… | |
… | |
4 | |
4 | |
5 | #include <assert.h> |
5 | #include <assert.h> |
6 | #include <string.h> |
6 | #include <string.h> |
7 | #include <stdlib.h> |
7 | #include <stdlib.h> |
8 | #include <stdio.h> |
8 | #include <stdio.h> |
|
|
9 | #include <limits.h> |
9 | #include <float.h> |
10 | #include <float.h> |
10 | |
11 | |
11 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
12 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
12 | # define snprintf _snprintf // C compilers have this in stdio.h |
13 | # define snprintf _snprintf // C compilers have this in stdio.h |
13 | #endif |
14 | #endif |
… | |
… | |
15 | // some old perls do not have this, try to make it work, no |
16 | // some old perls do not have this, try to make it work, no |
16 | // guarentees, though. if it breaks, you get to keep the pieces. |
17 | // guarentees, though. if it breaks, you get to keep the pieces. |
17 | #ifndef UTF8_MAXBYTES |
18 | #ifndef UTF8_MAXBYTES |
18 | # define UTF8_MAXBYTES 13 |
19 | # define UTF8_MAXBYTES 13 |
19 | #endif |
20 | #endif |
|
|
21 | |
|
|
22 | #define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2) |
20 | |
23 | |
21 | #define F_ASCII 0x00000001UL |
24 | #define F_ASCII 0x00000001UL |
22 | #define F_LATIN1 0x00000002UL |
25 | #define F_LATIN1 0x00000002UL |
23 | #define F_UTF8 0x00000004UL |
26 | #define F_UTF8 0x00000004UL |
24 | #define F_INDENT 0x00000008UL |
27 | #define F_INDENT 0x00000008UL |
… | |
… | |
50 | |
53 | |
51 | #define SB do { |
54 | #define SB do { |
52 | #define SE } while (0) |
55 | #define SE } while (0) |
53 | |
56 | |
54 | #if __GNUC__ >= 3 |
57 | #if __GNUC__ >= 3 |
55 | # define expect(expr,value) __builtin_expect ((expr),(value)) |
58 | # define expect(expr,value) __builtin_expect ((expr), (value)) |
56 | # define inline inline |
59 | # define INLINE static inline |
57 | #else |
60 | #else |
58 | # define expect(expr,value) (expr) |
61 | # define expect(expr,value) (expr) |
59 | # define inline static |
62 | # define INLINE static |
60 | #endif |
63 | #endif |
61 | |
64 | |
62 | #define expect_false(expr) expect ((expr) != 0, 0) |
65 | #define expect_false(expr) expect ((expr) != 0, 0) |
63 | #define expect_true(expr) expect ((expr) != 0, 1) |
66 | #define expect_true(expr) expect ((expr) != 0, 1) |
|
|
67 | |
|
|
68 | #define IN_RANGE_INC(type,val,beg,end) \ |
|
|
69 | ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
|
|
70 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
64 | |
71 | |
65 | #ifdef USE_ITHREADS |
72 | #ifdef USE_ITHREADS |
66 | # define JSON_SLOW 1 |
73 | # define JSON_SLOW 1 |
67 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
74 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
68 | #else |
75 | #else |
… | |
… | |
71 | #endif |
78 | #endif |
72 | |
79 | |
73 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
80 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
74 | static SV *json_true, *json_false; |
81 | static SV *json_true, *json_false; |
75 | |
82 | |
|
|
83 | enum { |
|
|
84 | INCR_M_WS = 0, // initial whitespace skipping, must be 0 |
|
|
85 | INCR_M_STR, // inside string |
|
|
86 | INCR_M_BS, // inside backslash |
|
|
87 | INCR_M_JSON // outside anything, count nesting |
|
|
88 | }; |
|
|
89 | |
|
|
90 | #define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON) |
|
|
91 | |
76 | typedef struct { |
92 | typedef struct { |
77 | U32 flags; |
93 | U32 flags; |
78 | SV *cb_object; |
94 | SV *cb_object; |
79 | HV *cb_sk_object; |
95 | HV *cb_sk_object; |
|
|
96 | |
|
|
97 | // for the incremental parser |
|
|
98 | SV *incr_text; // the source text so far |
|
|
99 | STRLEN incr_pos; // the current offset into the text |
|
|
100 | int incr_nest; // {[]}-nesting level |
|
|
101 | int incr_mode; |
80 | } JSON; |
102 | } JSON; |
81 | |
103 | |
82 | ///////////////////////////////////////////////////////////////////////////// |
104 | ///////////////////////////////////////////////////////////////////////////// |
83 | // utility functions |
