for (;;) |
|
|
1510 | { |
|
|
1511 | //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 |
|
|
1512 | switch (self->incr_mode) |
|
|
1513 | { |
|
|
1514 | // only used for intiial whitespace skipping |
|
|
1515 | case INCR_M_WS: |
|
|
1516 | for (;;) |
|
|
1517 | { |
|
|
1518 | if (*p > 0x20) |
|
|
1519 | { |
|
|
1520 | self->incr_mode = INCR_M_JSON; |
|
|
1521 | goto incr_m_json; |
|
|
1522 | } |
|
|
1523 | else if (!*p) |
|
|
1524 | goto interrupt; |
|
|
1525 | |
|
|
1526 | ++p; |
|
|
1527 | } |
|
|
1528 | |
|
|
1529 | // skip a single char inside a string (for \\-processing) |
|
|
1530 | case INCR_M_BS: |
|
|
1531 | if (!*p) |
|
|
1532 | goto interrupt; |
|
|
1533 | |
|
|
1534 | ++p; |
|
|
1535 | self->incr_mode = INCR_M_STR; |
|
|
1536 | goto incr_m_str; |
|
|
1537 | |
|
|
1538 | // inside a string |
|
|
1539 | case INCR_M_STR: |
|
|
1540 | incr_m_str: |
|
|
1541 | for (;;) |
|
|
1542 | { |
|
|
1543 | if (*p == '"') |
|
|
1544 | { |
|
|
1545 | ++p; |
|
|
1546 | self->incr_mode = INCR_M_JSON; |
|
|
1547 | |
|
|
1548 | if (!self->incr_nest) |
|
|
1549 | goto interrupt; |
|
|
1550 | |
|
|
1551 | goto incr_m_json; |
|
|
1552 | } |
|
|
1553 | else if (*p == '\\') |
|
|
1554 | { |
|
|
1555 | ++p; // "virtually" consumes character after \ |
|
|
1556 | |
|
|
1557 | if (!*p) // if at end of string we have to switch modes |
|
|
1558 | { |
|
|
1559 | self->incr_mode = INCR_M_BS; |
|
|
1560 | goto interrupt; |
|
|
1561 | } |
|
|
1562 | } |
|
|
1563 | else if (!*p) |
|
|
1564 | goto interrupt; |
|
|
1565 | |
|
|
1566 | ++p; |
|
|
1567 | } |
|
|
1568 | |
|
|
1569 | // after initial ws, outside string |
|
|
1570 | case INCR_M_JSON: |
|
|
1571 | incr_m_json: |
|
|
1572 | for (;;) |
|
|
1573 | { |
|
|
1574 | switch (*p++) |
|
|
1575 | { |
|
|
1576 | case 0: |
|
|
1577 | --p; |
|
|
1578 | goto interrupt; |
|
|
1579 | |
|
|
1580 | case 0x09: |
|
|
1581 | case 0x0a: |
|
|
1582 | case 0x0d: |
|
|
1583 | case 0x20: |
|
|
1584 | if (!self->incr_nest) |
|
|
1585 | { |
|
|
1586 | --p; // do not eat the whitespace, let the next round do it |
|
|
1587 | goto interrupt; |
|
|
1588 | } |
|
|
1589 | break; |
|
|
1590 | |
|
|
1591 | case '"': |
|
|
1592 | self->incr_mode = INCR_M_STR; |
|
|
1593 | goto incr_m_str; |
|
|
1594 | |
|
|
1595 | case '[': |
|
|
1596 | case '{': |
|
|
1597 | if (++self->incr_nest > self->max_depth) |
|
|
1598 | croak (ERR_NESTING_EXCEEDED); |
|
|
1599 | break; |
|
|
1600 | |
|
|
1601 | case ']': |
|
|
1602 | case '}': |
|
|
1603 | if (--self->incr_nest <= 0) |
|
|
1604 | goto interrupt; |
|
|
1605 | } |
|
|
1606 | } |
|
|
1607 | } |
|
|
1608 | |
|
|
1609 | modechange: |
|
|
1610 | ; |
|
|
1611 | } |
|
|
1612 | |
|
|
1613 | interrupt: |
|
|
1614 | self->incr_pos = p - SvPVX (self->incr_text); |
|
|
1615 | //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D |
|
|
1616 | } |
|
|
1617 | |
|
|
1618 | ///////////////////////////////////////////////////////////////////////////// |
1375 | // XS interface functions |
1619 | // XS interface functions |
1376 | |
1620 | |
1377 | MODULE = JSON::XS PACKAGE = JSON::XS |
1621 | MODULE = JSON::XS PACKAGE = JSON::XS |
1378 | |
1622 | |
1379 | BOOT: |
1623 | BOOT: |
… | |
… | |
1388 | : -1; |
1632 | : -1; |
