… | |
… | |
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. if it breaks, you get to keep the pieces. |
|
|
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); |
… | |
… | |
96 | U32 flags; // F_* |
118 | U32 flags; // F_* |
97 | U32 indent; // indentation level |
119 | U32 indent; // indentation level |
98 | U32 maxdepth; // max. indentation/recursion level |
120 | U32 maxdepth; // max. indentation/recursion level |
99 | } enc_t; |
121 | } enc_t; |
100 | |
122 | |
101 | static void |
123 | inline void |
102 | need (enc_t *enc, STRLEN len) |
124 | need (enc_t *enc, STRLEN len) |
103 | { |
125 | { |
104 | if (enc->cur + len >= enc->end) |
126 | if (expect_false (enc->cur + len >= enc->end)) |
105 | { |
127 | { |
106 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
128 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
107 | SvGROW (enc->sv, cur + len + 1); |
129 | SvGROW (enc->sv, cur + len + 1); |
108 | enc->cur = SvPVX (enc->sv) + cur; |
130 | enc->cur = SvPVX (enc->sv) + cur; |
109 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); |
131 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
110 | } |
132 | } |
111 | } |
133 | } |
112 | |
134 | |
113 | static void |
135 | inline void |
114 | encode_ch (enc_t *enc, char ch) |
136 | encode_ch (enc_t *enc, char ch) |
115 | { |
137 | { |
116 | need (enc, 1); |
138 | need (enc, 1); |
117 | *enc->cur++ = ch; |
139 | *enc->cur++ = ch; |
118 | } |
140 | } |
… | |
… | |
126 | |
148 | |
127 | while (str < end) |
149 | while (str < end) |
128 | { |
150 | { |
129 | unsigned char ch = *(unsigned char *)str; |
151 | unsigned char ch = *(unsigned char *)str; |
130 | |
152 | |
131 | if (ch >= 0x20 && ch < 0x80) // most common case |
153 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
132 | { |
154 | { |
133 | if (ch == '"') // but with slow exceptions |
155 | if (expect_false (ch == '"')) // but with slow exceptions |
134 | { |
156 | { |
135 | need (enc, len += 1); |
157 | need (enc, len += 1); |
136 | *enc->cur++ = '\\'; |
158 | *enc->cur++ = '\\'; |
137 | *enc->cur++ = '"'; |
159 | *enc->cur++ = '"'; |
138 | } |
160 | } |
139 | else if (ch == '\\') |
161 | else if (expect_false (ch == '\\')) |
140 | { |
162 | { |
141 | need (enc, len += 1); |
163 | need (enc, len += 1); |
142 | *enc->cur++ = '\\'; |
164 | *enc->cur++ = '\\'; |
143 | *enc->cur++ = '\\'; |
165 | *enc->cur++ = '\\'; |
144 | } |
166 | } |
… | |
… | |
162 | STRLEN clen; |
184 | STRLEN clen; |
163 | UV uch; |
185 | UV uch; |
164 | |
186 | |
165 | if (is_utf8) |
187 | if (is_utf8) |
166 | { |
188 | { |
167 | //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); |
|
|
168 | uch = decode_utf8 (str, end - str, &clen); |
189 | uch = decode_utf8 (str, end - str, &clen); |
169 | if (clen == (STRLEN)-1) |
190 | if (clen == (STRLEN)-1) |
170 | 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); |
171 | } |
192 | } |
172 | else |
193 | else |
… | |
… | |
176 | } |
197 | } |
177 | |
198 | |
178 | if (uch > 0x10FFFFUL) |
199 | if (uch > 0x10FFFFUL) |
179 | 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); |
180 | |
201 | |
181 | if (uch < 0x80 || enc->flags & F_ASCII) |