105 | // utility functions |
84 | |
106 | |
85 | inline void |
107 | INLINE void |
86 | shrink (SV *sv) |
108 | shrink (SV *sv) |
87 | { |
109 | { |
88 | sv_utf8_downgrade (sv, 1); |
110 | sv_utf8_downgrade (sv, 1); |
89 | if (SvLEN (sv) > SvCUR (sv) + 1) |
111 | if (SvLEN (sv) > SvCUR (sv) + 1) |
90 | { |
112 | { |
… | |
… | |
99 | // decode an utf-8 character and return it, or (UV)-1 in |
121 | // decode an utf-8 character and return it, or (UV)-1 in |
100 | // case of an error. |
122 | // case of an error. |
101 | // we special-case "safe" characters from U+80 .. U+7FF, |
123 | // we special-case "safe" characters from U+80 .. U+7FF, |
102 | // but use the very good perl function to parse anything else. |
124 | // but use the very good perl function to parse anything else. |
103 | // note that we never call this function for a ascii codepoints |
125 | // note that we never call this function for a ascii codepoints |
104 | inline UV |
126 | INLINE UV |
105 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
127 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
106 | { |
128 | { |
107 | if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) |
129 | if (expect_true (len >= 2 |
108 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
130 | && IN_RANGE_INC (char, s[0], 0xc2, 0xdf) |
109 | else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
131 | && IN_RANGE_INC (char, s[1], 0x80, 0xbf))) |
110 | { |
132 | { |
111 | *clen = 2; |
133 | *clen = 2; |
112 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
134 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
113 | } |
135 | } |
114 | else |
136 | else |
115 | { |
137 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
116 | *clen = (STRLEN)-1; |
138 | } |
117 | return (UV)-1; |
139 | |
118 | } |
140 | // likewise for encoding, also never called for ascii codepoints |
|
|
141 | // this function takes advantage of this fact, although current gccs |
|
|
142 | // seem to optimise the check for >= 0x80 away anyways |
|
|
143 | INLINE unsigned char * |
|
|
144 | encode_utf8 (unsigned char *s, UV ch) |
|
|
145 | { |
|
|
146 | if (expect_false (ch < 0x000080)) |
|
|
147 | *s++ = ch; |
|
|
148 | else if (expect_true (ch < 0x000800)) |
|
|
149 | *s++ = 0xc0 | ( ch >> 6), |
|
|
150 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
151 | else if ( ch < 0x010000) |
|
|
152 | *s++ = 0xe0 | ( ch >> 12), |
|
|
153 | *s++ = 0x80 | ((ch >> 6) & 0x3f), |
|
|
154 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
155 | else if ( ch < 0x110000) |
|
|
156 | *s++ = 0xf0 | ( ch >> 18), |
|
|
157 | *s++ = 0x80 | ((ch >> 12) & 0x3f), |
|
|
158 | *s++ = 0x80 | ((ch >> 6) & 0x3f), |
|
|
159 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
160 | |
|
|
161 | return s; |
119 | } |
162 | } |
120 | |
163 | |
121 | ///////////////////////////////////////////////////////////////////////////// |
164 | ///////////////////////////////////////////////////////////////////////////// |
122 | // encoder |
165 | // encoder |
123 | |
166 | |
… | |
… | |
128 | char *end; // SvEND (sv) |
171 | char *end; // SvEND (sv) |
129 | SV *sv; // result scalar |
172 | SV *sv; // result scalar |
130 | JSON json; |
173 | JSON json; |
131 | U32 indent; // indentation level |
174 | U32 indent; // indentation level |
132 | U32 maxdepth; // max. indentation/recursion level |
175 | U32 maxdepth; // max. indentation/recursion level |
|
|
176 | UV limit; // escape character values >= this value when encoding |
133 | } enc_t; |
177 | } enc_t; |
134 | |
178 | |
135 | inline void |
179 | INLINE void |
136 | need (enc_t *enc, STRLEN len) |
180 | need (enc_t *enc, STRLEN len) |
137 | { |
181 | { |
138 | if (expect_false (enc->cur + len >= enc->end)) |
182 | if (expect_false (enc->cur + len >= enc->end)) |
139 | { |
183 | { |
140 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
184 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
… | |
… | |
142 | enc->cur = SvPVX (enc->sv) + cur; |