1389 | |
1633 | |
1390 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1634 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1391 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
1635 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
1392 | |
1636 | |
1393 | json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); |
1637 | json_true = get_bool ("JSON::XS::true"); |
1394 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1638 | json_false = get_bool ("JSON::XS::false"); |
1395 | } |
1639 | } |
1396 | |
1640 | |
1397 | PROTOTYPES: DISABLE |
1641 | PROTOTYPES: DISABLE |
1398 | |
1642 | |
1399 | void CLONE (...) |
1643 | void CLONE (...) |
… | |
… | |
1404 | void new (char *klass) |
1648 | void new (char *klass) |
1405 | PPCODE: |
1649 | PPCODE: |
1406 | { |
1650 | { |
1407 | SV *pv = NEWSV (0, sizeof (JSON)); |
1651 | SV *pv = NEWSV (0, sizeof (JSON)); |
1408 | SvPOK_only (pv); |
1652 | SvPOK_only (pv); |
1409 | Zero (SvPVX (pv), 1, JSON); |
1653 | json_init ((JSON *)SvPVX (pv)); |
1410 | ((JSON *)SvPVX (pv))->flags = F_DEFAULT; |
1654 | XPUSHs (sv_2mortal (sv_bless ( |
1411 | XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH))); |
1655 | newRV_noinc (pv), |
|
|
1656 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
|
|
1657 | ))); |
1412 | } |
1658 | } |
1413 | |
1659 | |
1414 | void ascii (JSON *self, int enable = 1) |
1660 | void ascii (JSON *self, int enable = 1) |
1415 | ALIAS: |
1661 | ALIAS: |
1416 | ascii = F_ASCII |
1662 | ascii = F_ASCII |
… | |
… | |
1423 | pretty = F_PRETTY |
1669 | pretty = F_PRETTY |
1424 | allow_nonref = F_ALLOW_NONREF |
1670 | allow_nonref = F_ALLOW_NONREF |
1425 | shrink = F_SHRINK |
1671 | shrink = F_SHRINK |
1426 | allow_blessed = F_ALLOW_BLESSED |
1672 | allow_blessed = F_ALLOW_BLESSED |
1427 | convert_blessed = F_CONV_BLESSED |
1673 | convert_blessed = F_CONV_BLESSED |
|
|
1674 | relaxed = F_RELAXED |
|
|
1675 | allow_unknown = F_ALLOW_UNKNOWN |
1428 | PPCODE: |
1676 | PPCODE: |
1429 | { |
1677 | { |
1430 | if (enable) |
1678 | if (enable) |
1431 | self->flags |= ix; |
1679 | self->flags |= ix; |
1432 | else |
1680 | else |
1433 | self->flags &= ~ix; |
1681 | self->flags &= ~ix; |
1434 | |
1682 | |
1435 | XPUSHs (ST (0)); |
1683 | XPUSHs (ST (0)); |
1436 | } |
1684 | } |
1437 | |
1685 | |
1438 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
1686 | void get_ascii (JSON *self) |
|
|
1687 | ALIAS: |
|
|
1688 | get_ascii = F_ASCII |
|
|
1689 | get_latin1 = F_LATIN1 |
|
|
1690 | get_utf8 = F_UTF8 |
|
|
1691 | get_indent = F_INDENT |
|
|
1692 | get_canonical = F_CANONICAL |
|
|
1693 | get_space_before = F_SPACE_BEFORE |
|
|
1694 | get_space_after = F_SPACE_AFTER |
|
|
1695 | get_allow_nonref = F_ALLOW_NONREF |
|
|
1696 | get_shrink = F_SHRINK |
|
|
1697 | get_allow_blessed = F_ALLOW_BLESSED |
|
|
1698 | get_convert_blessed = F_CONV_BLESSED |
|
|
1699 | get_relaxed = F_RELAXED |
|
|
1700 | get_allow_unknown = F_ALLOW_UNKNOWN |
1439 | PPCODE: |
1701 | PPCODE: |
1440 | { |
1702 | XPUSHs (boolSV (self->flags & ix)); |
1441 | UV log2 = 0; |
|
|
1442 | |
1703 | |
1443 | if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
1704 | void max_depth (JSON *self, U32 max_depth = 0x80000000UL) |