202 | if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF)) |
182 | { |
203 | { |
183 | if (uch > 0xFFFFUL) |
204 | if (uch > 0xFFFFUL) |
184 | { |
205 | { |
185 | need (enc, len += 11); |
206 | need (enc, len += 11); |
186 | sprintf (enc->cur, "\\u%04x\\u%04x", |
207 | sprintf (enc->cur, "\\u%04x\\u%04x", |
… | |
… | |
200 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
221 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
201 | } |
222 | } |
202 | |
223 | |
203 | str += clen; |
224 | str += clen; |
204 | } |
225 | } |
|
|
226 | else if (enc->flags & F_LATIN1) |
|
|
227 | { |
|
|
228 | *enc->cur++ = uch; |
|
|
229 | str += clen; |
|
|
230 | } |
205 | else if (is_utf8) |
231 | else if (is_utf8) |
206 | { |
232 | { |
207 | need (enc, len += clen); |
233 | need (enc, len += clen); |
208 | do |
234 | do |
209 | { |
235 | { |
… | |
… | |
211 | } |
237 | } |
212 | while (--clen); |
238 | while (--clen); |
213 | } |
239 | } |
214 | else |
240 | else |
215 | { |
241 | { |
216 | 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 |
217 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
243 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
218 | ++str; |
244 | ++str; |
219 | } |
245 | } |
220 | } |
246 | } |
221 | } |
247 | } |
… | |
… | |
223 | |
249 | |
224 | --len; |
250 | --len; |
225 | } |
251 | } |
226 | } |
252 | } |
227 | |
253 | |
228 | static void |
254 | inline void |
229 | encode_indent (enc_t *enc) |
255 | encode_indent (enc_t *enc) |
230 | { |
256 | { |
231 | if (enc->flags & F_INDENT) |
257 | if (enc->flags & F_INDENT) |
232 | { |
258 | { |
233 | int spaces = enc->indent * INDENT_STEP; |
259 | int spaces = enc->indent * INDENT_STEP; |
… | |
… | |
236 | memset (enc->cur, ' ', spaces); |
262 | memset (enc->cur, ' ', spaces); |
237 | enc->cur += spaces; |
263 | enc->cur += spaces; |
238 | } |
264 | } |
239 | } |
265 | } |
240 | |
266 | |
241 | static void |
267 | inline void |
242 | encode_space (enc_t *enc) |
268 | encode_space (enc_t *enc) |
243 | { |
269 | { |
244 | need (enc, 1); |
270 | need (enc, 1); |
245 | encode_ch (enc, ' '); |
271 | encode_ch (enc, ' '); |
246 | } |
272 | } |
247 | |
273 | |
248 | static void |
274 | inline void |
249 | encode_nl (enc_t *enc) |
275 | encode_nl (enc_t *enc) |
250 | { |
276 | { |
251 | if (enc->flags & F_INDENT) |
277 | if (enc->flags & F_INDENT) |
252 | { |
278 | { |
253 | need (enc, 1); |
279 | need (enc, 1); |
254 | encode_ch (enc, '\n'); |
280 | encode_ch (enc, '\n'); |
255 | } |
281 | } |
256 | } |
282 | } |
257 | |
283 | |
258 | static void |
284 | inline void |
259 | encode_comma (enc_t *enc) |
285 | encode_comma (enc_t *enc) |
260 | { |
286 | { |
261 | encode_ch (enc, ','); |
287 | encode_ch (enc, ','); |
262 | |
288 | |
263 | if (enc->flags & F_INDENT) |
289 | if (enc->flags & F_INDENT) |
… | |
… | |
362 | // actually, this is mostly due to the stupid so-called |
388 | // actually, this is mostly due to the stupid so-called |
363 | // security workaround added somewhere in 5.8.x. |
389 | // security workaround added somewhere in 5.8.x. |
364 | // that randomises hash orderings |
390 | // that randomises hash orderings |
