… | |
… | |
3 | #include "XSUB.h" |
3 | #include "XSUB.h" |
4 | |
4 | |
5 | #include "assert.h" |
5 | #include "assert.h" |
6 | #include "string.h" |
6 | #include "string.h" |
7 | #include "stdlib.h" |
7 | #include "stdlib.h" |
|
|
8 | #include "stdio.h" |
|
|
9 | |
|
|
10 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
|
|
11 | # define snprintf _snprintf // C compilers have this in stdio.h |
|
|
12 | #endif |
|
|
13 | |
|
|
14 | // some old perls do not have this, try to make it work, no |
|
|
15 | // guarentees, though. |
|
|
16 | #ifndef UTF8_MAXBYTES |
|
|
17 | # define UTF8_MAXBYTES 13 |
|
|
18 | #endif |
8 | |
19 | |
9 | #define F_ASCII 0x00000001UL |
20 | #define F_ASCII 0x00000001UL |
|
|
21 | #define F_LATIN1 0x00000002UL |
10 | #define F_UTF8 0x00000002UL |
22 | #define F_UTF8 0x00000004UL |
11 | #define F_INDENT 0x00000004UL |
23 | #define F_INDENT 0x00000008UL |
12 | #define F_CANONICAL 0x00000008UL |
24 | #define F_CANONICAL 0x00000010UL |
13 | #define F_SPACE_BEFORE 0x00000010UL |
25 | #define F_SPACE_BEFORE 0x00000020UL |
14 | #define F_SPACE_AFTER 0x00000020UL |
26 | #define F_SPACE_AFTER 0x00000040UL |
15 | #define F_ALLOW_NONREF 0x00000080UL |
27 | #define F_ALLOW_NONREF 0x00000100UL |
16 | #define F_SHRINK 0x00000100UL |
28 | #define F_SHRINK 0x00000200UL |
17 | #define F_MAXDEPTH 0xf8000000UL |
29 | #define F_MAXDEPTH 0xf8000000UL |
18 | #define S_MAXDEPTH 27 |
30 | #define S_MAXDEPTH 27 |
19 | |
31 | |
20 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
32 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
21 | |
33 | |
22 | // F_SELFCONVERT? <=> to_json/toJson |
34 | // F_SELFCONVERT? <=> to_json/toJson |
23 | // F_BLESSED? <=> { $__class__$ => } |
35 | // F_BLESSED? <=> { $__class__$ => } |
24 | |
36 | |
25 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
37 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
26 | #define F_DEFAULT (12UL << S_MAXDEPTH) |
38 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
27 | |
39 | |
28 | #define INIT_SIZE 32 // initial scalar size to be allocated |
40 | #define INIT_SIZE 32 // initial scalar size to be allocated |
29 | #define INDENT_STEP 3 // spaces per indentation level |
41 | #define INDENT_STEP 3 // spaces per indentation level |
30 | |
42 | |
31 | #define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character |
|
|
32 | #define SHORT_STRING_LEN 512 // special-case strings of up to this size |
43 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
33 | |
44 | |
34 | #define SB do { |
45 | #define SB do { |
35 | #define SE } while (0) |
46 | #define SE } while (0) |
|
|
47 | |
|
|
48 | #if __GNUC__ >= 3 |
|
|
49 | # define expect(expr,value) __builtin_expect ((expr),(value)) |
|
|
50 | # define inline inline |
|
|
51 | #else |
|
|
52 | # define expect(expr,value) (expr) |
|
|
53 | # define inline static |
|
|
54 | #endif |
|
|
55 | |
|
|
56 | #define expect_false(expr) expect ((expr) != 0, 0) |
|
|
57 | #define expect_true(expr) expect ((expr) != 0, 1) |
36 | |
58 | |
37 | static HV *json_stash; // JSON::XS:: |
59 | static HV *json_stash; // JSON::XS:: |
38 | |
60 | |
39 | ///////////////////////////////////////////////////////////////////////////// |
61 | ///////////////////////////////////////////////////////////////////////////// |
40 | // utility functions |
62 | // utility functions |
… | |
… | |
65 | // decode an utf-8 character and return it, or (UV)-1 in |
87 | // decode an utf-8 character and return it, or (UV)-1 in |
66 | // case of an error. |
88 | // case of an error. |
67 | // we special-case "safe" characters from U+80 .. U+7FF, |
89 | // we special-case "safe" characters from U+80 .. U+7FF, |
68 | // but use the very good perl function to parse anything else. |
90 | // but use the very good perl function to parse anything else. |
69 | // note that we never call this function for a ascii codepoints |
91 | // note that we never call this function for a ascii codepoints |
70 | static UV |
92 | inline UV |
71 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
93 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
72 | { |
94 | { |
73 | if (s[0] > 0xdf || s[0] < 0xc2) |
95 | if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) |
74 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
96 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
