… | |
… | |
9 | |
9 | |
10 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
10 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
11 | # define snprintf _snprintf // C compilers have this in stdio.h |
11 | # define snprintf _snprintf // C compilers have this in stdio.h |
12 | #endif |
12 | #endif |
13 | |
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 |
|
|
19 | |
14 | #define F_ASCII 0x00000001UL |
20 | #define F_ASCII 0x00000001UL |
|
|
21 | #define F_LATIN1 0x00000002UL |
15 | #define F_UTF8 0x00000002UL |
22 | #define F_UTF8 0x00000004UL |
16 | #define F_INDENT 0x00000004UL |
23 | #define F_INDENT 0x00000008UL |
17 | #define F_CANONICAL 0x00000008UL |
24 | #define F_CANONICAL 0x00000010UL |
18 | #define F_SPACE_BEFORE 0x00000010UL |
25 | #define F_SPACE_BEFORE 0x00000020UL |
19 | #define F_SPACE_AFTER 0x00000020UL |
26 | #define F_SPACE_AFTER 0x00000040UL |
20 | #define F_ALLOW_NONREF 0x00000080UL |
27 | #define F_ALLOW_NONREF 0x00000100UL |
21 | #define F_SHRINK 0x00000100UL |
28 | #define F_SHRINK 0x00000200UL |
|
|
29 | #define F_ALLOW_BLESSED 0x00000400UL |
|
|
30 | #define F_CONV_BLESSED 0x00000800UL // NYI |
22 | #define F_MAXDEPTH 0xf8000000UL |
31 | #define F_MAXDEPTH 0xf8000000UL |
23 | #define S_MAXDEPTH 27 |
32 | #define S_MAXDEPTH 27 |
24 | |
33 | |
25 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
34 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
26 | |
|
|
27 | // F_SELFCONVERT? <=> to_json/toJson |
|
|
28 | // F_BLESSED? <=> { $__class__$ => } |
|
|
29 | |
35 | |
30 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
36 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
31 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
37 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
32 | |
38 | |
33 | #define INIT_SIZE 32 // initial scalar size to be allocated |
39 | #define INIT_SIZE 32 // initial scalar size to be allocated |
34 | #define INDENT_STEP 3 // spaces per indentation level |
40 | #define INDENT_STEP 3 // spaces per indentation level |
35 | |
41 | |
36 | #define SHORT_STRING_LEN 512 // special-case strings of up to this size |
42 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
37 | |
43 | |
38 | #define SB do { |
44 | #define SB do { |
39 | #define SE } while (0) |
45 | #define SE } while (0) |
40 | |
46 | |
|
|
47 | #if __GNUC__ >= 3 |
|
|
48 | # define expect(expr,value) __builtin_expect ((expr),(value)) |
|
|
49 | # define inline inline |
|
|
50 | #else |
|
|
51 | # define expect(expr,value) (expr) |
|
|
52 | # define inline static |
|
|
53 | #endif |
|
|
54 | |
|
|
55 | #define expect_false(expr) expect ((expr) != 0, 0) |
|
|
56 | #define expect_true(expr) expect ((expr) != 0, 1) |
|
|
57 | |
41 | static HV *json_stash; // JSON::XS:: |
58 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
|
|
59 | static SV *json_true, *json_false; |
42 | |
60 | |
43 | ///////////////////////////////////////////////////////////////////////////// |
61 | ///////////////////////////////////////////////////////////////////////////// |
44 | // utility functions |
62 | // utility functions |
45 | |
63 | |
46 | static UV * |
64 | static UV * |
… | |
… | |
69 | // 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 |
70 | // case of an error. |
88 | // case of an error. |
71 | // we special-case "safe" characters from U+80 .. U+7FF, |
89 | // we special-case "safe" characters from U+80 .. U+7FF, |
72 | // but use the very good perl function to parse anything else. |
90 | // but use the very good perl function to parse anything else. |
73 | // note that we never call this function for a ascii codepoints |
91 | // note that we never call this function for a ascii codepoints |
74 | static UV |
92 | inline UV |
75 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
93 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
76 | { |
94 | { |
77 | if (s[0] > 0xdf || s[0] < 0xc2) |
95 | if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) |
78 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
96 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
79 | else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