365 | if (enc->flags & F_CANONICAL) |
391 | if (enc->flags & F_CANONICAL) |
366 | { |
392 | { |
367 | HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode |
|
|
368 | 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 |
369 | |
400 | |
370 | i = 0; |
401 | i = 0; |
371 | while ((he = hv_iternext (hv))) |
402 | while ((he = hv_iternext (hv))) |
372 | { |
403 | { |
373 | hes [i++] = he; |
404 | hes [i++] = he; |
… | |
… | |
408 | |
439 | |
409 | encode_nl (enc); |
440 | encode_nl (enc); |
410 | } |
441 | } |
411 | else |
442 | else |
412 | { |
443 | { |
413 | SV *sv; |
|
|
414 | HE *he = hv_iternext (hv); |
444 | HE *he = hv_iternext (hv); |
415 | |
445 | |
416 | for (;;) |
446 | for (;;) |
417 | { |
447 | { |
418 | encode_indent (enc); |
448 | encode_indent (enc); |
… | |
… | |
433 | |
463 | |
434 | // encode objects, arrays and special \0=false and \1=true values. |
464 | // encode objects, arrays and special \0=false and \1=true values. |
435 | static void |
465 | static void |
436 | encode_rv (enc_t *enc, SV *sv) |
466 | encode_rv (enc_t *enc, SV *sv) |
437 | { |
467 | { |
|
|
468 | svtype svt; |
|
|
469 | |
438 | SvGETMAGIC (sv); |
470 | SvGETMAGIC (sv); |
439 | |
|
|
440 | svtype svt = SvTYPE (sv); |
471 | svt = SvTYPE (sv); |
441 | |
472 | |
442 | if (svt == SVt_PVHV) |
473 | if (svt == SVt_PVHV) |
443 | encode_hv (enc, (HV *)sv); |
474 | encode_hv (enc, (HV *)sv); |
444 | else if (svt == SVt_PVAV) |
475 | else if (svt == SVt_PVAV) |
445 | encode_av (enc, (AV *)sv); |
476 | encode_av (enc, (AV *)sv); |
… | |
… | |
471 | encode_str (enc, str, len, SvUTF8 (sv)); |
502 | encode_str (enc, str, len, SvUTF8 (sv)); |
472 | encode_ch (enc, '"'); |
503 | encode_ch (enc, '"'); |
473 | } |
504 | } |
474 | else if (SvNOKp (sv)) |
505 | else if (SvNOKp (sv)) |
475 | { |
506 | { |
|
|
507 | // trust that perl will do the right thing w.r.t. JSON syntax. |
476 | need (enc, NV_DIG + 32); |
508 | need (enc, NV_DIG + 32); |
477 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
509 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
478 | enc->cur += strlen (enc->cur); |
510 | enc->cur += strlen (enc->cur); |
479 | } |
511 | } |
480 | else if (SvIOKp (sv)) |
512 | else if (SvIOKp (sv)) |
481 | { |
513 | { |
482 | 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); |
483 | enc->cur += |
519 | enc->cur += |
484 | SvIsUV(sv) |
520 | SvIsUV(sv) |
485 | ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) |
521 | ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv)) |
486 | : 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 | } |
487 | } |
551 | } |
488 | else if (SvROK (sv)) |
552 | else if (SvROK (sv)) |
489 | encode_rv (enc, SvRV (sv)); |
553 | encode_rv (enc, SvRV (sv)); |
490 | else if (!SvOK (sv)) |
554 | else if (!SvOK (sv)) |
491 | encode_str (enc, "null", 4, 0); |
555 | encode_str (enc, "null", 4, 0); |
… | |
… | |
495 | } |
559 | } |
496 | |
560 | |
497 | static SV * |
561 | static SV * |
498 | encode_json (SV *scalar, U32 flags) |
562 | encode_json (SV *scalar, U32 flags) |
499 | { |
563 | { |
|
|
564 | enc_t enc; |
|
|
565 | |
500 | if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
