… | |
… | |
4 | |
4 | |
5 | #include <assert.h> |
5 | #include <assert.h> |
6 | #include <string.h> |
6 | #include <string.h> |
7 | #include <stdlib.h> |
7 | #include <stdlib.h> |
8 | #include <stdio.h> |
8 | #include <stdio.h> |
|
|
9 | #include <limits.h> |
9 | #include <float.h> |
10 | #include <float.h> |
10 | |
11 | |
11 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
12 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
12 | # define snprintf _snprintf // C compilers have this in stdio.h |
13 | # define snprintf _snprintf // C compilers have this in stdio.h |
13 | #endif |
14 | #endif |
… | |
… | |
15 | // some old perls do not have this, try to make it work, no |
16 | // some old perls do not have this, try to make it work, no |
16 | // guarentees, though. if it breaks, you get to keep the pieces. |
17 | // guarentees, though. if it breaks, you get to keep the pieces. |
17 | #ifndef UTF8_MAXBYTES |
18 | #ifndef UTF8_MAXBYTES |
18 | # define UTF8_MAXBYTES 13 |
19 | # define UTF8_MAXBYTES 13 |
19 | #endif |
20 | #endif |
|
|
21 | |
|
|
22 | #define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2) |
20 | |
23 | |
21 | #define F_ASCII 0x00000001UL |
24 | #define F_ASCII 0x00000001UL |
22 | #define F_LATIN1 0x00000002UL |
25 | #define F_LATIN1 0x00000002UL |
23 | #define F_UTF8 0x00000004UL |
26 | #define F_UTF8 0x00000004UL |
24 | #define F_INDENT 0x00000008UL |
27 | #define F_INDENT 0x00000008UL |
… | |
… | |
27 | #define F_SPACE_AFTER 0x00000040UL |
30 | #define F_SPACE_AFTER 0x00000040UL |
28 | #define F_ALLOW_NONREF 0x00000100UL |
31 | #define F_ALLOW_NONREF 0x00000100UL |
29 | #define F_SHRINK 0x00000200UL |
32 | #define F_SHRINK 0x00000200UL |
30 | #define F_ALLOW_BLESSED 0x00000400UL |
33 | #define F_ALLOW_BLESSED 0x00000400UL |
31 | #define F_CONV_BLESSED 0x00000800UL |
34 | #define F_CONV_BLESSED 0x00000800UL |
32 | #define F_MAXDEPTH 0xf8000000UL |
|
|
33 | #define S_MAXDEPTH 27 |
|
|
34 | #define F_MAXSIZE 0x01f00000UL |
35 | #define F_RELAXED 0x00001000UL |
35 | #define S_MAXSIZE 20 |
36 | #define F_ALLOW_UNKNOWN 0x00002000UL |
36 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
37 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
37 | |
38 | |
38 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
|
|
39 | #define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE )) |
|
|
40 | |
|
|
41 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
39 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
42 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
|
|
43 | |
40 | |
44 | #define INIT_SIZE 32 // initial scalar size to be allocated |
41 | #define INIT_SIZE 32 // initial scalar size to be allocated |
45 | #define INDENT_STEP 3 // spaces per indentation level |
42 | #define INDENT_STEP 3 // spaces per indentation level |
46 | |
43 | |
47 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
44 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
48 | |
45 | |
49 | #define SB do { |
46 | #define SB do { |
50 | #define SE } while (0) |
47 | #define SE } while (0) |
51 | |
48 | |
52 | #if __GNUC__ >= 3 |
49 | #if __GNUC__ >= 3 |
53 | # define expect(expr,value) __builtin_expect ((expr),(value)) |
50 | # define expect(expr,value) __builtin_expect ((expr), (value)) |
54 | # define inline inline |
51 | # define INLINE static inline |
55 | #else |
52 | #else |
56 | # define expect(expr,value) (expr) |
53 | # define expect(expr,value) (expr) |
57 | # define inline static |
54 | # define INLINE static |
58 | #endif |
55 | #endif |
59 | |
56 | |
60 | #define expect_false(expr) expect ((expr) != 0, 0) |
57 | #define expect_false(expr) expect ((expr) != 0, 0) |
61 | #define expect_true(expr) expect ((expr) != 0, 1) |
58 | #define expect_true(expr) expect ((expr) != 0, 1) |