97 | else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
80 | { |
98 | { |
81 | *clen = 2; |
99 | *clen = 2; |
82 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
100 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
… | |
… | |
100 | U32 flags; // F_* |
118 | U32 flags; // F_* |
101 | U32 indent; // indentation level |
119 | U32 indent; // indentation level |
102 | U32 maxdepth; // max. indentation/recursion level |
120 | U32 maxdepth; // max. indentation/recursion level |
103 | } enc_t; |
121 | } enc_t; |
104 | |
122 | |
105 | static void |
123 | inline void |
106 | need (enc_t *enc, STRLEN len) |
124 | need (enc_t *enc, STRLEN len) |
107 | { |
125 | { |
108 | if (enc->cur + len >= enc->end) |
126 | if (expect_false (enc->cur + len >= enc->end)) |
109 | { |
127 | { |
110 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
128 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
111 | SvGROW (enc->sv, cur + len + 1); |
129 | SvGROW (enc->sv, cur + len + 1); |
112 | enc->cur = SvPVX (enc->sv) + cur; |
130 | enc->cur = SvPVX (enc->sv) + cur; |
113 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
131 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
114 | } |
132 | } |
115 | } |
133 | } |
116 | |
134 | |
117 | static void |
135 | inline void |
118 | encode_ch (enc_t *enc, char ch) |
136 | encode_ch (enc_t *enc, char ch) |
119 | { |
137 | { |
120 | need (enc, 1); |
138 | need (enc, 1); |
121 | *enc->cur++ = ch; |
139 | *enc->cur++ = ch; |
122 | } |
140 | } |
… | |
… | |
130 | |
148 | |
131 | while (str < end) |
149 | while (str < end) |
132 | { |
150 | { |
133 | unsigned char ch = *(unsigned char *)str; |
151 | unsigned char ch = *(unsigned char *)str; |
134 | |
152 | |
135 | if (ch >= 0x20 && ch < 0x80) // most common case |
153 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
136 | { |
154 | { |
137 | if (ch == '"') // but with slow exceptions |
155 | if (expect_false (ch == '"')) // but with slow exceptions |
138 | { |
156 | { |
139 | need (enc, len += 1); |
157 | need (enc, len += 1); |
140 | *enc->cur++ = '\\'; |
158 | *enc->cur++ = '\\'; |
141 | *enc->cur++ = '"'; |
159 | *enc->cur++ = '"'; |
142 | } |
160 | } |
143 | else if (ch == '\\') |
161 | else if (expect_false (ch == '\\')) |
144 | { |
162 | { |
145 | need (enc, len += 1); |
163 | need (enc, len += 1); |
146 | *enc->cur++ = '\\'; |
164 | *enc->cur++ = '\\'; |
147 | *enc->cur++ = '\\'; |
165 | *enc->cur++ = '\\'; |
148 | } |
166 | } |
… | |
… | |
166 | STRLEN clen; |
184 | STRLEN clen; |
167 | UV uch; |
185 | UV uch; |
168 | |
186 | |
169 | if (is_utf8) |
187 | if (is_utf8) |
170 | { |
188 | { |
171 | //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); |
|
|
172 | uch = decode_utf8 (str, end - str, &clen); |
189 | uch = decode_utf8 (str, end - str, &clen); |
173 | if (clen == (STRLEN)-1) |
190 | if (clen == (STRLEN)-1) |
174 | 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); |
175 | } |
192 | } |
176 | else |
193 | else |
… | |
… | |
180 | } |
197 | } |
181 | |
198 | |
182 | if (uch > 0x10FFFFUL) |
199 | if (uch > 0x10FFFFUL) |
183 | 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); |
184 | |
201 | |
185 | if (uch < 0x80 || enc->flags & F_ASCII) |
202 | if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF)) |
186 | { |
203 | { |
187 | if (uch > 0xFFFFUL) |
204 | if (uch > 0xFFFFUL) |
188 | { |
205 | { |
189 | need (enc, len += 11); |
206 | need (enc, len += 11); |
190 | sprintf (enc->cur, "\\u%04x\\u%04x", |
207 | sprintf (enc->cur, "\\u%04x\\u%04x", |
… | |
… | |
204 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
221 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
205 | } |
222 | } |
206 | |
223 | |
207 | str += clen; |
224 | str += clen; |
208 | } |
225 | } |
|
|
226 | else if (enc->flags & F_LATIN1) |
|
|
227 | { |
|
|
228 | *enc->cur++ = uch; |
|
|
229 | str += clen; |
|
|
230 | } |
209 | else if (is_utf8) |
231 | else if (is_utf8) |
210 | { |
232 | { |
211 | need (enc, len += clen); |
233 | need (enc, len += clen); |
212 | do |
234 | do |
213 | { |
235 | { |
… | |
… | |
227 | |
249 | |
228 | --len; |
250 | --len; |
229 | } |
251 | } |
230 | } |
252 | } |