566 | if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
501 | 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)"); |
502 | |
568 | |
503 | enc_t enc; |
|
|
504 | enc.flags = flags; |
569 | enc.flags = flags; |
505 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
570 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
506 | enc.cur = SvPVX (enc.sv); |
571 | enc.cur = SvPVX (enc.sv); |
507 | enc.end = SvEND (enc.sv); |
572 | enc.end = SvEND (enc.sv); |
508 | enc.indent = 0; |
573 | enc.indent = 0; |
509 | enc.maxdepth = DEC_DEPTH (flags); |
574 | enc.maxdepth = DEC_DEPTH (flags); |
510 | |
575 | |
511 | SvPOK_only (enc.sv); |
576 | SvPOK_only (enc.sv); |
512 | encode_sv (&enc, scalar); |
577 | encode_sv (&enc, scalar); |
513 | |
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 | |
514 | if (!(flags & (F_ASCII | F_UTF8))) |
582 | if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
515 | SvUTF8_on (enc.sv); |
583 | SvUTF8_on (enc.sv); |
516 | |
|
|
517 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
|
|
518 | |
584 | |
519 | if (enc.flags & F_SHRINK) |
585 | if (enc.flags & F_SHRINK) |
520 | shrink (enc.sv); |
586 | shrink (enc.sv); |
521 | |
587 | |
522 | return enc.sv; |
588 | return enc.sv; |
… | |
… | |
534 | U32 flags; // F_* |
600 | U32 flags; // F_* |
535 | U32 depth; // recursion depth |
601 | U32 depth; // recursion depth |
536 | U32 maxdepth; // recursion depth limit |
602 | U32 maxdepth; // recursion depth limit |
537 | } dec_t; |
603 | } dec_t; |
538 | |
604 | |
539 | static void |
605 | inline void |
540 | decode_ws (dec_t *dec) |
606 | decode_ws (dec_t *dec) |
541 | { |
607 | { |
542 | for (;;) |
608 | for (;;) |
543 | { |
609 | { |
544 | char ch = *dec->cur; |
610 | char ch = *dec->cur; |
… | |
… | |
570 | decode_4hex (dec_t *dec) |
636 | decode_4hex (dec_t *dec) |
571 | { |
637 | { |
572 | signed char d1, d2, d3, d4; |
638 | signed char d1, d2, d3, d4; |
573 | unsigned char *cur = (unsigned char *)dec->cur; |
639 | unsigned char *cur = (unsigned char *)dec->cur; |
574 | |
640 | |
575 | 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"); |
576 | 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"); |
577 | 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"); |
578 | 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"); |
579 | |
645 | |
580 | dec->cur += 4; |
646 | dec->cur += 4; |
581 | |
647 | |
582 | return ((UV)d1) << 12 |
648 | return ((UV)d1) << 12 |
583 | | ((UV)d2) << 8 |
649 | | ((UV)d2) << 8 |
… | |
… | |
591 | static SV * |
657 | static SV * |
592 | decode_str (dec_t *dec) |
658 | decode_str (dec_t *dec) |
593 | { |
659 | { |
594 | SV *sv = 0; |
660 | SV *sv = 0; |
595 | int utf8 = 0; |
661 | int utf8 = 0; |
|
|
662 | char *dec_cur = dec->cur; |
596 | |
663 | |
597 | do |
664 | do |
598 | { |
665 | { |
599 | char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; |
666 | char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; |
600 | char *cur = buf; |
667 | char *cur = buf; |
601 | |
668 | |
602 | do |
669 | do |
603 | { |
670 | { |
604 | unsigned char ch = *(unsigned char *)dec->cur++; |
671 | unsigned char ch = *(unsigned char *)dec_cur++; |