186 | enc->cur = SvPVX (enc->sv) + cur; |
143 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
187 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
144 | } |
188 | } |
145 | } |
189 | } |
146 | |
190 | |
147 | inline void |
191 | INLINE void |
148 | encode_ch (enc_t *enc, char ch) |
192 | encode_ch (enc_t *enc, char ch) |
149 | { |
193 | { |
150 | need (enc, 1); |
194 | need (enc, 1); |
151 | *enc->cur++ = ch; |
195 | *enc->cur++ = ch; |
152 | } |
196 | } |
… | |
… | |
206 | { |
250 | { |
207 | uch = ch; |
251 | uch = ch; |
208 | clen = 1; |
252 | clen = 1; |
209 | } |
253 | } |
210 | |
254 | |
211 | if (uch > 0x10FFFFUL) |
255 | if (uch < 0x80/*0x20*/ || uch >= enc->limit) |
212 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
|
|
213 | |
|
|
214 | if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF)) |
|
|
215 | { |
256 | { |
216 | if (uch > 0xFFFFUL) |
257 | if (uch >= 0x10000UL) |
217 | { |
258 | { |
|
|
259 | if (uch >= 0x110000UL) |
|
|
260 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
|
|
261 | |
218 | need (enc, len += 11); |
262 | need (enc, len += 11); |
219 | sprintf (enc->cur, "\\u%04x\\u%04x", |
263 | sprintf (enc->cur, "\\u%04x\\u%04x", |
220 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
264 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
221 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
265 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
222 | enc->cur += 12; |
266 | enc->cur += 12; |
… | |
… | |
250 | while (--clen); |
294 | while (--clen); |
251 | } |
295 | } |
252 | else |
296 | else |
253 | { |
297 | { |
254 | need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
298 | need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
255 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
299 | enc->cur = encode_utf8 (enc->cur, uch); |
256 | ++str; |
300 | ++str; |
257 | } |
301 | } |
258 | } |
302 | } |
259 | } |
303 | } |
260 | } |
304 | } |
261 | |
305 | |
262 | --len; |
306 | --len; |
263 | } |
307 | } |
264 | } |
308 | } |
265 | |
309 | |
266 | inline void |
310 | INLINE void |
267 | encode_indent (enc_t *enc) |
311 | encode_indent (enc_t *enc) |
268 | { |
312 | { |
269 | if (enc->json.flags & F_INDENT) |
313 | if (enc->json.flags & F_INDENT) |
270 | { |
314 | { |
271 | int spaces = enc->indent * INDENT_STEP; |
315 | int spaces = enc->indent * INDENT_STEP; |
… | |
… | |
274 | memset (enc->cur, ' ', spaces); |
318 | memset (enc->cur, ' ', spaces); |
275 | enc->cur += spaces; |
319 | enc->cur += spaces; |
276 | } |
320 | } |
277 | } |
321 | } |
278 | |
322 | |
279 | inline void |
323 | INLINE void |
280 | encode_space (enc_t *enc) |
324 | encode_space (enc_t *enc) |
281 | { |
325 | { |
282 | need (enc, 1); |
326 | need (enc, 1); |
283 | encode_ch (enc, ' '); |
327 | encode_ch (enc, ' '); |
284 | } |
328 | } |
285 | |
329 | |
286 | inline void |
330 | INLINE void |
287 | encode_nl (enc_t *enc) |
331 | encode_nl (enc_t *enc) |
288 | { |
332 | { |
289 | if (enc->json.flags & F_INDENT) |
333 | if (enc->json.flags & F_INDENT) |
290 | { |
334 | { |
291 | need (enc, 1); |
335 | need (enc, 1); |
292 | encode_ch (enc, '\n'); |
336 | encode_ch (enc, '\n'); |
293 | } |
337 | } |
294 | } |
338 | } |
295 | |
339 | |
296 | inline void |
340 | INLINE void |
297 | encode_comma (enc_t *enc) |
341 | encode_comma (enc_t *enc) |
298 | { |
342 | { |
299 | encode_ch (enc, ','); |
343 | encode_ch (enc, ','); |
300 | |
344 | |
301 | if (enc->json.flags & F_INDENT) |
345 | if (enc->json.flags & F_INDENT) |
… | |
… | |
394 | |
438 | |
395 | static void |
439 | static void |
396 | encode_hv (enc_t *enc, HV *hv) |
440 | encode_hv (enc_t *enc, HV *hv) |
397 | { |
441 | { |
398 | HE *he; |
442 | HE *he; |
399 | int count; |
|
|
400 | |
443 | |
401 | if (enc->indent >= enc->maxdepth) |
444 | if (enc->indent >= enc->maxdepth) |
402 | croak ("data structure too deep (hit recursion limit)"); |
445 | croak ("data structure too deep (hit recursion limit)"); |
403 | |
446 | |