231 | |
253 | |
232 | static void |
254 | inline void |
233 | encode_indent (enc_t *enc) |
255 | encode_indent (enc_t *enc) |
234 | { |
256 | { |
235 | if (enc->flags & F_INDENT) |
257 | if (enc->flags & F_INDENT) |
236 | { |
258 | { |
237 | int spaces = enc->indent * INDENT_STEP; |
259 | int spaces = enc->indent * INDENT_STEP; |
… | |
… | |
240 | memset (enc->cur, ' ', spaces); |
262 | memset (enc->cur, ' ', spaces); |
241 | enc->cur += spaces; |
263 | enc->cur += spaces; |
242 | } |
264 | } |
243 | } |
265 | } |
244 | |
266 | |
245 | static void |
267 | inline void |
246 | encode_space (enc_t *enc) |
268 | encode_space (enc_t *enc) |
247 | { |
269 | { |
248 | need (enc, 1); |
270 | need (enc, 1); |
249 | encode_ch (enc, ' '); |
271 | encode_ch (enc, ' '); |
250 | } |
272 | } |
251 | |
273 | |
252 | static void |
274 | inline void |
253 | encode_nl (enc_t *enc) |
275 | encode_nl (enc_t *enc) |
254 | { |
276 | { |
255 | if (enc->flags & F_INDENT) |
277 | if (enc->flags & F_INDENT) |
256 | { |
278 | { |
257 | need (enc, 1); |
279 | need (enc, 1); |
258 | encode_ch (enc, '\n'); |
280 | encode_ch (enc, '\n'); |
259 | } |
281 | } |
260 | } |
282 | } |
261 | |
283 | |
262 | static void |
284 | inline void |
263 | encode_comma (enc_t *enc) |
285 | encode_comma (enc_t *enc) |
264 | { |
286 | { |
265 | encode_ch (enc, ','); |
287 | encode_ch (enc, ','); |
266 | |
288 | |
267 | if (enc->flags & F_INDENT) |
289 | if (enc->flags & F_INDENT) |
… | |
… | |
366 | // actually, this is mostly due to the stupid so-called |
388 | // actually, this is mostly due to the stupid so-called |
367 | // security workaround added somewhere in 5.8.x. |
389 | // security workaround added somewhere in 5.8.x. |
368 | // that randomises hash orderings |
390 | // that randomises hash orderings |
369 | if (enc->flags & F_CANONICAL) |
391 | if (enc->flags & F_CANONICAL) |
370 | { |
392 | { |
371 | HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode |
|
|
372 | 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 |
373 | |
400 | |
374 | i = 0; |
401 | i = 0; |
375 | while ((he = hv_iternext (hv))) |
402 | while ((he = hv_iternext (hv))) |
376 | { |
403 | { |
377 | hes [i++] = he; |
404 | hes [i++] = he; |
… | |
… | |
441 | svtype svt; |
468 | svtype svt; |
442 | |
469 | |
443 | SvGETMAGIC (sv); |
470 | SvGETMAGIC (sv); |
444 | svt = SvTYPE (sv); |
471 | svt = SvTYPE (sv); |
445 | |
472 | |
|
|
473 | if (expect_false (SvOBJECT (sv))) |
|
|
474 | { |
|
|
475 | if (SvSTASH (sv) == json_boolean_stash) |
|
|
476 | { |
|
|
477 | if (SvIV (sv) == 0) |
|
|
478 | encode_str (enc, "false", 5, 0); |
|
|
479 | else |
|
|
480 | encode_str (enc, "true", 4, 0); |
|
|
481 | } |
|
|
482 | else |
|
|
483 | { |
|
|
484 | #if 0 |
|
|
485 | if (0 && sv_derived_from (rv, "JSON::Literal")) |
|
|
486 | { |
|
|
487 | // not yet |
|
|
488 | } |
|
|
489 | #endif |
|
|
490 | if (enc->flags & F_CONV_BLESSED) |
|
|
491 | { |
|
|
492 | // we re-bless the reference to get overload and other niceties right |
|
|
493 | GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1); |
|
|
494 | |
|
|
495 | if (to_json) |
|
|
496 | { |
|
|
497 | dSP; |
|
|
498 | ENTER; |
|
|
499 | SAVETMPS; |
|
|
500 | PUSHMARK (SP); |
|
|
501 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); |
|
|
502 | |
|
|
503 | // calling with G_SCALAR ensures that we always get a 1 reutrn value |
|
|
504 | // check anyways. |
|
|
505 | PUTBACK; |
|
|
506 | assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR)); |
|
|
507 | SPAGAIN; |
|
|
508 | |
|
|
509 | encode_sv (enc, POPs); |
|
|
510 | |
|
|
511 | FREETMPS; |
|
|
512 | LEAVE; |
|
|
513 | } |
|
|
514 | else if (enc->flags & F_ALLOW_BLESSED) |
|
|
515 | encode_str (enc, "null", 4, 0); |
|
|
516 | else |
|
|
517 | croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it", |
|
|
518 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
519 | } |
|
|
520 | else if (enc->flags & F_ALLOW_BLESSED) |
|
|
521 | encode_str (enc, "null", 4, 0); |
|
|
522 | else |
|
|
523 | croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", |
|
|
524 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
525 | } |
|
|