605 | |
672 | |
606 | if (ch == '"') |
673 | if (expect_false (ch == '"')) |
607 | { |
674 | { |
608 | --dec->cur; |
675 | --dec_cur; |
609 | break; |
676 | break; |
610 | } |
677 | } |
611 | else if (ch == '\\') |
678 | else if (expect_false (ch == '\\')) |
612 | { |
679 | { |
613 | switch (*dec->cur) |
680 | switch (*dec_cur) |
614 | { |
681 | { |
615 | case '\\': |
682 | case '\\': |
616 | case '/': |
683 | case '/': |
617 | case '"': *cur++ = *dec->cur++; break; |
684 | case '"': *cur++ = *dec_cur++; break; |
618 | |
685 | |
619 | case 'b': ++dec->cur; *cur++ = '\010'; break; |
686 | case 'b': ++dec_cur; *cur++ = '\010'; break; |
620 | case 't': ++dec->cur; *cur++ = '\011'; break; |
687 | case 't': ++dec_cur; *cur++ = '\011'; break; |
621 | case 'n': ++dec->cur; *cur++ = '\012'; break; |
688 | case 'n': ++dec_cur; *cur++ = '\012'; break; |
622 | case 'f': ++dec->cur; *cur++ = '\014'; break; |
689 | case 'f': ++dec_cur; *cur++ = '\014'; break; |
623 | case 'r': ++dec->cur; *cur++ = '\015'; break; |
690 | case 'r': ++dec_cur; *cur++ = '\015'; break; |
624 | |
691 | |
625 | case 'u': |
692 | case 'u': |
626 | { |
693 | { |
627 | UV lo, hi; |
694 | UV lo, hi; |
628 | ++dec->cur; |
695 | ++dec_cur; |
629 | |
696 | |
|
|
697 | dec->cur = dec_cur; |
630 | hi = decode_4hex (dec); |
698 | hi = decode_4hex (dec); |
|
|
699 | dec_cur = dec->cur; |
631 | if (hi == (UV)-1) |
700 | if (hi == (UV)-1) |
632 | goto fail; |
701 | goto fail; |
633 | |
702 | |
634 | // possibly a surrogate pair |
703 | // possibly a surrogate pair |
635 | if (hi >= 0xd800) |
704 | if (hi >= 0xd800) |
636 | if (hi < 0xdc00) |
705 | if (hi < 0xdc00) |
637 | { |
706 | { |
638 | if (dec->cur [0] != '\\' || dec->cur [1] != 'u') |
707 | if (dec_cur [0] != '\\' || dec_cur [1] != 'u') |
639 | ERR ("missing low surrogate character in surrogate pair"); |
708 | ERR ("missing low surrogate character in surrogate pair"); |
640 | |
709 | |
641 | dec->cur += 2; |
710 | dec_cur += 2; |
642 | |
711 | |
|
|
712 | dec->cur = dec_cur; |
643 | lo = decode_4hex (dec); |
713 | lo = decode_4hex (dec); |
|
|
714 | dec_cur = dec->cur; |
644 | if (lo == (UV)-1) |
715 | if (lo == (UV)-1) |
645 | goto fail; |
716 | goto fail; |
646 | |
717 | |
647 | if (lo < 0xdc00 || lo >= 0xe000) |
718 | if (lo < 0xdc00 || lo >= 0xe000) |
648 | ERR ("surrogate pair expected"); |
719 | ERR ("surrogate pair expected"); |
… | |
… | |
662 | *cur++ = hi; |
733 | *cur++ = hi; |
663 | } |
734 | } |
664 | break; |
735 | break; |
665 | |
736 | |
666 | default: |
737 | default: |
667 | --dec->cur; |
738 | --dec_cur; |
668 | ERR ("illegal backslash escape sequence in string"); |
739 | ERR ("illegal backslash escape sequence in string"); |
669 | } |
740 | } |
670 | } |
741 | } |
671 | else if (ch >= 0x20 && ch <= 0x7f) |
742 | else if (expect_true (ch >= 0x20 && ch <= 0x7f)) |
672 | *cur++ = ch; |
743 | *cur++ = ch; |
673 | else if (ch >= 0x80) |
744 | else if (ch >= 0x80) |
674 | { |
745 | { |
675 | --dec->cur; |
|
|
676 | |
|
|
677 | STRLEN clen; |
746 | STRLEN clen; |
|
|
747 | UV uch; |
|
|
748 | |
|
|
749 | --dec_cur; |
|
|
750 | |