75 | else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
97 | else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
76 | { |
98 | { |
77 | *clen = 2; |
99 | *clen = 2; |
78 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
100 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
79 | } |
101 | } |
80 | else |
102 | else |
|
|
103 | { |
|
|
104 | *clen = (STRLEN)-1; |
81 | return (UV)-1; |
105 | return (UV)-1; |
|
|
106 | } |
82 | } |
107 | } |
83 | |
108 | |
84 | ///////////////////////////////////////////////////////////////////////////// |
109 | ///////////////////////////////////////////////////////////////////////////// |
85 | // encoder |
110 | // encoder |
86 | |
111 | |
… | |
… | |
93 | U32 flags; // F_* |
118 | U32 flags; // F_* |
94 | U32 indent; // indentation level |
119 | U32 indent; // indentation level |
95 | U32 maxdepth; // max. indentation/recursion level |
120 | U32 maxdepth; // max. indentation/recursion level |
96 | } enc_t; |
121 | } enc_t; |
97 | |
122 | |
98 | static void |
123 | inline void |
99 | need (enc_t *enc, STRLEN len) |
124 | need (enc_t *enc, STRLEN len) |
100 | { |
125 | { |
101 | if (enc->cur + len >= enc->end) |
126 | if (expect_false (enc->cur + len >= enc->end)) |
102 | { |
127 | { |
103 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
128 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
104 | SvGROW (enc->sv, cur + len + 1); |
129 | SvGROW (enc->sv, cur + len + 1); |
105 | enc->cur = SvPVX (enc->sv) + cur; |
130 | enc->cur = SvPVX (enc->sv) + cur; |
106 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); |
131 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
107 | } |
132 | } |
108 | } |
133 | } |
109 | |
134 | |
110 | static void |
135 | inline void |
111 | encode_ch (enc_t *enc, char ch) |
136 | encode_ch (enc_t *enc, char ch) |
112 | { |
137 | { |
113 | need (enc, 1); |
138 | need (enc, 1); |
114 | *enc->cur++ = ch; |
139 | *enc->cur++ = ch; |
115 | } |
140 | } |
… | |
… | |
123 | |
148 | |
124 | while (str < end) |
149 | while (str < end) |
125 | { |
150 | { |
126 | unsigned char ch = *(unsigned char *)str; |
151 | unsigned char ch = *(unsigned char *)str; |
127 | |
152 | |
128 | if (ch >= 0x20 && ch < 0x80) // most common case |
153 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
129 | { |
154 | { |
130 | if (ch == '"') // but with slow exceptions |
155 | if (expect_false (ch == '"')) // but with slow exceptions |
131 | { |
156 | { |
132 | need (enc, len += 1); |
157 | need (enc, len += 1); |
133 | *enc->cur++ = '\\'; |
158 | *enc->cur++ = '\\'; |
134 | *enc->cur++ = '"'; |
159 | *enc->cur++ = '"'; |
135 | } |
160 | } |
136 | else if (ch == '\\') |
161 | else if (expect_false (ch == '\\')) |
137 | { |
162 | { |
138 | need (enc, len += 1); |
163 | need (enc, len += 1); |
139 | *enc->cur++ = '\\'; |
164 | *enc->cur++ = '\\'; |
140 | *enc->cur++ = '\\'; |
165 | *enc->cur++ = '\\'; |
141 | } |
166 | } |
… | |
… | |
159 | STRLEN clen; |
184 | STRLEN clen; |
160 | UV uch; |
185 | UV uch; |
161 | |
186 | |
162 | if (is_utf8) |
187 | if (is_utf8) |
163 | { |
188 | { |
164 | //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); |
|
|
165 | uch = decode_utf8 (str, end - str, &clen); |
189 | uch = decode_utf8 (str, end - str, &clen); |
166 | if (clen == (STRLEN)-1) |
190 | if (clen == (STRLEN)-1) |
167 | croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
191 | croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
168 | } |
192 | } |
169 | else |
193 | else |
… | |
… | |
173 | } |
197 | } |
174 | |
198 | |
175 | if (uch > 0x10FFFFUL) |
199 | if (uch > 0x10FFFFUL) |
176 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
200 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
177 | |
201 | |
178 | if (uch < 0x80 || enc->flags & F_ASCII) |
202 | if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF)) |
179 | { |
203 | { |
180 | if (uch > 0xFFFFUL) |
204 | if (uch > 0xFFFFUL) |
181 | { |
205 | { |
182 | need (enc, len += 11); |
206 | need (enc, len += 11); |
183 | sprintf (enc->cur, "\\u%04x\\u%04x", |
207 | sprintf (enc->cur, "\\u%04x\\u%04x", |
… | |
… | |
197 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
221 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
198 | } |
222 | } |
199 | |
223 | |
200 | str += clen; |
224 | str += clen; |
201 | } |
225 | } |
|
|
226 | else if (enc->flags & F_LATIN1) |