526 | } |
446 | if (svt == SVt_PVHV) |
527 | else if (svt == SVt_PVHV) |
447 | encode_hv (enc, (HV *)sv); |
528 | encode_hv (enc, (HV *)sv); |
448 | else if (svt == SVt_PVAV) |
529 | else if (svt == SVt_PVAV) |
449 | encode_av (enc, (AV *)sv); |
530 | encode_av (enc, (AV *)sv); |
450 | else if (svt < SVt_PVAV) |
531 | else if (svt < SVt_PVAV) |
451 | { |
532 | { |
… | |
… | |
475 | encode_str (enc, str, len, SvUTF8 (sv)); |
556 | encode_str (enc, str, len, SvUTF8 (sv)); |
476 | encode_ch (enc, '"'); |
557 | encode_ch (enc, '"'); |
477 | } |
558 | } |
478 | else if (SvNOKp (sv)) |
559 | else if (SvNOKp (sv)) |
479 | { |
560 | { |
|
|
561 | // trust that perl will do the right thing w.r.t. JSON syntax. |
480 | need (enc, NV_DIG + 32); |
562 | need (enc, NV_DIG + 32); |
481 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
563 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
482 | enc->cur += strlen (enc->cur); |
564 | enc->cur += strlen (enc->cur); |
483 | } |
565 | } |
484 | else if (SvIOKp (sv)) |
566 | else if (SvIOKp (sv)) |
485 | { |
567 | { |
486 | need (enc, 64); |
568 | // we assume we can always read an IV as a UV |
|
|
569 | if (SvUV (sv) & ~(UV)0x7fff) |
|
|
570 | { |
|
|
571 | // large integer, use the (rather slow) snprintf way. |
|
|
572 | need (enc, sizeof (UV) * 3); |
487 | enc->cur += |
573 | enc->cur += |
488 | SvIsUV(sv) |
574 | SvIsUV(sv) |
489 | ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) |
575 | ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv)) |
490 | : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); |
576 | : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv)); |
|
|
577 | } |
|
|
578 | else |
|
|
579 | { |
|
|
580 | // optimise the "small number case" |
|
|
581 | // code will likely be branchless and use only a single multiplication |
|
|
582 | I32 i = SvIV (sv); |
|
|
583 | U32 u; |
|
|
584 | char digit, nz = 0; |
|
|
585 | |
|
|
586 | need (enc, 6); |
|
|
587 | |
|
|
588 | *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0; |
|
|
589 | u = i < 0 ? -i : i; |
|
|
590 | |
|
|
591 | // convert to 4.28 fixed-point representation |
|
|
592 | u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits |
|
|
593 | |
|
|
594 | // now output digit by digit, each time masking out the integer part |
|
|
595 | // and multiplying by 5 while moving the decimal point one to the right, |
|
|
596 | // resulting in a net multiplication by 10. |
|
|
597 | // we always write the digit to memory but conditionally increment |
|
|
598 | // the pointer, to ease the usage of conditional move instructions. |
|
|
599 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; |
|
|
600 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; |
|
|
601 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; |
|
|
602 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; |
|
|
603 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
|
|
604 | } |
491 | } |
605 | } |
492 | else if (SvROK (sv)) |
606 | else if (SvROK (sv)) |
493 | encode_rv (enc, SvRV (sv)); |
607 | encode_rv (enc, SvRV (sv)); |
494 | else if (!SvOK (sv)) |
608 | else if (!SvOK (sv)) |
495 | encode_str (enc, "null", 4, 0); |
609 | encode_str (enc, "null", 4, 0); |
… | |
… | |
514 | enc.maxdepth = DEC_DEPTH (flags); |
628 | enc.maxdepth = DEC_DEPTH (flags); |
515 | |
629 | |
516 | SvPOK_only (enc.sv); |
630 | SvPOK_only (enc.sv); |
517 | encode_sv (&enc, scalar); |
631 | encode_sv (&enc, scalar); |
518 | |
632 | |
519 | if (!(flags & (F_ASCII | F_UTF8))) |
|
|
520 | SvUTF8_on (enc.sv); |
|
|
521 | |
|
|
522 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
633 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
523 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
634 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
|
|
635 | |
|
|
636 | if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
|
|
637 | SvUTF8_on (enc.sv); |
524 | |
638 | |
525 | if (enc.flags & F_SHRINK) |
639 | if (enc.flags & F_SHRINK) |
526 | shrink (enc.sv); |
640 | shrink (enc.sv); |
527 | |
641 | |
528 | return enc.sv; |
642 | return enc.sv; |
… | |
… | |
540 | U32 flags; // F_* |
654 | U32 flags; // F_* |