404 | encode_ch (enc, '{'); |
447 | encode_ch (enc, '{'); |
… | |
… | |
619 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
662 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
620 | enc->cur += strlen (enc->cur); |
663 | enc->cur += strlen (enc->cur); |
621 | } |
664 | } |
622 | else if (SvIOKp (sv)) |
665 | else if (SvIOKp (sv)) |
623 | { |
666 | { |
624 | // we assume we can always read an IV as a UV |
667 | // we assume we can always read an IV as a UV and vice versa |
625 | if (SvUV (sv) & ~(UV)0x7fff) |
668 | // we assume two's complement |
626 | { |
669 | // we assume no aliasing issues in the union |
627 | // large integer, use the (rather slow) snprintf way. |
670 | if (SvIsUV (sv) ? SvUVX (sv) <= 59000 |
628 | need (enc, sizeof (UV) * 3); |
671 | : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000) |
629 | enc->cur += |
|
|
630 | SvIsUV(sv) |
|
|
631 | ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv)) |
|
|
632 | : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv)); |
|
|
633 | } |
|
|
634 | else |
|
|
635 | { |
672 | { |
636 | // optimise the "small number case" |
673 | // optimise the "small number case" |
637 | // code will likely be branchless and use only a single multiplication |
674 | // code will likely be branchless and use only a single multiplication |
|
|
675 | // works for numbers up to 59074 |
638 | I32 i = SvIV (sv); |
676 | I32 i = SvIVX (sv); |
639 | U32 u; |
677 | U32 u; |
640 | char digit, nz = 0; |
678 | char digit, nz = 0; |
641 | |
679 | |
642 | need (enc, 6); |
680 | need (enc, 6); |
643 | |
681 | |
… | |
… | |
649 | |
687 | |
650 | // now output digit by digit, each time masking out the integer part |
688 | // now output digit by digit, each time masking out the integer part |
651 | // and multiplying by 5 while moving the decimal point one to the right, |
689 | // and multiplying by 5 while moving the decimal point one to the right, |
652 | // resulting in a net multiplication by 10. |
690 | // resulting in a net multiplication by 10. |
653 | // we always write the digit to memory but conditionally increment |
691 | // we always write the digit to memory but conditionally increment |
654 | // the pointer, to ease the usage of conditional move instructions. |
692 | // the pointer, to enable the use of conditional move instructions. |
655 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; |
693 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5; |
656 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; |
694 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5; |
657 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; |
695 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5; |
658 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; |
696 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5; |
659 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
697 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
|
|
698 | } |
|
|
699 | else |
|
|
700 | { |
|
|
701 | // large integer, use the (rather slow) snprintf way. |
|
|
702 | need (enc, IVUV_MAXCHARS); |
|
|
703 | enc->cur += |
|
|
704 | SvIsUV(sv) |
|
|
705 | ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) |
|
|
706 | : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); |
660 | } |
707 | } |
661 | } |
708 | } |
662 | else if (SvROK (sv)) |
709 | else if (SvROK (sv)) |
663 | encode_rv (enc, SvRV (sv)); |
710 | encode_rv (enc, SvRV (sv)); |
664 | else if (!SvOK (sv)) |
711 | else if (!SvOK (sv)) |
… | |
… | |
680 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
727 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
681 | enc.cur = SvPVX (enc.sv); |
728 | enc.cur = SvPVX (enc.sv); |
682 | enc.end = SvEND (enc.sv); |
729 | enc.end = SvEND (enc.sv); |
683 | enc.indent = 0; |
730 | enc.indent = 0; |
684 | enc.maxdepth = DEC_DEPTH (enc.json.flags); |
731 | enc.maxdepth = DEC_DEPTH (enc.json.flags); |
|
|
732 | enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
|
|