62 | |
59 | |
|
|
60 | #define IN_RANGE_INC(type,val,beg,end) \ |
|
|
61 | ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
|
|
62 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
|
|
63 | |
|
|
64 | #define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)" |
|
|
65 | |
|
|
66 | #ifdef USE_ITHREADS |
|
|
67 | # define JSON_SLOW 1 |
|
|
68 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
|
|
69 | #else |
|
|
70 | # define JSON_SLOW 0 |
|
|
71 | # define JSON_STASH json_stash |
|
|
72 | #endif |
|
|
73 | |
63 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
74 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
64 | static SV *json_true, *json_false; |
75 | static SV *json_true, *json_false; |
65 | |
76 | |
|
|
77 | enum { |
|
|
78 | INCR_M_WS = 0, // initial whitespace skipping, must be 0 |
|
|
79 | INCR_M_STR, // inside string |
|
|
80 | INCR_M_BS, // inside backslash |
|
|
81 | INCR_M_JSON // outside anything, count nesting |
|
|
82 | }; |
|
|
83 | |
|
|
84 | #define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) |
|
|
85 | |
66 | typedef struct { |
86 | typedef struct { |
67 | U32 flags; |
87 | U32 flags; |
|
|
88 | U32 max_depth; |
|
|
89 | STRLEN max_size; |
|
|
90 | |
68 | SV *cb_object; |
91 | SV *cb_object; |
69 | HV *cb_sk_object; |
92 | HV *cb_sk_object; |
|
|
93 | |
|
|
94 | // for the incremental parser |
|
|
95 | SV *incr_text; // the source text so far |
|
|
96 | STRLEN incr_pos; // the current offset into the text |
|
|
97 | int incr_nest; // {[]}-nesting level |
|
|
98 | unsigned char incr_mode; |
70 | } JSON; |
99 | } JSON; |
|
|
100 | |
|
|
101 | INLINE void |
|
|
102 | json_init (JSON *json) |
|
|
103 | { |
|
|
104 | Zero (json, 1, JSON); |
|
|
105 | json->max_depth = 512; |
|
|
106 | } |
71 | |
107 | |
72 | ///////////////////////////////////////////////////////////////////////////// |
108 | ///////////////////////////////////////////////////////////////////////////// |
73 | // utility functions |
109 | // utility functions |
74 | |
110 | |
75 | inline void |
111 | INLINE SV * |
|
|
112 | get_bool (const char *name) |
|
|
113 | { |
|
|
114 | SV *sv = get_sv (name, 1); |
|
|
115 | |
|
|
116 | SvREADONLY_on (sv); |
|
|
117 | SvREADONLY_on (SvRV (sv)); |
|
|
118 | |
|
|
119 | return sv; |
|
|
120 | } |
|
|
121 | |
|
|
122 | INLINE void |
76 | shrink (SV *sv) |
123 | shrink (SV *sv) |
77 | { |
124 | { |
78 | sv_utf8_downgrade (sv, 1); |
125 | sv_utf8_downgrade (sv, 1); |
|
|
126 | |
79 | if (SvLEN (sv) > SvCUR (sv) + 1) |
127 | if (SvLEN (sv) > SvCUR (sv) + 1) |
80 | { |
128 | { |
81 | #ifdef SvPV_shrink_to_cur |
129 | #ifdef SvPV_shrink_to_cur |
82 | SvPV_shrink_to_cur (sv); |
130 | SvPV_shrink_to_cur (sv); |
83 | #elif defined (SvPV_renew) |
131 | #elif defined (SvPV_renew) |
… | |
… | |
89 | // decode an utf-8 character and return it, or (UV)-1 in |
137 | // decode an utf-8 character and return it, or (UV)-1 in |
90 | // case of an error. |
138 | // case of an error. |
91 | // we special-case "safe" characters from U+80 .. U+7FF, |
139 | // we special-case "safe" characters from U+80 .. U+7FF, |
92 | // but use the very good perl function to parse anything else. |
140 | // but use the very good perl function to parse anything else. |
93 | // note that we never call this function for a ascii codepoints |
141 | // note that we never call this function for a ascii codepoints |
94 | inline UV |
142 | INLINE UV |
95 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
143 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
96 | { |
144 | { |
97 | if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) |
145 | if (expect_true (len >= 2 |
98 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