541 | U32 depth; // recursion depth |
655 | U32 depth; // recursion depth |
542 | U32 maxdepth; // recursion depth limit |
656 | U32 maxdepth; // recursion depth limit |
543 | } dec_t; |
657 | } dec_t; |
544 | |
658 | |
545 | static void |
659 | inline void |
546 | decode_ws (dec_t *dec) |
660 | decode_ws (dec_t *dec) |
547 | { |
661 | { |
548 | for (;;) |
662 | for (;;) |
549 | { |
663 | { |
550 | char ch = *dec->cur; |
664 | char ch = *dec->cur; |
… | |
… | |
576 | decode_4hex (dec_t *dec) |
690 | decode_4hex (dec_t *dec) |
577 | { |
691 | { |
578 | signed char d1, d2, d3, d4; |
692 | signed char d1, d2, d3, d4; |
579 | unsigned char *cur = (unsigned char *)dec->cur; |
693 | unsigned char *cur = (unsigned char *)dec->cur; |
580 | |
694 | |
581 | d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); |
695 | d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected"); |
582 | d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); |
696 | d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected"); |
583 | d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); |
697 | d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected"); |
584 | d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); |
698 | d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected"); |
585 | |
699 | |
586 | dec->cur += 4; |
700 | dec->cur += 4; |
587 | |
701 | |
588 | return ((UV)d1) << 12 |
702 | return ((UV)d1) << 12 |
589 | | ((UV)d2) << 8 |
703 | | ((UV)d2) << 8 |
… | |
… | |
597 | static SV * |
711 | static SV * |
598 | decode_str (dec_t *dec) |
712 | decode_str (dec_t *dec) |
599 | { |
713 | { |
600 | SV *sv = 0; |
714 | SV *sv = 0; |
601 | int utf8 = 0; |
715 | int utf8 = 0; |
|
|
716 | char *dec_cur = dec->cur; |
602 | |
717 | |
603 | do |
718 | do |
604 | { |
719 | { |
605 | char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; |
720 | char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; |
606 | char *cur = buf; |
721 | char *cur = buf; |
607 | |
722 | |
608 | do |
723 | do |
609 | { |
724 | { |
610 | unsigned char ch = *(unsigned char *)dec->cur++; |
725 | unsigned char ch = *(unsigned char *)dec_cur++; |
611 | |
726 | |
612 | if (ch == '"') |
727 | if (expect_false (ch == '"')) |
613 | { |
728 | { |
614 | --dec->cur; |
729 | --dec_cur; |
615 | break; |
730 | break; |
616 | } |
731 | } |
617 | else if (ch == '\\') |
732 | else if (expect_false (ch == '\\')) |
618 | { |
733 | { |
619 | switch (*dec->cur) |
734 | switch (*dec_cur) |
620 | { |
735 | { |
621 | case '\\': |
736 | case '\\': |
622 | case '/': |
737 | case '/': |
623 | case '"': *cur++ = *dec->cur++; break; |
738 | case '"': *cur++ = *dec_cur++; break; |
624 | |
739 | |
625 | case 'b': ++dec->cur; *cur++ = '\010'; break; |
740 | case 'b': ++dec_cur; *cur++ = '\010'; break; |
626 | case 't': ++dec->cur; *cur++ = '\011'; break; |
741 | case 't': ++dec_cur; *cur++ = '\011'; break; |
627 | case 'n': ++dec->cur; *cur++ = '\012'; break; |
742 | case 'n': ++dec_cur; *cur++ = '\012'; break; |
628 | case 'f': ++dec->cur; *cur++ = '\014'; break; |
743 | case 'f': ++dec_cur; *cur++ = '\014'; break; |
629 | case 'r': ++dec->cur; *cur++ = '\015'; break; |
744 | case 'r': ++dec_cur; *cur++ = '\015'; break; |
630 | |
745 | |
631 | case 'u': |
746 | case 'u': |
632 | { |
747 | { |
633 | UV lo, hi; |
748 | UV lo, hi; |
634 | ++dec->cur; |
749 | ++dec_cur; |
635 | |
750 | |
|
|
751 | dec->cur = dec_cur; |
636 | hi = decode_4hex (dec); |
752 | hi = decode_4hex (dec); |
|
|
753 | dec_cur = dec->cur; |
637 | if (hi == (UV)-1) |
754 | if (hi == (UV)-1) |
638 | goto fail; |
755 | goto fail; |
639 | |
756 | |
640 | // possibly a surrogate pair |
757 | // possibly a surrogate pair |
641 | if (hi >= 0xd800) |
758 | if (hi >= 0xd800) |
642 | if (hi < 0xdc00) |
759 | if (hi < 0xdc00) |
643 | { |
760 | { |
644 | if (dec->cur [0] != '\\' || dec->cur [1] != 'u') |
761 | if (dec_cur [0] != '\\' || dec_cur [1] != 'u') |
645 | ERR ("missing low surrogate character in surrogate pair"); |
762 | ERR ("missing low surrogate character in surrogate pair"); |
646 | |
763 | |
647 | dec->cur += 2; |
764 | dec_cur += 2; |
648 | |
765 | |