733 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
|
|
734 | : 0x110000UL; |
685 | |
735 | |
686 | SvPOK_only (enc.sv); |
736 | SvPOK_only (enc.sv); |
687 | encode_sv (&enc, scalar); |
737 | encode_sv (&enc, scalar); |
688 | |
738 | |
689 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
739 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
… | |
… | |
710 | JSON json; |
760 | JSON json; |
711 | U32 depth; // recursion depth |
761 | U32 depth; // recursion depth |
712 | U32 maxdepth; // recursion depth limit |
762 | U32 maxdepth; // recursion depth limit |
713 | } dec_t; |
763 | } dec_t; |
714 | |
764 | |
715 | inline void |
765 | INLINE void |
716 | decode_comment (dec_t *dec) |
766 | decode_comment (dec_t *dec) |
717 | { |
767 | { |
718 | // only '#'-style comments allowed a.t.m. |
768 | // only '#'-style comments allowed a.t.m. |
719 | |
769 | |
720 | while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) |
770 | while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) |
721 | ++dec->cur; |
771 | ++dec->cur; |
722 | } |
772 | } |
723 | |
773 | |
724 | inline void |
774 | INLINE void |
725 | decode_ws (dec_t *dec) |
775 | decode_ws (dec_t *dec) |
726 | { |
776 | { |
727 | for (;;) |
777 | for (;;) |
728 | { |
778 | { |
729 | char ch = *dec->cur; |
779 | char ch = *dec->cur; |
… | |
… | |
855 | |
905 | |
856 | if (hi >= 0x80) |
906 | if (hi >= 0x80) |
857 | { |
907 | { |
858 | utf8 = 1; |
908 | utf8 = 1; |
859 | |
909 | |
860 | cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); |
910 | cur = encode_utf8 (cur, hi); |
861 | } |
911 | } |
862 | else |
912 | else |
863 | *cur++ = hi; |
913 | *cur++ = hi; |
864 | } |
914 | } |
865 | break; |
915 | break; |
… | |
… | |
867 | default: |
917 | default: |
868 | --dec_cur; |
918 | --dec_cur; |
869 | ERR ("illegal backslash escape sequence in string"); |
919 | ERR ("illegal backslash escape sequence in string"); |
870 | } |
920 | } |
871 | } |
921 | } |
872 | else if (expect_true (ch >= 0x20 && ch <= 0x7f)) |
922 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
873 | *cur++ = ch; |
923 | *cur++ = ch; |
874 | else if (ch >= 0x80) |
924 | else if (ch >= 0x80) |
875 | { |
925 | { |
876 | STRLEN clen; |
926 | STRLEN clen; |
877 | UV uch; |
927 | UV uch; |
… | |
… | |
1000 | |
1050 | |
1001 | if (!is_nv) |
1051 | if (!is_nv) |
1002 | { |
1052 | { |
1003 | int len = dec->cur - start; |
1053 | int len = dec->cur - start; |
1004 | |
1054 | |
1005 | // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so |
1055 | // special case the rather common 1..5-digit-int case |
1006 | if (*start == '-') |
1056 | if (*start == '-') |
1007 | switch (len) |
1057 | switch (len) |
1008 | { |
1058 | { |
1009 | case 2: return newSViv (-( start [1] - '0' * 1)); |
1059 | case 2: return newSViv (-( start [1] - '0' * 1)); |
1010 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
1060 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
1011 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
1061 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
1012 | case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
1062 | case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
|
|
1063 | case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
1013 | } |
1064 | } |
1014 | else |
1065 | else |
1015 | switch (len) |
1066 | switch (len) |
1016 | { |
1067 | { |
1017 | case 1: return newSViv ( start [0] - '0' * 1); |
1068 | case 1: return newSViv ( start [0] - '0' * 1); |
1018 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1069 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1019 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
1070 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
1020 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
1071 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
|
|
1072 | case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
1021 | } |
1073 | } |
1022 | |
1074 | |
1023 | { |
1075 | { |
1024 | UV uv; |
1076 | UV uv; |