|
|
227 | { |
|
|
228 | *enc->cur++ = uch; |
|
|
229 | str += clen; |
|
|
230 | } |
202 | else if (is_utf8) |
231 | else if (is_utf8) |
203 | { |
232 | { |
204 | need (enc, len += clen); |
233 | need (enc, len += clen); |
205 | do |
234 | do |
206 | { |
235 | { |
… | |
… | |
208 | } |
237 | } |
209 | while (--clen); |
238 | while (--clen); |
210 | } |
239 | } |
211 | else |
240 | else |
212 | { |
241 | { |
213 | need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed |
242 | need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
214 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
243 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
215 | ++str; |
244 | ++str; |
216 | } |
245 | } |
217 | } |
246 | } |
218 | } |
247 | } |
… | |
… | |
220 | |
249 | |
221 | --len; |
250 | --len; |
222 | } |
251 | } |
223 | } |
252 | } |
224 | |
253 | |
225 | static void |
254 | inline void |
226 | encode_indent (enc_t *enc) |
255 | encode_indent (enc_t *enc) |
227 | { |
256 | { |
228 | if (enc->flags & F_INDENT) |
257 | if (enc->flags & F_INDENT) |
229 | { |
258 | { |
230 | int spaces = enc->indent * INDENT_STEP; |
259 | int spaces = enc->indent * INDENT_STEP; |
… | |
… | |
233 | memset (enc->cur, ' ', spaces); |
262 | memset (enc->cur, ' ', spaces); |
234 | enc->cur += spaces; |
263 | enc->cur += spaces; |
235 | } |
264 | } |
236 | } |
265 | } |
237 | |
266 | |
238 | static void |
267 | inline void |
239 | encode_space (enc_t *enc) |
268 | encode_space (enc_t *enc) |
240 | { |
269 | { |
241 | need (enc, 1); |
270 | need (enc, 1); |
242 | encode_ch (enc, ' '); |
271 | encode_ch (enc, ' '); |
243 | } |
272 | } |
244 | |
273 | |
245 | static void |
274 | inline void |
246 | encode_nl (enc_t *enc) |
275 | encode_nl (enc_t *enc) |
247 | { |
276 | { |
248 | if (enc->flags & F_INDENT) |
277 | if (enc->flags & F_INDENT) |
249 | { |
278 | { |
250 | need (enc, 1); |
279 | need (enc, 1); |
251 | encode_ch (enc, '\n'); |
280 | encode_ch (enc, '\n'); |
252 | } |
281 | } |
253 | } |
282 | } |
254 | |
283 | |
255 | static void |
284 | inline void |
256 | encode_comma (enc_t *enc) |
285 | encode_comma (enc_t *enc) |
257 | { |
286 | { |
258 | encode_ch (enc, ','); |
287 | encode_ch (enc, ','); |
259 | |
288 | |
260 | if (enc->flags & F_INDENT) |
289 | if (enc->flags & F_INDENT) |
… | |
… | |
359 | // actually, this is mostly due to the stupid so-called |
388 | // actually, this is mostly due to the stupid so-called |
360 | // security workaround added somewhere in 5.8.x. |
389 | // security workaround added somewhere in 5.8.x. |
361 | // that randomises hash orderings |
390 | // that randomises hash orderings |
362 | if (enc->flags & F_CANONICAL) |
391 | if (enc->flags & F_CANONICAL) |
363 | { |
392 | { |
364 | HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode |
|
|
365 | int fast = 1; |
393 | int fast = 1; |
|
|
394 | HE *he; |
|
|
395 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
|
|
396 | HE **hes = _alloca (count * sizeof (HE)); |
|
|
397 | #else |
|
|
398 | HE *hes [count]; // if your compiler dies here, you need to enable C99 mode |
|
|
399 | #endif |
366 | |
400 | |
367 | i = 0; |
401 | i = 0; |
368 | while ((he = hv_iternext (hv))) |
402 | while ((he = hv_iternext (hv))) |
369 | { |
403 | { |
370 | hes [i++] = he; |
404 | hes [i++] = he; |
… | |
… | |
405 | |
439 | |
406 | encode_nl (enc); |
440 | encode_nl (enc); |
407 | } |
441 | } |
408 | else |
442 | else |
409 | { |
443 | { |
410 | SV *sv; |
|
|
411 | HE *he = hv_iternext (hv); |
444 | HE *he = hv_iternext (hv); |
412 | |
445 | |
413 | for (;;) |
446 | for (;;) |
414 | { |
447 | { |
415 | encode_indent (enc); |
448 | encode_indent (enc); |
… | |
… | |
430 | |
463 | |
431 | // encode objects, arrays and special \0=false and \1=true values. |
464 | // encode objects, arrays and special \0=false and \1=true values. |
432 | static void |
465 | static void |
433 | encode_rv (enc_t *enc, SV *sv) |
466 | encode_rv (enc_t *enc, SV *sv) |
434 | { |
467 | { |
|
|
468 | svtype svt; |
|
|
469 | |
435 | SvGETMAGIC (sv); |
470 | SvGETMAGIC (sv); |
436 | |
|
|
437 | svtype svt = SvTYPE (sv); |
471 | svt = SvTYPE (sv); |
438 | |
472 | |
439 | if (svt == SVt_PVHV) |
473 | if (svt == SVt_PVHV) |
440 | encode_hv (enc, (HV *)sv); |
474 | encode_hv (enc, (HV *)sv); |
441 | else if (svt == SVt_PVAV) |