|
|
766 | dec->cur = dec_cur; |
649 | lo = decode_4hex (dec); |
767 | lo = decode_4hex (dec); |
|
|
768 | dec_cur = dec->cur; |
650 | if (lo == (UV)-1) |
769 | if (lo == (UV)-1) |
651 | goto fail; |
770 | goto fail; |
652 | |
771 | |
653 | if (lo < 0xdc00 || lo >= 0xe000) |
772 | if (lo < 0xdc00 || lo >= 0xe000) |
654 | ERR ("surrogate pair expected"); |
773 | ERR ("surrogate pair expected"); |
… | |
… | |
668 | *cur++ = hi; |
787 | *cur++ = hi; |
669 | } |
788 | } |
670 | break; |
789 | break; |
671 | |
790 | |
672 | default: |
791 | default: |
673 | --dec->cur; |
792 | --dec_cur; |
674 | ERR ("illegal backslash escape sequence in string"); |
793 | ERR ("illegal backslash escape sequence in string"); |
675 | } |
794 | } |
676 | } |
795 | } |
677 | else if (ch >= 0x20 && ch <= 0x7f) |
796 | else if (expect_true (ch >= 0x20 && ch <= 0x7f)) |
678 | *cur++ = ch; |
797 | *cur++ = ch; |
679 | else if (ch >= 0x80) |
798 | else if (ch >= 0x80) |
680 | { |
799 | { |
681 | STRLEN clen; |
800 | STRLEN clen; |
682 | UV uch; |
801 | UV uch; |
683 | |
802 | |
684 | --dec->cur; |
803 | --dec_cur; |
685 | |
804 | |
686 | uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); |
805 | uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
687 | if (clen == (STRLEN)-1) |
806 | if (clen == (STRLEN)-1) |
688 | ERR ("malformed UTF-8 character in JSON string"); |
807 | ERR ("malformed UTF-8 character in JSON string"); |
689 | |
808 | |
690 | do |
809 | do |
691 | *cur++ = *dec->cur++; |
810 | *cur++ = *dec_cur++; |
692 | while (--clen); |
811 | while (--clen); |
693 | |
812 | |
694 | utf8 = 1; |
813 | utf8 = 1; |
695 | } |
814 | } |
696 | else |
815 | else |
697 | { |
816 | { |
698 | --dec->cur; |
817 | --dec_cur; |
699 | |
818 | |
700 | if (!ch) |
819 | if (!ch) |
701 | ERR ("unexpected end of string while parsing JSON string"); |
820 | ERR ("unexpected end of string while parsing JSON string"); |
702 | else |
821 | else |
703 | ERR ("invalid character encountered while parsing JSON string"); |
822 | ERR ("invalid character encountered while parsing JSON string"); |
… | |
… | |
716 | } |
835 | } |
717 | else |
836 | else |
718 | sv = newSVpvn (buf, len); |
837 | sv = newSVpvn (buf, len); |
719 | } |
838 | } |
720 | } |
839 | } |
721 | while (*dec->cur != '"'); |
840 | while (*dec_cur != '"'); |
722 | |
841 | |
723 | ++dec->cur; |
842 | ++dec_cur; |
724 | |
843 | |
725 | if (sv) |
844 | if (sv) |
726 | { |
845 | { |
727 | SvPOK_only (sv); |
846 | SvPOK_only (sv); |
728 | *SvEND (sv) = 0; |
847 | *SvEND (sv) = 0; |
… | |
… | |
731 | SvUTF8_on (sv); |
850 | SvUTF8_on (sv); |
732 | } |
851 | } |
733 | else |
852 | else |
734 | sv = newSVpvn ("", 0); |
853 | sv = newSVpvn ("", 0); |
735 | |
854 | |
|
|
855 | dec->cur = dec_cur; |
736 | return sv; |
856 | return sv; |
737 | |
857 | |
738 | fail: |
858 | fail: |
|
|
859 | dec->cur = dec_cur; |
739 | return 0; |
860 | return 0; |
740 | } |
861 | } |
741 | |
862 | |
742 | static SV * |
863 | static SV * |
743 | decode_num (dec_t *dec) |
864 | decode_num (dec_t *dec) |
… | |
… | |
801 | is_nv = 1; |
922 | is_nv = 1; |
802 | } |
923 | } |
803 | |
924 | |
804 | if (!is_nv) |
925 | if (!is_nv) |
805 | { |
926 | { |
806 | UV uv; |
927 | // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so |
807 | int numtype = grok_number (start, dec->cur - start, &uv); |
928 | if (*start == '-') |
808 | if (numtype & IS_NUMBER_IN_UV) |
929 | switch (dec->cur - start) |
809 | if (numtype & IS_NUMBER_NEG) |
|
|
810 | { |
930 | { |
811 | if (uv < (UV)IV_MIN) |
931 | case 2: return newSViv (-( start [1] - '0' * 1)); |
812 | return newSViv (-(IV)uv); |
932 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
|
|
933 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
|
|
934 | case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
813 | } |
935 | } |
|
|
936 | else |
|
|
937 | switch (dec->cur - start) |
|
|
938 | { |
|
|
939 | case 1: return newSViv ( start [0] - '0' * 1); |
|
|
940 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
|
|
941 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
|
|
942 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
|
|