1025 | int numtype = grok_number (start, len, &uv); |
1077 | int numtype = grok_number (start, len, &uv); |
… | |
… | |
1133 | char *p = dec->cur; |
1185 | char *p = dec->cur; |
1134 | char *e = p + 24; // only try up to 24 bytes |
1186 | char *e = p + 24; // only try up to 24 bytes |
1135 | |
1187 | |
1136 | for (;;) |
1188 | for (;;) |
1137 | { |
1189 | { |
1138 | // the >= 0x80 is true on most architectures |
1190 | // the >= 0x80 is false on most architectures |
1139 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1191 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1140 | { |
1192 | { |
1141 | // slow path, back up and use decode_str |
1193 | // slow path, back up and use decode_str |
1142 | SV *key = decode_str (dec); |
1194 | SV *key = decode_str (dec); |
1143 | if (!key) |
1195 | if (!key) |
… | |
… | |
1274 | |
1326 | |
1275 | static SV * |
1327 | static SV * |
1276 | decode_sv (dec_t *dec) |
1328 | decode_sv (dec_t *dec) |
1277 | { |
1329 | { |
1278 | // the beauty of JSON: you need exactly one character lookahead |
1330 | // the beauty of JSON: you need exactly one character lookahead |
1279 | // to parse anything. |
1331 | // to parse everything. |
1280 | switch (*dec->cur) |
1332 | switch (*dec->cur) |
1281 | { |
1333 | { |
1282 | case '"': ++dec->cur; return decode_str (dec); |
1334 | case '"': ++dec->cur; return decode_str (dec); |
1283 | case '[': ++dec->cur; return decode_av (dec); |
1335 | case '[': ++dec->cur; return decode_av (dec); |
1284 | case '{': ++dec->cur; return decode_hv (dec); |
1336 | case '{': ++dec->cur; return decode_hv (dec); |
1285 | |
1337 | |
1286 | case '-': |
1338 | case '-': |
1287 | case '0': case '1': case '2': case '3': case '4': |
1339 | case '0': case '1': case '2': case '3': case '4': |
1288 | case '5': case '6': case '7': case '8': case '9': |
1340 | case '5': case '6': case '7': case '8': case '9': |
1289 | return decode_num (dec); |
1341 | return decode_num (dec); |
… | |
… | |
1335 | fail: |
1387 | fail: |
1336 | return 0; |
1388 | return 0; |
1337 | } |
1389 | } |
1338 | |
1390 | |
1339 | static SV * |
1391 | static SV * |
1340 | decode_json (SV *string, JSON *json, UV *offset_return) |
1392 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1341 | { |
1393 | { |
1342 | dec_t dec; |
1394 | dec_t dec; |
1343 | UV offset; |
1395 | STRLEN offset; |
1344 | SV *sv; |
1396 | SV *sv; |
1345 | |
1397 | |
1346 | SvGETMAGIC (string); |
1398 | SvGETMAGIC (string); |
1347 | SvUPGRADE (string, SVt_PV); |
1399 | SvUPGRADE (string, SVt_PV); |
1348 | |
1400 | |
… | |
… | |
1421 | |
1473 | |
1422 | return sv; |
1474 | return sv; |
1423 | } |
1475 | } |
1424 | |
1476 | |
1425 | ///////////////////////////////////////////////////////////////////////////// |
1477 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1478 | // incremental parser |
|
|
1479 | |
|
|
1480 | static void |
|
|
1481 | incr_parse (JSON *self) |
|
|
1482 | { |
|
|
1483 | const char *p = SvPVX (self->incr_text) + self->incr_pos; |
|
|
1484 | |
|
|
1485 | for (;;) |
|
|
1486 | { |
|
|
1487 | //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D |
|
|
1488 | switch (self->incr_mode) |
|
|
1489 | { |
|
|
1490 | // only used for intiial whitespace skipping |
|
|
1491 | case INCR_M_WS: |
|
|
1492 | for (;;) |
|
|
1493 | { |
|
|
1494 | if (*p > 0x20) |
|
|
1495 | { |
|
|
1496 | self->incr_mode = INCR_M_JSON; |
|
|
1497 | goto incr_m_json; |
|
|
1498 | } |
|
|
1499 | else if (!*p) |
|
|
1500 | goto interrupt; |
|
|
1501 | |
|
|
1502 | ++p; |
|
|
1503 | } |
|
|
1504 | |
|
|
1505 | // skip a single char inside a string (for \\-processing) |
|
|
1506 | case INCR_M_BS: |
|
|
1507 | if (!*p) |
|
|
1508 | goto interrupt; |
|
|
1509 | |
|
|
1510 | ++p; |
|
|
1511 | self->incr_mode = INCR_M_STR; |
|
|
1512 | goto incr_m_str; |
|
|
1513 | |
|
|
1514 | // inside a string |
|
|
1515 | case INCR_M_STR: |
|
|
1516 | incr_m_str: |
|
|
1517 | for (;;) |
|
|
1518 | { |
|
|
1519 | if (*p == '"') |
|
|
1520 | { |
|
|
1521 | ++p; |