146 | && IN_RANGE_INC (char, s[0], 0xc2, 0xdf) |
99 | else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
147 | && IN_RANGE_INC (char, s[1], 0x80, 0xbf))) |
100 | { |
148 | { |
101 | *clen = 2; |
149 | *clen = 2; |
102 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
150 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
103 | } |
151 | } |
104 | else |
152 | else |
105 | { |
153 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
106 | *clen = (STRLEN)-1; |
154 | } |
107 | return (UV)-1; |
155 | |
108 | } |
156 | // likewise for encoding, also never called for ascii codepoints |
|
|
157 | // this function takes advantage of this fact, although current gccs |
|
|
158 | // seem to optimise the check for >= 0x80 away anyways |
|
|
159 | INLINE unsigned char * |
|
|
160 | encode_utf8 (unsigned char *s, UV ch) |
|
|
161 | { |
|
|
162 | if (expect_false (ch < 0x000080)) |
|
|
163 | *s++ = ch; |
|
|
164 | else if (expect_true (ch < 0x000800)) |
|
|
165 | *s++ = 0xc0 | ( ch >> 6), |
|
|
166 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
167 | else if ( ch < 0x010000) |
|
|
168 | *s++ = 0xe0 | ( ch >> 12), |
|
|
169 | *s++ = 0x80 | ((ch >> 6) & 0x3f), |
|
|
170 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
171 | else if ( ch < 0x110000) |
|
|
172 | *s++ = 0xf0 | ( ch >> 18), |
|
|
173 | *s++ = 0x80 | ((ch >> 12) & 0x3f), |
|
|
174 | *s++ = 0x80 | ((ch >> 6) & 0x3f), |
|
|
175 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
176 | |
|
|
177 | return s; |
109 | } |
178 | } |
110 | |
179 | |
111 | ///////////////////////////////////////////////////////////////////////////// |
180 | ///////////////////////////////////////////////////////////////////////////// |
112 | // encoder |
181 | // encoder |
113 | |
182 | |
… | |
… | |
117 | char *cur; // SvPVX (sv) + current output position |
186 | char *cur; // SvPVX (sv) + current output position |
118 | char *end; // SvEND (sv) |
187 | char *end; // SvEND (sv) |
119 | SV *sv; // result scalar |
188 | SV *sv; // result scalar |
120 | JSON json; |
189 | JSON json; |
121 | U32 indent; // indentation level |
190 | U32 indent; // indentation level |
122 | U32 maxdepth; // max. indentation/recursion level |
191 | UV limit; // escape character values >= this value when encoding |
123 | } enc_t; |
192 | } enc_t; |
124 | |
193 | |
125 | inline void |
194 | INLINE void |
126 | need (enc_t *enc, STRLEN len) |
195 | need (enc_t *enc, STRLEN len) |
127 | { |
196 | { |
128 | if (expect_false (enc->cur + len >= enc->end)) |
197 | if (expect_false (enc->cur + len >= enc->end)) |
129 | { |
198 | { |
130 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
199 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
… | |
… | |
132 | enc->cur = SvPVX (enc->sv) + cur; |
201 | enc->cur = SvPVX (enc->sv) + cur; |
133 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
202 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
134 | } |
203 | } |
135 | } |
204 | } |
136 | |
205 | |
137 | inline void |
206 | INLINE void |
138 | encode_ch (enc_t *enc, char ch) |
207 | encode_ch (enc_t *enc, char ch) |
139 | { |
208 | { |
140 | need (enc, 1); |
209 | need (enc, 1); |
141 | *enc->cur++ = ch; |
210 | *enc->cur++ = ch; |
142 | } |
211 | } |
… | |
… | |
196 | { |
265 | { |
197 | uch = ch; |
266 | uch = ch; |
198 | clen = 1; |
267 | clen = 1; |
199 | } |
268 | } |
200 | |
269 | |
201 | if (uch > 0x10FFFFUL) |
270 | if (uch < 0x80/*0x20*/ || uch >= enc->limit) |
202 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
|
|
203 | |
|
|
204 | if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF)) |
|
|
205 | { |
271 | { |
206 | if (uch > 0xFFFFUL) |
272 | if (uch >= 0x10000UL) |
207 | { |
273 | { |
|
|
274 | if (uch >= 0x110000UL) |
|
|