943 | } |
|
|
944 | |
|
|
945 | { |
|
|
946 | UV uv; |
|
|
947 | int numtype = grok_number (start, dec->cur - start, &uv); |
|
|
948 | if (numtype & IS_NUMBER_IN_UV) |
|
|
949 | if (numtype & IS_NUMBER_NEG) |
|
|
950 | { |
|
|
951 | if (uv < (UV)IV_MIN) |
|
|
952 | return newSViv (-(IV)uv); |
|
|
953 | } |
814 | else |
954 | else |
815 | return newSVuv (uv); |
955 | return newSVuv (uv); |
|
|
956 | |
|
|
957 | // here would likely be the place for bigint support |
816 | } |
958 | } |
|
|
959 | } |
817 | |
960 | |
|
|
961 | // if we ever support bigint or bigfloat, this is the place for bigfloat |
818 | return newSVnv (Atof (start)); |
962 | return newSVnv (Atof (start)); |
819 | |
963 | |
820 | fail: |
964 | fail: |
821 | return 0; |
965 | return 0; |
822 | } |
966 | } |
… | |
… | |
923 | |
1067 | |
924 | static SV * |
1068 | static SV * |
925 | decode_sv (dec_t *dec) |
1069 | decode_sv (dec_t *dec) |
926 | { |
1070 | { |
927 | decode_ws (dec); |
1071 | decode_ws (dec); |
|
|
1072 | |
|
|
1073 | // the beauty of JSON: you need exactly one character lookahead |
|
|
1074 | // to parse anything. |
928 | switch (*dec->cur) |
1075 | switch (*dec->cur) |
929 | { |
1076 | { |
930 | case '"': ++dec->cur; return decode_str (dec); |
1077 | case '"': ++dec->cur; return decode_str (dec); |
931 | case '[': ++dec->cur; return decode_av (dec); |
1078 | case '[': ++dec->cur; return decode_av (dec); |
932 | case '{': ++dec->cur; return decode_hv (dec); |
1079 | case '{': ++dec->cur; return decode_hv (dec); |
… | |
… | |
938 | |
1085 | |
939 | case 't': |
1086 | case 't': |
940 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1087 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
941 | { |
1088 | { |
942 | dec->cur += 4; |
1089 | dec->cur += 4; |
943 | return newSViv (1); |
1090 | return SvREFCNT_inc (json_true); |
944 | } |
1091 | } |
945 | else |
1092 | else |
946 | ERR ("'true' expected"); |
1093 | ERR ("'true' expected"); |
947 | |
1094 | |
948 | break; |
1095 | break; |
949 | |
1096 | |
950 | case 'f': |
1097 | case 'f': |
951 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1098 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
952 | { |
1099 | { |
953 | dec->cur += 5; |
1100 | dec->cur += 5; |
954 | return newSViv (0); |
1101 | return SvREFCNT_inc (json_false); |
955 | } |
1102 | } |
956 | else |
1103 | else |
957 | ERR ("'false' expected"); |
1104 | ERR ("'false' expected"); |
958 | |
1105 | |
959 | break; |
1106 | break; |
… | |
… | |
977 | fail: |
1124 | fail: |
978 | return 0; |
1125 | return 0; |
979 | } |
1126 | } |
980 | |
1127 | |
981 | static SV * |
1128 | static SV * |
982 | decode_json (SV *string, U32 flags) |
1129 | decode_json (SV *string, U32 flags, UV *offset_return) |
983 | { |
1130 | { |
984 | dec_t dec; |
1131 | dec_t dec; |
|
|
1132 | UV offset; |
985 | SV *sv; |
1133 | SV *sv; |
986 | |
1134 | |
987 | SvGETMAGIC (string); |
1135 | SvGETMAGIC (string); |
988 | SvUPGRADE (string, SVt_PV); |
1136 | SvUPGRADE (string, SVt_PV); |
989 | |
1137 | |
… | |
… | |
999 | dec.end = SvEND (string); |
1147 | dec.end = SvEND (string); |
1000 | dec.err = 0; |
1148 | dec.err = 0; |
1001 | dec.depth = 0; |
1149 | dec.depth = 0; |
1002 | dec.maxdepth = DEC_DEPTH (dec.flags); |
1150 | dec.maxdepth = DEC_DEPTH (dec.flags); |
1003 | |
1151 | |
1004 | *dec.end = 0; // this should basically be a nop, too, but make sure its there |
1152 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1005 | sv = decode_sv (&dec); |
1153 | sv = decode_sv (&dec); |
1006 | |
1154 | |
|
|
1155 | if (!(offset_return || !sv)) |
|
|
1156 | { |
|
|
1157 | // check for trailing garbage |
|
|
1158 | decode_ws (&dec); |
|
|
1159 | |
|
|
1160 | if (*dec.cur) |
|
|
1161 | { |
|
|
1162 | dec.err = "garbage after JSON object"; |
|
|
1163 | SvREFCNT_dec (sv); |
|
|
1164 | sv = 0; |
|
|
1165 | } |
|
|
1166 | } |
|
|
1167 | |
|
|
1168 | if (offset_return || !sv) |
|
|
1169 | { |
|
|
1170 | offset = dec.flags & F_UTF8 |
|
|
1171 | ? dec.cur - SvPVX (string) |
|
|
1172 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
1173 | |
|
|
1174 | if (offset_return) |
|
|
1175 | *offset_return = offset; |
|
|