475 | else if (svt == SVt_PVAV) |
442 | encode_av (enc, (AV *)sv); |
476 | encode_av (enc, (AV *)sv); |
… | |
… | |
468 | encode_str (enc, str, len, SvUTF8 (sv)); |
502 | encode_str (enc, str, len, SvUTF8 (sv)); |
469 | encode_ch (enc, '"'); |
503 | encode_ch (enc, '"'); |
470 | } |
504 | } |
471 | else if (SvNOKp (sv)) |
505 | else if (SvNOKp (sv)) |
472 | { |
506 | { |
|
|
507 | // trust that perl will do the right thing w.r.t. JSON syntax. |
473 | need (enc, NV_DIG + 32); |
508 | need (enc, NV_DIG + 32); |
474 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
509 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
475 | enc->cur += strlen (enc->cur); |
510 | enc->cur += strlen (enc->cur); |
476 | } |
511 | } |
477 | else if (SvIOKp (sv)) |
512 | else if (SvIOKp (sv)) |
478 | { |
513 | { |
479 | need (enc, 64); |
514 | // we assume we can always read an IV as a UV |
|
|
515 | if (SvUV (sv) & ~(UV)0x7fff) |
|
|
516 | { |
|
|
517 | // large integer, use the (rather slow) snprintf way. |
|
|
518 | need (enc, sizeof (UV) * 3); |
480 | enc->cur += |
519 | enc->cur += |
481 | SvIsUV(sv) |
520 | SvIsUV(sv) |
482 | ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) |
521 | ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv)) |
483 | : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); |
522 | : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv)); |
|
|
523 | } |
|
|
524 | else |
|
|
525 | { |
|
|
526 | // optimise the "small number case" |
|
|
527 | // code will likely be branchless and use only a single multiplication |
|
|
528 | I32 i = SvIV (sv); |
|
|
529 | U32 u; |
|
|
530 | char digit, nz = 0; |
|
|
531 | |
|
|
532 | need (enc, 6); |
|
|
533 | |
|
|
534 | *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0; |
|
|
535 | u = i < 0 ? -i : i; |
|
|
536 | |
|
|
537 | // convert to 4.28 fixed-point representation |
|
|
538 | u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits |
|
|
539 | |
|
|
540 | // now output digit by digit, each time masking out the integer part |
|
|
541 | // and multiplying by 5 while moving the decimal point one to the right, |
|
|
542 | // resulting in a net multiplication by 10. |
|
|
543 | // we always write the digit to memory but conditionally increment |
|
|
544 | // the pointer, to ease the usage of conditional move instructions. |
|
|
545 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; |
|
|
546 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; |
|
|
547 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; |
|
|
548 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; |
|
|
549 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
|
|
550 | } |
484 | } |
551 | } |
485 | else if (SvROK (sv)) |
552 | else if (SvROK (sv)) |
486 | encode_rv (enc, SvRV (sv)); |
553 | encode_rv (enc, SvRV (sv)); |
487 | else if (!SvOK (sv)) |
554 | else if (!SvOK (sv)) |
488 | encode_str (enc, "null", 4, 0); |
555 | encode_str (enc, "null", 4, 0); |
… | |
… | |
492 | } |
559 | } |
493 | |
560 | |
494 | static SV * |
561 | static SV * |
495 | encode_json (SV *scalar, U32 flags) |
562 | encode_json (SV *scalar, U32 flags) |
496 | { |
563 | { |
|
|
564 | enc_t enc; |
|
|
565 | |
497 | if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
566 | if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
498 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
567 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
499 | |
568 | |
500 | enc_t enc; |
|
|
501 | enc.flags = flags; |
569 | enc.flags = flags; |
502 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
570 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
503 | enc.cur = SvPVX (enc.sv); |
571 | enc.cur = SvPVX (enc.sv); |
504 | enc.end = SvEND (enc.sv); |
572 | enc.end = SvEND (enc.sv); |
505 | enc.indent = 0; |
573 | enc.indent = 0; |
506 | enc.maxdepth = DEC_DEPTH (flags); |
574 | enc.maxdepth = DEC_DEPTH (flags); |
507 | |
575 | |
508 | SvPOK_only (enc.sv); |
576 | SvPOK_only (enc.sv); |
509 | encode_sv (&enc, scalar); |
577 | encode_sv (&enc, scalar); |
510 | |
578 | |
|
|
579 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
|
|
580 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
|
|
581 | |
511 | if (!(flags & (F_ASCII | F_UTF8))) |
582 | if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
512 | SvUTF8_on (enc.sv); |
583 | SvUTF8_on (enc.sv); |
513 | |