678 | UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); |
751 | uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
679 | if (clen == (STRLEN)-1) |
752 | if (clen == (STRLEN)-1) |
680 | ERR ("malformed UTF-8 character in JSON string"); |
753 | ERR ("malformed UTF-8 character in JSON string"); |
681 | |
754 | |
682 | do |
755 | do |
683 | *cur++ = *dec->cur++; |
756 | *cur++ = *dec_cur++; |
684 | while (--clen); |
757 | while (--clen); |
685 | |
758 | |
686 | utf8 = 1; |
759 | utf8 = 1; |
687 | } |
760 | } |
688 | else |
761 | else |
689 | { |
762 | { |
690 | --dec->cur; |
763 | --dec_cur; |
691 | |
764 | |
692 | if (!ch) |
765 | if (!ch) |
693 | ERR ("unexpected end of string while parsing json string"); |
766 | ERR ("unexpected end of string while parsing JSON string"); |
694 | else |
767 | else |
695 | ERR ("invalid character encountered while parsing json string"); |
768 | ERR ("invalid character encountered while parsing JSON string"); |
696 | } |
769 | } |
697 | } |
770 | } |
698 | while (cur < buf + SHORT_STRING_LEN); |
771 | while (cur < buf + SHORT_STRING_LEN); |
699 | |
772 | |
|
|
773 | { |
700 | STRLEN len = cur - buf; |
774 | STRLEN len = cur - buf; |
701 | |
775 | |
702 | if (sv) |
776 | if (sv) |
703 | { |
777 | { |
704 | SvGROW (sv, SvCUR (sv) + len + 1); |
778 | SvGROW (sv, SvCUR (sv) + len + 1); |
705 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
779 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
706 | SvCUR_set (sv, SvCUR (sv) + len); |
780 | SvCUR_set (sv, SvCUR (sv) + len); |
707 | } |
781 | } |
708 | else |
782 | else |
709 | sv = newSVpvn (buf, len); |
783 | sv = newSVpvn (buf, len); |
710 | } |
784 | } |
|
|
785 | } |
711 | while (*dec->cur != '"'); |
786 | while (*dec_cur != '"'); |
712 | |
787 | |
713 | ++dec->cur; |
788 | ++dec_cur; |
714 | |
789 | |
715 | if (sv) |
790 | if (sv) |
716 | { |
791 | { |
717 | SvPOK_only (sv); |
792 | SvPOK_only (sv); |
718 | *SvEND (sv) = 0; |
793 | *SvEND (sv) = 0; |
… | |
… | |
721 | SvUTF8_on (sv); |
796 | SvUTF8_on (sv); |
722 | } |
797 | } |
723 | else |
798 | else |
724 | sv = newSVpvn ("", 0); |
799 | sv = newSVpvn ("", 0); |
725 | |
800 | |
|
|
801 | dec->cur = dec_cur; |
726 | return sv; |
802 | return sv; |
727 | |
803 | |
728 | fail: |
804 | fail: |
|
|
805 | dec->cur = dec_cur; |
729 | return 0; |
806 | return 0; |
730 | } |
807 | } |
731 | |
808 | |
732 | static SV * |
809 | static SV * |
733 | decode_num (dec_t *dec) |
810 | decode_num (dec_t *dec) |
… | |
… | |
791 | is_nv = 1; |
868 | is_nv = 1; |
792 | } |
869 | } |
793 | |
870 | |
794 | if (!is_nv) |
871 | if (!is_nv) |
795 | { |
872 | { |
796 | UV uv; |
873 | // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so |
797 | int numtype = grok_number (start, dec->cur - start, &uv); |
874 | if (*start == '-') |
798 | if (numtype & IS_NUMBER_IN_UV) |
875 | switch (dec->cur - start) |
799 | if (numtype & IS_NUMBER_NEG) |
|
|
800 | { |
876 | { |
801 | if (uv < (UV)IV_MIN) |
877 | case 2: return newSViv (-( start [1] - '0' * 1)); |
802 | 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)); |
803 | } |
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 | } |
804 | else |