|
|
1522 | self->incr_mode = INCR_M_JSON; |
|
|
1523 | |
|
|
1524 | if (!self->incr_nest) |
|
|
1525 | goto interrupt; |
|
|
1526 | |
|
|
1527 | goto incr_m_json; |
|
|
1528 | } |
|
|
1529 | else if (*p == '\\') |
|
|
1530 | { |
|
|
1531 | ++p; // "virtually" consumes character after \ |
|
|
1532 | |
|
|
1533 | if (!*p) // if at end of string we have to switch modes |
|
|
1534 | { |
|
|
1535 | self->incr_mode = INCR_M_BS; |
|
|
1536 | goto interrupt; |
|
|
1537 | } |
|
|
1538 | } |
|
|
1539 | else if (!*p) |
|
|
1540 | goto interrupt; |
|
|
1541 | |
|
|
1542 | ++p; |
|
|
1543 | } |
|
|
1544 | |
|
|
1545 | // after initial ws, outside string |
|
|
1546 | case INCR_M_JSON: |
|
|
1547 | incr_m_json: |
|
|
1548 | for (;;) |
|
|
1549 | { |
|
|
1550 | switch (*p++) |
|
|
1551 | { |
|
|
1552 | case 0: |
|
|
1553 | --p; |
|
|
1554 | goto interrupt; |
|
|
1555 | |
|
|
1556 | case 0x09: |
|
|
1557 | case 0x0a: |
|
|
1558 | case 0x0d: |
|
|
1559 | case 0x20: |
|
|
1560 | if (!self->incr_nest) |
|
|
1561 | { |
|
|
1562 | --p; // do not eat the whitespace, let the next round do it |
|
|
1563 | goto interrupt; |
|
|
1564 | } |
|
|
1565 | break; |
|
|
1566 | |
|
|
1567 | case '"': |
|
|
1568 | self->incr_mode = INCR_M_STR; |
|
|
1569 | goto incr_m_str; |
|
|
1570 | |
|
|
1571 | case '[': |
|
|
1572 | case '{': |
|
|
1573 | ++self->incr_nest; |
|
|
1574 | break; |
|
|
1575 | |
|
|
1576 | case ']': |
|
|
1577 | case '}': |
|
|
1578 | if (!--self->incr_nest) |
|
|
1579 | goto interrupt; |
|
|
1580 | } |
|
|
1581 | } |
|
|
1582 | } |
|
|
1583 | |
|
|
1584 | modechange: |
|
|
1585 | ; |
|
|
1586 | } |
|
|
1587 | |
|
|
1588 | interrupt: |
|
|
1589 | self->incr_pos = p - SvPVX (self->incr_text); |
|
|
1590 | //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D |
|
|
1591 | } |
|
|
1592 | |
|
|
1593 | ///////////////////////////////////////////////////////////////////////////// |
1426 | // XS interface functions |
1594 | // XS interface functions |
1427 | |
1595 | |
1428 | MODULE = JSON::XS PACKAGE = JSON::XS |
1596 | MODULE = JSON::XS PACKAGE = JSON::XS |
1429 | |
1597 | |
1430 | BOOT: |
1598 | BOOT: |
… | |
… | |
1457 | { |
1625 | { |
1458 | SV *pv = NEWSV (0, sizeof (JSON)); |
1626 | SV *pv = NEWSV (0, sizeof (JSON)); |
1459 | SvPOK_only (pv); |
1627 | SvPOK_only (pv); |
1460 | Zero (SvPVX (pv), 1, JSON); |
1628 | Zero (SvPVX (pv), 1, JSON); |
1461 | ((JSON *)SvPVX (pv))->flags = F_DEFAULT; |
1629 | ((JSON *)SvPVX (pv))->flags = F_DEFAULT; |
1462 | XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), JSON_STASH))); |
1630 | XPUSHs (sv_2mortal (sv_bless ( |
|
|
1631 | newRV_noinc (pv), |
|
|
1632 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
|
|
1633 | ))); |
1463 | } |
1634 | } |
1464 | |
1635 | |
1465 | void ascii (JSON *self, int enable = 1) |
1636 | void ascii (JSON *self, int enable = 1) |
1466 | ALIAS: |
1637 | ALIAS: |
1467 | ascii = F_ASCII |
1638 | ascii = F_ASCII |
… | |
… | |
1494 | get_utf8 = F_UTF8 |
1665 | get_utf8 = F_UTF8 |
1495 | get_indent = F_INDENT |
1666 | get_indent = F_INDENT |
1496 | get_canonical = F_CANONICAL |
1667 | get_canonical = F_CANONICAL |
1497 | get_space_before = F_SPACE_BEFORE |
1668 | get_space_before = F_SPACE_BEFORE |
1498 | get_space_after = F_SPACE_AFTER |
1669 | get_space_after = F_SPACE_AFTER |
1499 | get_pretty = F_PRETTY |
|
|
1500 | get_allow_nonref = F_ALLOW_NONREF |
1670 | get_allow_nonref = F_ALLOW_NONREF |
1501 | get_shrink = F_SHRINK |
1671 | get_shrink = F_SHRINK |
1502 | get_allow_blessed = F_ALLOW_BLESSED |
1672 | get_allow_blessed = F_ALLOW_BLESSED |
1503 | get_convert_blessed = F_CONV_BLESSED |
1673 | get_convert_blessed = F_CONV_BLESSED |
1504 | get_relaxed = F_RELAXED |
1674 | get_relaxed = F_RELAXED |
… | |
… | |
1588 | XPUSHs (decode_json (jsonstr, self, 0)); |
1758 | XPUSHs (decode_json (jsonstr, self, 0)); |
1589 | |
1759 | |
1590 | void decode_prefix (JSON *self, SV *jsonstr) |
1760 | void decode_prefix (JSON *self, SV *jsonstr) |
1591 | PPCODE: |
1761 | PPCODE: |