|
|
514 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
|
|
515 | |
584 | |
516 | if (enc.flags & F_SHRINK) |
585 | if (enc.flags & F_SHRINK) |
517 | shrink (enc.sv); |
586 | shrink (enc.sv); |
518 | |
587 | |
519 | return enc.sv; |
588 | return enc.sv; |
… | |
… | |
531 | U32 flags; // F_* |
600 | U32 flags; // F_* |
532 | U32 depth; // recursion depth |
601 | U32 depth; // recursion depth |
533 | U32 maxdepth; // recursion depth limit |
602 | U32 maxdepth; // recursion depth limit |
534 | } dec_t; |
603 | } dec_t; |
535 | |
604 | |
536 | static void |
605 | inline void |
537 | decode_ws (dec_t *dec) |
606 | decode_ws (dec_t *dec) |
538 | { |
607 | { |
539 | for (;;) |
608 | for (;;) |
540 | { |
609 | { |
541 | char ch = *dec->cur; |
610 | char ch = *dec->cur; |
… | |
… | |
567 | decode_4hex (dec_t *dec) |
636 | decode_4hex (dec_t *dec) |
568 | { |
637 | { |
569 | signed char d1, d2, d3, d4; |
638 | signed char d1, d2, d3, d4; |
570 | unsigned char *cur = (unsigned char *)dec->cur; |
639 | unsigned char *cur = (unsigned char *)dec->cur; |
571 | |
640 | |
572 | d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); |
641 | d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected"); |
573 | d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); |
642 | d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected"); |
574 | d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); |
643 | d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected"); |
575 | d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); |
644 | d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected"); |
576 | |
645 | |
577 | dec->cur += 4; |
646 | dec->cur += 4; |
578 | |
647 | |
579 | return ((UV)d1) << 12 |
648 | return ((UV)d1) << 12 |
580 | | ((UV)d2) << 8 |
649 | | ((UV)d2) << 8 |
… | |
… | |
588 | static SV * |
657 | static SV * |
589 | decode_str (dec_t *dec) |
658 | decode_str (dec_t *dec) |
590 | { |
659 | { |
591 | SV *sv = 0; |
660 | SV *sv = 0; |
592 | int utf8 = 0; |
661 | int utf8 = 0; |
|
|
662 | char *dec_cur = dec->cur; |
593 | |
663 | |
594 | do |
664 | do |
595 | { |
665 | { |
596 | char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; |
666 | char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; |
597 | char *cur = buf; |
667 | char *cur = buf; |
598 | |
668 | |
599 | do |
669 | do |
600 | { |
670 | { |
601 | unsigned char ch = *(unsigned char *)dec->cur++; |
671 | unsigned char ch = *(unsigned char *)dec_cur++; |
602 | |
672 | |
603 | if (ch == '"') |
673 | if (expect_false (ch == '"')) |
604 | { |
674 | { |
605 | --dec->cur; |
675 | --dec_cur; |
606 | break; |
676 | break; |
607 | } |
677 | } |
608 | else if (ch == '\\') |
678 | else if (expect_false (ch == '\\')) |
609 | { |
679 | { |
610 | switch (*dec->cur) |
680 | switch (*dec_cur) |
611 | { |
681 | { |
612 | case '\\': |
682 | case '\\': |
613 | case '/': |
683 | case '/': |
614 | case '"': *cur++ = *dec->cur++; break; |
684 | case '"': *cur++ = *dec_cur++; break; |
615 | |
685 | |
616 | case 'b': ++dec->cur; *cur++ = '\010'; break; |
686 | case 'b': ++dec_cur; *cur++ = '\010'; break; |
617 | case 't': ++dec->cur; *cur++ = '\011'; break; |
687 | case 't': ++dec_cur; *cur++ = '\011'; break; |
618 | case 'n': ++dec->cur; *cur++ = '\012'; break; |
688 | case 'n': ++dec_cur; *cur++ = '\012'; break; |
619 | case 'f': ++dec->cur; *cur++ = '\014'; break; |
689 | case 'f': ++dec_cur; *cur++ = '\014'; break; |
620 | case 'r': ++dec->cur; *cur++ = '\015'; break; |
690 | case 'r': ++dec_cur; *cur++ = '\015'; break; |
621 | |
691 | |
622 | case 'u': |
692 | case 'u': |
623 | { |
693 | { |
624 | UV lo, hi; |
694 | UV lo, hi; |
625 | ++dec->cur; |
695 | ++dec_cur; |
626 | |
696 | |
|
|
697 | dec->cur = dec_cur; |
627 | hi = decode_4hex (dec); |
698 | hi = decode_4hex (dec); |
|
|
699 | dec_cur = dec->cur; |
628 | if (hi == (UV)-1) |
700 | if (hi == (UV)-1) |
629 | goto fail; |
701 | goto fail; |
630 | |
702 | |
631 | // possibly a surrogate pair |
703 | // possibly a surrogate pair |
632 | if (hi >= 0xd800) |
704 | if (hi >= 0xd800) |
633 | if (hi < 0xdc00) |
705 | if (hi < 0xdc00) |
634 | { |
706 | { |
635 | if (dec->cur [0] != '\\' || dec->cur [1] != 'u') |
707 | if (dec_cur [0] != '\\' || dec_cur [1] != 'u') |
636 | ERR ("missing low surrogate character in surrogate pair"); |