900 | else |
805 | return newSVuv (uv); |
901 | return newSVuv (uv); |
|
|
902 | |
|
|
903 | // here would likely be the place for bigint support |
806 | } |
904 | } |
|
|
905 | } |
807 | |
906 | |
|
|
907 | // if we ever support bigint or bigfloat, this is the place for bigfloat |
808 | return newSVnv (Atof (start)); |
908 | return newSVnv (Atof (start)); |
809 | |
909 | |
810 | fail: |
910 | fail: |
811 | return 0; |
911 | return 0; |
812 | } |
912 | } |
… | |
… | |
913 | |
1013 | |
914 | static SV * |
1014 | static SV * |
915 | decode_sv (dec_t *dec) |
1015 | decode_sv (dec_t *dec) |
916 | { |
1016 | { |
917 | decode_ws (dec); |
1017 | decode_ws (dec); |
|
|
1018 | |
|
|
1019 | // the beauty of JSON: you need exactly one character lookahead |
|
|
1020 | // to parse anything. |
918 | switch (*dec->cur) |
1021 | switch (*dec->cur) |
919 | { |
1022 | { |
920 | case '"': ++dec->cur; return decode_str (dec); |
1023 | case '"': ++dec->cur; return decode_str (dec); |
921 | case '[': ++dec->cur; return decode_av (dec); |
1024 | case '[': ++dec->cur; return decode_av (dec); |
922 | case '{': ++dec->cur; return decode_hv (dec); |
1025 | case '{': ++dec->cur; return decode_hv (dec); |
… | |
… | |
958 | ERR ("'null' expected"); |
1061 | ERR ("'null' expected"); |
959 | |
1062 | |
960 | break; |
1063 | break; |
961 | |
1064 | |
962 | default: |
1065 | default: |
963 | ERR ("malformed json string, neither array, object, number, string or atom"); |
1066 | ERR ("malformed JSON string, neither array, object, number, string or atom"); |
964 | break; |
1067 | break; |
965 | } |
1068 | } |
966 | |
1069 | |
967 | fail: |
1070 | fail: |
968 | return 0; |
1071 | return 0; |
969 | } |
1072 | } |
970 | |
1073 | |
971 | static SV * |
1074 | static SV * |
972 | decode_json (SV *string, U32 flags) |
1075 | decode_json (SV *string, U32 flags, UV *offset_return) |
973 | { |
1076 | { |
|
|
1077 | dec_t dec; |
|
|
1078 | UV offset; |
974 | SV *sv; |
1079 | SV *sv; |
975 | |
1080 | |
|
|
1081 | SvGETMAGIC (string); |
976 | SvUPGRADE (string, SVt_PV); |
1082 | SvUPGRADE (string, SVt_PV); |
977 | |
1083 | |
978 | if (flags & F_UTF8) |
1084 | if (flags & F_UTF8) |
979 | sv_utf8_downgrade (string, 0); |
1085 | sv_utf8_downgrade (string, 0); |
980 | else |
1086 | else |
981 | sv_utf8_upgrade (string); |
1087 | sv_utf8_upgrade (string); |
982 | |
1088 | |
983 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1089 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
984 | |
1090 | |
985 | dec_t dec; |
|
|
986 | dec.flags = flags; |
1091 | dec.flags = flags; |
987 | dec.cur = SvPVX (string); |
1092 | dec.cur = SvPVX (string); |
988 | dec.end = SvEND (string); |
1093 | dec.end = SvEND (string); |
989 | dec.err = 0; |
1094 | dec.err = 0; |
990 | dec.depth = 0; |
1095 | dec.depth = 0; |
991 | dec.maxdepth = DEC_DEPTH (dec.flags); |
1096 | dec.maxdepth = DEC_DEPTH (dec.flags); |
992 | |
1097 | |
993 | *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 |
994 | sv = decode_sv (&dec); |
1099 | sv = decode_sv (&dec); |
995 | |
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 | |
996 | if (!sv) |
1124 | if (!sv) |
997 | { |
1125 | { |