1176 | } |
|
|
1177 | |
1007 | if (!sv) |
1178 | if (!sv) |
1008 | { |
1179 | { |
1009 | IV offset = dec.flags & F_UTF8 |
|
|
1010 | ? dec.cur - SvPVX (string) |
|
|
1011 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
1012 | SV *uni = sv_newmortal (); |
1180 | SV *uni = sv_newmortal (); |
1013 | |
1181 | |
1014 | // horrible hack to silence warning inside pv_uni_display |
1182 | // horrible hack to silence warning inside pv_uni_display |
1015 | COP cop = *PL_curcop; |
1183 | COP cop = *PL_curcop; |
1016 | cop.cop_warnings = pWARN_NONE; |
1184 | cop.cop_warnings = pWARN_NONE; |
… | |
… | |
1041 | |
1209 | |
1042 | BOOT: |
1210 | BOOT: |
1043 | { |
1211 | { |
1044 | int i; |
1212 | int i; |
1045 | |
1213 | |
1046 | memset (decode_hexdigit, 0xff, 256); |
|
|
1047 | |
|
|
1048 | for (i = 0; i < 256; ++i) |
1214 | for (i = 0; i < 256; ++i) |
1049 | decode_hexdigit [i] = |
1215 | decode_hexdigit [i] = |
1050 | i >= '0' && i <= '9' ? i - '0' |
1216 | i >= '0' && i <= '9' ? i - '0' |
1051 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1217 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1052 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1218 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1053 | : -1; |
1219 | : -1; |
1054 | |
1220 | |
1055 | json_stash = gv_stashpv ("JSON::XS", 1); |
1221 | json_stash = gv_stashpv ("JSON::XS" , 1); |
|
|
1222 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
|
|
1223 | |
|
|
1224 | json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); |
|
|
1225 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1056 | } |
1226 | } |
1057 | |
1227 | |
1058 | PROTOTYPES: DISABLE |
1228 | PROTOTYPES: DISABLE |
1059 | |
1229 | |
1060 | SV *new (char *dummy) |
1230 | SV *new (char *dummy) |
… | |
… | |
1063 | OUTPUT: |
1233 | OUTPUT: |
1064 | RETVAL |
1234 | RETVAL |
1065 | |
1235 | |
1066 | SV *ascii (SV *self, int enable = 1) |
1236 | SV *ascii (SV *self, int enable = 1) |
1067 | ALIAS: |
1237 | ALIAS: |
1068 | ascii = F_ASCII |
1238 | ascii = F_ASCII |
|
|
1239 | latin1 = F_LATIN1 |
1069 | utf8 = F_UTF8 |
1240 | utf8 = F_UTF8 |
1070 | indent = F_INDENT |
1241 | indent = F_INDENT |
1071 | canonical = F_CANONICAL |
1242 | canonical = F_CANONICAL |
1072 | space_before = F_SPACE_BEFORE |
1243 | space_before = F_SPACE_BEFORE |
1073 | space_after = F_SPACE_AFTER |
1244 | space_after = F_SPACE_AFTER |
1074 | pretty = F_PRETTY |
1245 | pretty = F_PRETTY |
1075 | allow_nonref = F_ALLOW_NONREF |
1246 | allow_nonref = F_ALLOW_NONREF |
1076 | shrink = F_SHRINK |
1247 | shrink = F_SHRINK |
|
|
1248 | allow_blessed = F_ALLOW_BLESSED |
|
|
1249 | convert_blessed = F_CONV_BLESSED |
1077 | CODE: |
1250 | CODE: |
1078 | { |
1251 | { |
1079 | UV *uv = SvJSON (self); |
1252 | UV *uv = SvJSON (self); |
1080 | if (enable) |
1253 | if (enable) |
1081 | *uv |= ix; |
1254 | *uv |= ix; |
… | |
… | |
1109 | PPCODE: |
1282 | PPCODE: |
1110 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
1283 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
1111 | |
1284 | |
1112 | void decode (SV *self, SV *jsonstr) |
1285 | void decode (SV *self, SV *jsonstr) |
1113 | PPCODE: |
1286 | PPCODE: |
1114 | XPUSHs (decode_json (jsonstr, *SvJSON (self))); |
1287 | XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); |
|
|
1288 | |
|
|
1289 | void decode_prefix (SV *self, SV *jsonstr) |
|
|
1290 | PPCODE: |
|
|
1291 | { |
|
|
1292 | UV offset; |
|
|
1293 | EXTEND (SP, 2); |
|
|
1294 | PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); |
|
|
1295 | PUSHs (sv_2mortal (newSVuv (offset))); |
|
|
1296 | } |
1115 | |
1297 | |
1116 | PROTOTYPES: ENABLE |
1298 | PROTOTYPES: ENABLE |
1117 | |
1299 | |
1118 | void to_json (SV *scalar) |
1300 | void to_json (SV *scalar) |
1119 | ALIAS: |
1301 | ALIAS: |
… | |
… | |
1123 | |
1305 | |
1124 | void from_json (SV *jsonstr) |
1306 | void from_json (SV *jsonstr) |
1125 | ALIAS: |
1307 | ALIAS: |
1126 | jsonToObj = 0 |
1308 | jsonToObj = 0 |
1127 | PPCODE: |
1309 | PPCODE: |
1128 | XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); |
1310 | XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0)); |
1129 | |
1311 | |