708 | ERR ("missing low surrogate character in surrogate pair"); |
637 | |
709 | |
638 | dec->cur += 2; |
710 | dec_cur += 2; |
639 | |
711 | |
|
|
712 | dec->cur = dec_cur; |
640 | lo = decode_4hex (dec); |
713 | lo = decode_4hex (dec); |
|
|
714 | dec_cur = dec->cur; |
641 | if (lo == (UV)-1) |
715 | if (lo == (UV)-1) |
642 | goto fail; |
716 | goto fail; |
643 | |
717 | |
644 | if (lo < 0xdc00 || lo >= 0xe000) |
718 | if (lo < 0xdc00 || lo >= 0xe000) |
645 | ERR ("surrogate pair expected"); |
719 | ERR ("surrogate pair expected"); |
… | |
… | |
659 | *cur++ = hi; |
733 | *cur++ = hi; |
660 | } |
734 | } |
661 | break; |
735 | break; |
662 | |
736 | |
663 | default: |
737 | default: |
664 | --dec->cur; |
738 | --dec_cur; |
665 | ERR ("illegal backslash escape sequence in string"); |
739 | ERR ("illegal backslash escape sequence in string"); |
666 | } |
740 | } |
667 | } |
741 | } |
668 | else if (ch >= 0x20 && ch <= 0x7f) |
742 | else if (expect_true (ch >= 0x20 && ch <= 0x7f)) |
669 | *cur++ = ch; |
743 | *cur++ = ch; |
670 | else if (ch >= 0x80) |
744 | else if (ch >= 0x80) |
671 | { |
745 | { |
672 | --dec->cur; |
|
|
673 | |
|
|
674 | STRLEN clen; |
746 | STRLEN clen; |
|
|
747 | UV uch; |
|
|
748 | |
|
|
749 | --dec_cur; |
|
|
750 | |
675 | UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); |
751 | uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
676 | if (clen == (STRLEN)-1) |
752 | if (clen == (STRLEN)-1) |
677 | ERR ("malformed UTF-8 character in JSON string"); |
753 | ERR ("malformed UTF-8 character in JSON string"); |
678 | |
754 | |
679 | do |
755 | do |
680 | { |
|
|
681 | *cur++ = *dec->cur++; |
756 | *cur++ = *dec_cur++; |
682 | } |
|
|
683 | while (--clen); |
757 | while (--clen); |
684 | |
758 | |
685 | utf8 = 1; |
759 | utf8 = 1; |
686 | } |
760 | } |
687 | else if (!ch) |
|
|
688 | ERR ("unexpected end of string while parsing json string"); |
|
|
689 | else |
761 | else |
|
|
762 | { |
|
|
763 | --dec_cur; |
|
|
764 | |
|
|
765 | if (!ch) |
|
|
766 | ERR ("unexpected end of string while parsing JSON string"); |
|
|
767 | else |
690 | ERR ("invalid character encountered"); |
768 | ERR ("invalid character encountered while parsing JSON string"); |
691 | |
769 | } |
692 | } |
770 | } |
693 | while (cur < buf + SHORT_STRING_LEN); |
771 | while (cur < buf + SHORT_STRING_LEN); |
694 | |
772 | |
|
|
773 | { |
695 | STRLEN len = cur - buf; |
774 | STRLEN len = cur - buf; |
696 | |
775 | |
697 | if (sv) |
776 | if (sv) |
698 | { |
777 | { |
699 | SvGROW (sv, SvCUR (sv) + len + 1); |
778 | SvGROW (sv, SvCUR (sv) + len + 1); |
700 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
779 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
701 | SvCUR_set (sv, SvCUR (sv) + len); |
780 | SvCUR_set (sv, SvCUR (sv) + len); |
702 | } |
781 | } |
703 | else |
782 | else |
704 | sv = newSVpvn (buf, len); |
783 | sv = newSVpvn (buf, len); |
705 | } |
784 | } |
|
|
785 | } |
706 | while (*dec->cur != '"'); |
786 | while (*dec_cur != '"'); |
707 | |
787 | |
708 | ++dec->cur; |
788 | ++dec_cur; |
709 | |
789 | |
710 | if (sv) |
790 | if (sv) |
711 | { |
791 | { |
712 | SvPOK_only (sv); |
792 | SvPOK_only (sv); |
713 | *SvEND (sv) = 0; |
793 | *SvEND (sv) = 0; |
… | |
… | |
716 | SvUTF8_on (sv); |
796 | SvUTF8_on (sv); |
717 | } |
797 | } |
718 | else |
798 | else |
719 | sv = newSVpvn ("", 0); |
799 | sv = newSVpvn ("", 0); |
720 | |
800 | |
|
|
801 | dec->cur = dec_cur; |
721 | return sv; |
802 | return sv; |
722 | |
803 | |
723 | fail: |
804 | fail: |
|
|
805 | dec->cur = dec_cur; |
724 | return 0; |
806 | return 0; |
725 | } |
807 | } |
726 | |
808 | |
727 | static SV * |
809 | static SV * |
728 | decode_num (dec_t *dec) |
810 | decode_num (dec_t *dec) |
… | |
… | |
786 | is_nv = 1; |
868 | is_nv = 1; |
787 | } |
869 | } |
788 | |
870 | |
789 | if (!is_nv) |
871 | if (!is_nv) |
790 | { |
872 | { |
791 | UV uv; |
873 | // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so |
792 | int numtype = grok_number (start, dec->cur - start, &uv); |
874 | if (*start == '-') |
793 | if (numtype & IS_NUMBER_IN_UV) |
875 | switch (dec->cur - start) |
794 | if (numtype & IS_NUMBER_NEG) |
|
|
795 | { |
876 | { |
796 | if (uv < (UV)IV_MIN) |
877 | case 2: return newSViv (-( start [1] - '0' * 1)); |
797 | return newSViv (-(IV)uv); |
878 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
|