1444 | |
1705 | PPCODE: |
1445 | while ((1UL << log2) < max_depth) |
1706 | self->max_depth = max_depth; |
1446 | ++log2; |
|
|
1447 | |
|
|
1448 | self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
|
|
1449 | |
|
|
1450 | XPUSHs (ST (0)); |
1707 | XPUSHs (ST (0)); |
1451 | } |
|
|
1452 | |
1708 | |
|
|
1709 | U32 get_max_depth (JSON *self) |
|
|
1710 | CODE: |
|
|
1711 | RETVAL = self->max_depth; |
|
|
1712 | OUTPUT: |
|
|
1713 | RETVAL |
|
|
1714 | |
1453 | void max_size (JSON *self, UV max_size = 0) |
1715 | void max_size (JSON *self, U32 max_size = 0) |
1454 | PPCODE: |
1716 | PPCODE: |
1455 | { |
1717 | self->max_size = max_size; |
1456 | UV log2 = 0; |
|
|
1457 | |
|
|
1458 | if (max_size > 0x80000000UL) max_size = 0x80000000UL; |
|
|
1459 | if (max_size == 1) max_size = 2; |
|
|
1460 | |
|
|
1461 | while ((1UL << log2) < max_size) |
|
|
1462 | ++log2; |
|
|
1463 | |
|
|
1464 | self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); |
|
|
1465 | |
|
|
1466 | XPUSHs (ST (0)); |
1718 | XPUSHs (ST (0)); |
1467 | } |
1719 | |
|
|
1720 | int get_max_size (JSON *self) |
|
|
1721 | CODE: |
|
|
1722 | RETVAL = self->max_size; |
|
|
1723 | OUTPUT: |
|
|
1724 | RETVAL |
1468 | |
1725 | |
1469 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1726 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1470 | PPCODE: |
1727 | PPCODE: |
1471 | { |
1728 | { |
1472 | SvREFCNT_dec (self->cb_object); |
1729 | SvREFCNT_dec (self->cb_object); |
… | |
… | |
1506 | XPUSHs (decode_json (jsonstr, self, 0)); |
1763 | XPUSHs (decode_json (jsonstr, self, 0)); |
1507 | |
1764 | |
1508 | void decode_prefix (JSON *self, SV *jsonstr) |
1765 | void decode_prefix (JSON *self, SV *jsonstr) |
1509 | PPCODE: |
1766 | PPCODE: |
1510 | { |
1767 | { |
1511 | UV offset; |
1768 | STRLEN offset; |
1512 | EXTEND (SP, 2); |
1769 | EXTEND (SP, 2); |
1513 | PUSHs (decode_json (jsonstr, self, &offset)); |
1770 | PUSHs (decode_json (jsonstr, self, &offset)); |
1514 | PUSHs (sv_2mortal (newSVuv (offset))); |
1771 | PUSHs (sv_2mortal (newSVuv (offset))); |
|
|
1772 | } |
|
|
1773 | |
|
|
1774 | void incr_parse (JSON *self, SV *jsonstr = 0) |
|
|
1775 | PPCODE: |
|
|
1776 | { |
|
|
1777 | if (!self->incr_text) |
|
|
1778 | self->incr_text = newSVpvn ("", 0); |
|
|
1779 | |
|
|
1780 | // append data, if any |
|
|
1781 | if (jsonstr) |
|
|
1782 | { |
|
|
1783 | if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) |
|
|
1784 | { |
|
|
1785 | /* utf-8-ness differs, need to upgrade */ |
|
|
1786 | sv_utf8_upgrade (self->incr_text); |
|
|
1787 | |
|
|
1788 | if (self->incr_pos) |
|
|
1789 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
|
|
1790 | - (U8 *)SvPVX (self->incr_text); |
|
|
1791 | } |
|
|
1792 | |
|
|
1793 | { |
|
|
1794 | STRLEN len; |
|
|
1795 | const char *str = SvPV (jsonstr, len); |
|
|
1796 | SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); |
|
|
1797 | Move (str, SvEND (self->incr_text), len, char); |
|
|
1798 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
|
|
1799 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
|
|
1800 | } |
|
|
1801 | } |
|
|
1802 | |
|
|
1803 | if (GIMME_V != G_VOID) |
|
|
1804 | do |
|
|