998 | IV offset = dec.flags & F_UTF8 |
|
|
999 | ? dec.cur - SvPVX (string) |
|
|
1000 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
1001 | SV *uni = sv_newmortal (); |
1126 | SV *uni = sv_newmortal (); |
1002 | |
1127 | |
1003 | // horrible hack to silence warning inside pv_uni_display |
1128 | // horrible hack to silence warning inside pv_uni_display |
1004 | COP cop = *PL_curcop; |
1129 | COP cop = *PL_curcop; |
1005 | cop.cop_warnings = pWARN_NONE; |
1130 | cop.cop_warnings = pWARN_NONE; |
… | |
… | |
1030 | |
1155 | |
1031 | BOOT: |
1156 | BOOT: |
1032 | { |
1157 | { |
1033 | int i; |
1158 | int i; |
1034 | |
1159 | |
1035 | memset (decode_hexdigit, 0xff, 256); |
|
|
1036 | |
|
|
1037 | for (i = 0; i < 256; ++i) |
1160 | for (i = 0; i < 256; ++i) |
1038 | decode_hexdigit [i] = |
1161 | decode_hexdigit [i] = |
1039 | i >= '0' && i <= '9' ? i - '0' |
1162 | i >= '0' && i <= '9' ? i - '0' |
1040 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1163 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1041 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1164 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
… | |
… | |
1053 | RETVAL |
1176 | RETVAL |
1054 | |
1177 | |
1055 | SV *ascii (SV *self, int enable = 1) |
1178 | SV *ascii (SV *self, int enable = 1) |
1056 | ALIAS: |
1179 | ALIAS: |
1057 | ascii = F_ASCII |
1180 | ascii = F_ASCII |
|
|
1181 | latin1 = F_LATIN1 |
1058 | utf8 = F_UTF8 |
1182 | utf8 = F_UTF8 |
1059 | indent = F_INDENT |
1183 | indent = F_INDENT |
1060 | canonical = F_CANONICAL |
1184 | canonical = F_CANONICAL |
1061 | space_before = F_SPACE_BEFORE |
1185 | space_before = F_SPACE_BEFORE |
1062 | space_after = F_SPACE_AFTER |
1186 | space_after = F_SPACE_AFTER |
… | |
… | |
1074 | RETVAL = newSVsv (self); |
1198 | RETVAL = newSVsv (self); |
1075 | } |
1199 | } |
1076 | OUTPUT: |
1200 | OUTPUT: |
1077 | RETVAL |
1201 | RETVAL |
1078 | |
1202 | |
1079 | SV *max_depth (SV *self, int max_depth = 0x80000000UL) |
1203 | SV *max_depth (SV *self, UV max_depth = 0x80000000UL) |
1080 | CODE: |
1204 | CODE: |
1081 | { |
1205 | { |
1082 | UV *uv = SvJSON (self); |
1206 | UV *uv = SvJSON (self); |
1083 | UV log2 = 0; |
1207 | UV log2 = 0; |
1084 | |
1208 | |
… | |
… | |
1098 | PPCODE: |
1222 | PPCODE: |
1099 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
1223 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
1100 | |
1224 | |
1101 | void decode (SV *self, SV *jsonstr) |
1225 | void decode (SV *self, SV *jsonstr) |
1102 | PPCODE: |
1226 | PPCODE: |
1103 | 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 | } |
1104 | |
1237 | |
1105 | PROTOTYPES: ENABLE |
1238 | PROTOTYPES: ENABLE |
1106 | |
1239 | |
1107 | void to_json (SV *scalar) |
1240 | void to_json (SV *scalar) |
1108 | ALIAS: |
1241 | ALIAS: |
… | |
… | |
1112 | |
1245 | |
1113 | void from_json (SV *jsonstr) |
1246 | void from_json (SV *jsonstr) |
1114 | ALIAS: |
1247 | ALIAS: |
1115 | jsonToObj = 0 |
1248 | jsonToObj = 0 |
1116 | PPCODE: |
1249 | PPCODE: |
1117 | XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); |
1250 | XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0)); |
1118 | |
1251 | |