|
879 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
|
|
880 | case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
798 | } |
881 | } |
|
|
882 | else |
|
|
883 | switch (dec->cur - start) |
|
|
884 | { |
|
|
885 | case 1: return newSViv ( start [0] - '0' * 1); |
|
|
886 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
|
|
887 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
|
|
888 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
|
|
889 | } |
|
|
890 | |
|
|
891 | { |
|
|
892 | UV uv; |
|
|
893 | int numtype = grok_number (start, dec->cur - start, &uv); |
|
|
894 | if (numtype & IS_NUMBER_IN_UV) |
|
|
895 | if (numtype & IS_NUMBER_NEG) |
|
|
896 | { |
|
|
897 | if (uv < (UV)IV_MIN) |
|
|
898 | return newSViv (-(IV)uv); |
|
|
899 | } |
799 | else |
900 | else |
800 | return newSVuv (uv); |
901 | return newSVuv (uv); |
|
|
902 | |
|
|
903 | // here would likely be the place for bigint support |
801 | } |
904 | } |
|
|
905 | } |
802 | |
906 | |
|
|
907 | // if we ever support bigint or bigfloat, this is the place for bigfloat |
803 | return newSVnv (Atof (start)); |
908 | return newSVnv (Atof (start)); |
804 | |
909 | |
805 | fail: |
910 | fail: |
806 | return 0; |
911 | return 0; |
807 | } |
912 | } |
… | |
… | |
908 | |
1013 | |
909 | static SV * |
1014 | static SV * |
910 | decode_sv (dec_t *dec) |
1015 | decode_sv (dec_t *dec) |
911 | { |
1016 | { |
912 | decode_ws (dec); |
1017 | decode_ws (dec); |
|
|
1018 | |
|
|
1019 | // the beauty of JSON: you need exactly one character lookahead |
|
|
1020 | // to parse anything. |
913 | switch (*dec->cur) |
1021 | switch (*dec->cur) |
914 | { |
1022 | { |
915 | case '"': ++dec->cur; return decode_str (dec); |
1023 | case '"': ++dec->cur; return decode_str (dec); |
916 | case '[': ++dec->cur; return decode_av (dec); |
1024 | case '[': ++dec->cur; return decode_av (dec); |
917 | case '{': ++dec->cur; return decode_hv (dec); |
1025 | case '{': ++dec->cur; return decode_hv (dec); |
… | |
… | |
953 | ERR ("'null' expected"); |
1061 | ERR ("'null' expected"); |
954 | |
1062 | |
955 | break; |
1063 | break; |
956 | |
1064 | |
957 | default: |
1065 | default: |
958 | ERR ("malformed json string, neither array, object, number, string or atom"); |
1066 | ERR ("malformed JSON string, neither array, object, number, string or atom"); |
959 | break; |
1067 | break; |
960 | } |
1068 | } |
961 | |
1069 | |
962 | fail: |
1070 | fail: |
963 | return 0; |
1071 | return 0; |
964 | } |
1072 | } |
965 | |
1073 | |
966 | static SV * |
1074 | static SV * |
967 | decode_json (SV *string, U32 flags) |
1075 | decode_json (SV *string, U32 flags, UV *offset_return) |
968 | { |
1076 | { |
|
|
1077 | dec_t dec; |
|
|
1078 | UV offset; |
969 | SV *sv; |
1079 | SV *sv; |
|
|
1080 | |
|
|
1081 | SvGETMAGIC (string); |
|
|
1082 | SvUPGRADE (string, SVt_PV); |
970 | |
1083 | |
971 | if (flags & F_UTF8) |
1084 | if (flags & F_UTF8) |
972 | sv_utf8_downgrade (string, 0); |
1085 | sv_utf8_downgrade (string, 0); |
973 | else |
1086 | else |
974 | sv_utf8_upgrade (string); |
1087 | sv_utf8_upgrade (string); |
975 | |
1088 | |
976 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1089 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
977 | |
1090 | |
978 | dec_t dec; |
|
|
979 | dec.flags = flags; |
1091 | dec.flags = flags; |
980 | dec.cur = SvPVX (string); |
1092 | dec.cur = SvPVX (string); |
981 | dec.end = SvEND (string); |
1093 | dec.end = SvEND (string); |
982 | dec.err = 0; |
1094 | dec.err = 0; |
983 | dec.depth = 0; |
1095 | dec.depth = 0; |
984 | dec.maxdepth = DEC_DEPTH (dec.flags); |
1096 | dec.maxdepth = DEC_DEPTH (dec.flags); |
985 | |
1097 | |
986 | *dec.end = 0; // this should basically be a nop, too, but make sure its there |
1098 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
987 | sv = decode_sv (&dec); |
1099 | sv = decode_sv (&dec); |
988 | |
1100 | |
|
|
1101 | if (!(offset_return || !sv)) |
|
|
1102 | { |
|
|
1103 | // check for trailing garbage |
|
|
1104 | decode_ws (&dec); |
|
|
1105 | |
|
|
1106 | if (*dec.cur) |
|
|
1107 | { |
|
|
1108 | dec.err = "garbage after JSON object"; |
|
|
1109 | SvREFCNT_dec (sv); |
|
|
1110 | sv = 0; |
|
|
1111 | } |
|
|
1112 | } |
|
|
1113 | |
|
|
1114 | if (offset_return || !sv) |
|
|
1115 | { |
|
|
1116 | offset = dec.flags & F_UTF8 |
|
|
1117 | ? dec.cur - SvPVX (string) |