1805 | { |
|
|
1806 | STRLEN offset; |
|
|
1807 | |
|
|
1808 | if (!INCR_DONE (self)) |
|
|
1809 | { |
|
|
1810 | incr_parse (self); |
|
|
1811 | |
|
|
1812 | if (self->incr_pos > self->max_size && self->max_size) |
|
|
1813 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
|
|
1814 | (unsigned long)self->incr_pos, (unsigned long)self->max_size); |
|
|
1815 | |
|
|
1816 | if (!INCR_DONE (self)) |
|
|
1817 | break; |
|
|
1818 | } |
|
|
1819 | |
|
|
1820 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
|
|
1821 | |
|
|
1822 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset); |
|
|
1823 | self->incr_pos -= offset; |
|
|
1824 | self->incr_nest = 0; |
|
|
1825 | self->incr_mode = 0; |
|
|
1826 | } |
|
|
1827 | while (GIMME_V == G_ARRAY); |
|
|
1828 | } |
|
|
1829 | |
|
|
1830 | SV *incr_text (JSON *self) |
|
|
1831 | ATTRS: lvalue |
|
|
1832 | CODE: |
|
|
1833 | { |
|
|
1834 | if (self->incr_pos) |
|
|
1835 | croak ("incr_text can not be called when the incremental parser already started parsing"); |
|
|
1836 | |
|
|
1837 | RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef; |
|
|
1838 | } |
|
|
1839 | OUTPUT: |
|
|
1840 | RETVAL |
|
|
1841 | |
|
|
1842 | void incr_skip (JSON *self) |
|
|
1843 | CODE: |
|
|
1844 | { |
|
|
1845 | if (self->incr_pos) |
|
|
1846 | { |
|
|
1847 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos); |
|
|
1848 | self->incr_pos = 0; |
|
|
1849 | self->incr_nest = 0; |
|
|
1850 | self->incr_mode = 0; |
|
|
1851 | } |
|
|
1852 | } |
|
|
1853 | |
|
|
1854 | void incr_reset (JSON *self) |
|
|
1855 | CODE: |
|
|
1856 | { |
|
|
1857 | SvREFCNT_dec (self->incr_text); |
|
|
1858 | self->incr_text = 0; |
|
|
1859 | self->incr_pos = 0; |
|
|
1860 | self->incr_nest = 0; |
|
|
1861 | self->incr_mode = 0; |
1515 | } |
1862 | } |
1516 | |
1863 | |
1517 | void DESTROY (JSON *self) |
1864 | void DESTROY (JSON *self) |
1518 | CODE: |
1865 | CODE: |
1519 | SvREFCNT_dec (self->cb_sk_object); |
1866 | SvREFCNT_dec (self->cb_sk_object); |
1520 | SvREFCNT_dec (self->cb_object); |
1867 | SvREFCNT_dec (self->cb_object); |
|
|
1868 | SvREFCNT_dec (self->incr_text); |
1521 | |
1869 | |
1522 | PROTOTYPES: ENABLE |
1870 | PROTOTYPES: ENABLE |
1523 | |
1871 | |
1524 | void to_json (SV *scalar) |
1872 | void encode_json (SV *scalar) |
|
|
1873 | ALIAS: |
|
|
1874 | to_json_ = 0 |
|
|
1875 | encode_json = F_UTF8 |
1525 | PPCODE: |
1876 | PPCODE: |
1526 | { |
1877 | { |
1527 | JSON json = { F_DEFAULT | F_UTF8 }; |
1878 | JSON json; |
|
|
1879 | json_init (&json); |
|
|
1880 | json.flags |= ix; |
1528 | XPUSHs (encode_json (scalar, &json)); |
1881 | XPUSHs (encode_json (scalar, &json)); |
1529 | } |
1882 | } |
1530 | |
1883 | |
1531 | void from_json (SV *jsonstr) |
1884 | void decode_json (SV *jsonstr) |
|
|
1885 | ALIAS: |
|
|
1886 | from_json_ = 0 |
|
|
1887 | decode_json = F_UTF8 |
1532 | PPCODE: |
1888 | PPCODE: |
1533 | { |
1889 | { |
1534 | JSON json = { F_DEFAULT | F_UTF8 }; |
1890 | JSON json; |
|
|
1891 | json_init (&json); |
|
|
1892 | json.flags |= ix; |
1535 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1893 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1536 | } |
1894 | } |
1537 | |
1895 | |
|
|
1896 | |