1592 | { |
1762 | { |
1593 | UV offset; |
1763 | STRLEN offset; |
1594 | EXTEND (SP, 2); |
1764 | EXTEND (SP, 2); |
1595 | PUSHs (decode_json (jsonstr, self, &offset)); |
1765 | PUSHs (decode_json (jsonstr, self, &offset)); |
1596 | PUSHs (sv_2mortal (newSVuv (offset))); |
1766 | PUSHs (sv_2mortal (newSVuv (offset))); |
|
|
1767 | } |
|
|
1768 | |
|
|
1769 | void incr_parse (JSON *self, SV *jsonstr = 0) |
|
|
1770 | PPCODE: |
|
|
1771 | { |
|
|
1772 | if (!self->incr_text) |
|
|
1773 | self->incr_text = newSVpvn ("", 0); |
|
|
1774 | |
|
|
1775 | // append data, if any |
|
|
1776 | if (jsonstr) |
|
|
1777 | { |
|
|
1778 | if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) |
|
|
1779 | { |
|
|
1780 | /* utf-8-ness differs, need to upgrade */ |
|
|
1781 | sv_utf8_upgrade (self->incr_text); |
|
|
1782 | |
|
|
1783 | if (self->incr_pos) |
|
|
1784 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
|
|
1785 | - (U8 *)SvPVX (self->incr_text); |
|
|
1786 | } |
|
|
1787 | |
|
|
1788 | { |
|
|
1789 | STRLEN len; |
|
|
1790 | const char *str = SvPV (jsonstr, len); |
|
|
1791 | SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); |
|
|
1792 | Move (str, SvEND (self->incr_text), len, char); |
|
|
1793 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
|
|
1794 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
|
|
1795 | } |
|
|
1796 | } |
|
|
1797 | |
|
|
1798 | if (GIMME_V != G_VOID) |
|
|
1799 | do |
|
|
1800 | { |
|
|
1801 | STRLEN offset; |
|
|
1802 | |
|
|
1803 | incr_parse (self); |
|
|
1804 | |
|
|
1805 | if (!INCR_DONE (self)) |
|
|
1806 | break; |
|
|
1807 | |
|
|
1808 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
|
|
1809 | |
|
|
1810 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset); |
|
|
1811 | self->incr_pos -= offset; |
|
|
1812 | self->incr_nest = 0; |
|
|
1813 | self->incr_mode = 0; |
|
|
1814 | } |
|
|
1815 | while (GIMME_V == G_ARRAY); |
|
|
1816 | } |
|
|
1817 | |
|
|
1818 | SV *incr_text (JSON *self) |
|
|
1819 | ATTRS: lvalue |
|
|
1820 | CODE: |
|
|
1821 | { |
|
|
1822 | if (self->incr_pos) |
|
|
1823 | croak ("incr_text can only be called after a successful incr_parse call in scalar context"); |
|
|
1824 | |
|
|
1825 | RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef; |
|
|
1826 | } |
|
|
1827 | OUTPUT: |
|
|
1828 | RETVAL |
|
|
1829 | |
|
|
1830 | void incr_skip (JSON *self) |
|
|
1831 | CODE: |
|
|
1832 | { |
|
|
1833 | if (self->incr_pos) |
|
|
1834 | { |
|
|
1835 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos); |
|
|
1836 | self->incr_pos = 0; |
|
|
1837 | self->incr_nest = 0; |
|
|
1838 | self->incr_mode = 0; |
|
|
1839 | } |
1597 | } |
1840 | } |
1598 | |
1841 | |
1599 | void DESTROY (JSON *self) |
1842 | void DESTROY (JSON *self) |
1600 | CODE: |
1843 | CODE: |
1601 | SvREFCNT_dec (self->cb_sk_object); |
1844 | SvREFCNT_dec (self->cb_sk_object); |
1602 | SvREFCNT_dec (self->cb_object); |
1845 | SvREFCNT_dec (self->cb_object); |
|
|
1846 | SvREFCNT_dec (self->incr_text); |
1603 | |
1847 | |
1604 | PROTOTYPES: ENABLE |
1848 | PROTOTYPES: ENABLE |
1605 | |
1849 | |
1606 | void to_json (SV *scalar) |
1850 | void encode_json (SV *scalar) |
|
|
1851 | ALIAS: |
|
|
1852 | to_json_ = 0 |
|
|
1853 | encode_json = F_UTF8 |
1607 | PPCODE: |
1854 | PPCODE: |
1608 | { |
1855 | { |
1609 | JSON json = { F_DEFAULT | F_UTF8 }; |
1856 | JSON json = { F_DEFAULT | ix }; |
1610 | XPUSHs (encode_json (scalar, &json)); |
1857 | XPUSHs (encode_json (scalar, &json)); |
1611 | } |
1858 | } |
1612 | |
1859 | |
1613 | void from_json (SV *jsonstr) |
1860 | void decode_json (SV *jsonstr) |
|
|
1861 | ALIAS: |
|
|
1862 | from_json_ = 0 |
|
|
1863 | decode_json = F_UTF8 |
1614 | PPCODE: |
1864 | PPCODE: |
1615 | { |
1865 | { |
1616 | JSON json = { F_DEFAULT | F_UTF8 }; |
1866 | JSON json = { F_DEFAULT | ix }; |
1617 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1867 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1618 | } |
1868 | } |
1619 | |
1869 | |
|
|
1870 | |