|
|
1118 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
1119 | |
|
|
1120 | if (offset_return) |
|
|
1121 | *offset_return = offset; |
|
|
1122 | } |
|
|
1123 | |
989 | if (!sv) |
1124 | if (!sv) |
990 | { |
1125 | { |
991 | IV offset = dec.flags & F_UTF8 |
|
|
992 | ? dec.cur - SvPVX (string) |
|
|
993 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
994 | SV *uni = sv_newmortal (); |
1126 | SV *uni = sv_newmortal (); |
995 | |
1127 | |
996 | // horrible hack to silence warning inside pv_uni_display |
1128 | // horrible hack to silence warning inside pv_uni_display |
997 | COP cop = *PL_curcop; |
1129 | COP cop = *PL_curcop; |
998 | cop.cop_warnings = pWARN_NONE; |
1130 | cop.cop_warnings = pWARN_NONE; |
… | |
… | |
1000 | SAVEVPTR (PL_curcop); |
1132 | SAVEVPTR (PL_curcop); |
1001 | PL_curcop = &cop; |
1133 | PL_curcop = &cop; |
1002 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1134 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1003 | LEAVE; |
1135 | LEAVE; |
1004 | |
1136 | |
1005 | croak ("%s, at character offset %d (%s)", |
1137 | croak ("%s, at character offset %d [\"%s\"]", |
1006 | dec.err, |
1138 | dec.err, |
1007 | (int)offset, |
1139 | (int)offset, |
1008 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1140 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1009 | } |
1141 | } |
1010 | |
1142 | |
… | |
… | |
1022 | MODULE = JSON::XS PACKAGE = JSON::XS |
1154 | MODULE = JSON::XS PACKAGE = JSON::XS |
1023 | |
1155 | |
1024 | BOOT: |
1156 | BOOT: |
1025 | { |
1157 | { |
1026 | int i; |
1158 | int i; |
1027 | |
|
|
1028 | memset (decode_hexdigit, 0xff, 256); |
|
|
1029 | |
1159 | |
1030 | for (i = 0; i < 256; ++i) |
1160 | for (i = 0; i < 256; ++i) |
1031 | decode_hexdigit [i] = |
1161 | decode_hexdigit [i] = |
1032 | i >= '0' && i <= '9' ? i - '0' |
1162 | i >= '0' && i <= '9' ? i - '0' |
1033 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1163 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
… | |
… | |
1046 | RETVAL |
1176 | RETVAL |
1047 | |
1177 | |
1048 | SV *ascii (SV *self, int enable = 1) |
1178 | SV *ascii (SV *self, int enable = 1) |
1049 | ALIAS: |
1179 | ALIAS: |
1050 | ascii = F_ASCII |
1180 | ascii = F_ASCII |
|
|
1181 | latin1 = F_LATIN1 |
1051 | utf8 = F_UTF8 |
1182 | utf8 = F_UTF8 |
1052 | indent = F_INDENT |
1183 | indent = F_INDENT |
1053 | canonical = F_CANONICAL |
1184 | canonical = F_CANONICAL |
1054 | space_before = F_SPACE_BEFORE |
1185 | space_before = F_SPACE_BEFORE |
1055 | space_after = F_SPACE_AFTER |
1186 | space_after = F_SPACE_AFTER |
… | |
… | |
1067 | RETVAL = newSVsv (self); |
1198 | RETVAL = newSVsv (self); |
1068 | } |
1199 | } |
1069 | OUTPUT: |
1200 | OUTPUT: |
1070 | RETVAL |
1201 | RETVAL |
1071 | |
1202 | |
1072 | SV *max_depth (SV *self, int max_depth = 0x80000000UL) |
1203 | SV *max_depth (SV *self, UV max_depth = 0x80000000UL) |
1073 | CODE: |
1204 | CODE: |
1074 | { |
1205 | { |
1075 | UV *uv = SvJSON (self); |
1206 | UV *uv = SvJSON (self); |
1076 | UV log2 = 0; |
1207 | UV log2 = 0; |
1077 | |
1208 | |
… | |
… | |
1091 | PPCODE: |
1222 | PPCODE: |
1092 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
1223 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
1093 | |
1224 | |
1094 | void decode (SV *self, SV *jsonstr) |
1225 | void decode (SV *self, SV *jsonstr) |
1095 | PPCODE: |
1226 | PPCODE: |
1096 | XPUSHs (decode_json (jsonstr, *SvJSON (self))); |
1227 | XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); |
|
|
1228 | |
|
|
1229 | void decode_prefix (SV *self, SV *jsonstr) |
|
|
1230 | PPCODE: |
|
|
1231 | { |
|
|
1232 | UV offset; |
|
|
1233 | EXTEND (SP, 2); |
|
|
1234 | PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); |
|
|
1235 | PUSHs (sv_2mortal (newSVuv (offset))); |
|
|
1236 | } |
1097 | |
1237 | |
1098 | PROTOTYPES: ENABLE |
1238 | PROTOTYPES: ENABLE |
1099 | |
1239 | |
1100 | void to_json (SV *scalar) |
1240 | void to_json (SV *scalar) |
1101 | ALIAS: |
1241 | ALIAS: |
… | |
… | |
1105 | |
1245 | |
1106 | void from_json (SV *jsonstr) |
1246 | void from_json (SV *jsonstr) |
1107 | ALIAS: |
1247 | ALIAS: |
1108 | jsonToObj = 0 |
1248 | jsonToObj = 0 |
1109 | PPCODE: |
1249 | PPCODE: |
1110 | XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); |
1250 | XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0)); |
1111 | |
1251 | |