275 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
|
|
276 | |
208 | need (enc, len += 11); |
277 | need (enc, len += 11); |
209 | sprintf (enc->cur, "\\u%04x\\u%04x", |
278 | sprintf (enc->cur, "\\u%04x\\u%04x", |
210 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
279 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
211 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
280 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
212 | enc->cur += 12; |
281 | enc->cur += 12; |
213 | } |
282 | } |
214 | else |
283 | else |
215 | { |
284 | { |
216 | static char hexdigit [16] = "0123456789abcdef"; |
|
|
217 | need (enc, len += 5); |
285 | need (enc, len += 5); |
218 | *enc->cur++ = '\\'; |
286 | *enc->cur++ = '\\'; |
219 | *enc->cur++ = 'u'; |
287 | *enc->cur++ = 'u'; |
220 | *enc->cur++ = hexdigit [ uch >> 12 ]; |
288 | *enc->cur++ = PL_hexdigit [ uch >> 12 ]; |
221 | *enc->cur++ = hexdigit [(uch >> 8) & 15]; |
289 | *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; |
222 | *enc->cur++ = hexdigit [(uch >> 4) & 15]; |
290 | *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; |
223 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
291 | *enc->cur++ = PL_hexdigit [(uch >> 0) & 15]; |
224 | } |
292 | } |
225 | |
293 | |
226 | str += clen; |
294 | str += clen; |
227 | } |
295 | } |
228 | else if (enc->json.flags & F_LATIN1) |
296 | else if (enc->json.flags & F_LATIN1) |
… | |
… | |
240 | while (--clen); |
308 | while (--clen); |
241 | } |
309 | } |
242 | else |
310 | else |
243 | { |
311 | { |
244 | need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
312 | need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
245 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
313 | enc->cur = encode_utf8 (enc->cur, uch); |
246 | ++str; |
314 | ++str; |
247 | } |
315 | } |
248 | } |
316 | } |
249 | } |
317 | } |
250 | } |
318 | } |
251 | |
319 | |
252 | --len; |
320 | --len; |
253 | } |
321 | } |
254 | } |
322 | } |
255 | |
323 | |
256 | inline void |
324 | INLINE void |
257 | encode_indent (enc_t *enc) |
325 | encode_indent (enc_t *enc) |
258 | { |
326 | { |
259 | if (enc->json.flags & F_INDENT) |
327 | if (enc->json.flags & F_INDENT) |
260 | { |
328 | { |
261 | int spaces = enc->indent * INDENT_STEP; |
329 | int spaces = enc->indent * INDENT_STEP; |
… | |
… | |
264 | memset (enc->cur, ' ', spaces); |
332 | memset (enc->cur, ' ', spaces); |
265 | enc->cur += spaces; |
333 | enc->cur += spaces; |
266 | } |
334 | } |
267 | } |
335 | } |
268 | |
336 | |
269 | inline void |
337 | INLINE void |
270 | encode_space (enc_t *enc) |
338 | encode_space (enc_t *enc) |
271 | { |
339 | { |
272 | need (enc, 1); |
340 | need (enc, 1); |
273 | encode_ch (enc, ' '); |
341 | encode_ch (enc, ' '); |
274 | } |
342 | } |
275 | |
343 | |
276 | inline void |
344 | INLINE void |
277 | encode_nl (enc_t *enc) |
345 | encode_nl (enc_t *enc) |
278 | { |
346 | { |
279 | if (enc->json.flags & F_INDENT) |
347 | if (enc->json.flags & F_INDENT) |
280 | { |
348 | { |
281 | need (enc, 1); |
349 | need (enc, 1); |
282 | encode_ch (enc, '\n'); |
350 | encode_ch (enc, '\n'); |
283 | } |
351 | } |
284 | } |
352 | } |
285 | |
353 | |
286 | inline void |
354 | INLINE void |
287 | encode_comma (enc_t *enc) |
355 | encode_comma (enc_t *enc) |
288 | { |
356 | { |
289 | encode_ch (enc, ','); |
357 | encode_ch (enc, ','); |
290 | |
358 | |
291 | if (enc->json.flags & F_INDENT) |
359 | if (enc->json.flags & F_INDENT) |
… | |
… | |
299 | static void |
367 | static void |
300 | encode_av (enc_t *enc, AV *av) |
368 | encode_av (enc_t *enc, AV *av) |
301 | { |
369 | { |
302 | int i, len = av_len (av); |
370 | int i, len = av_len (av); |
303 | |
371 | |
304 | if (enc->indent >= enc->maxdepth) |
372 | if (enc->indent >= enc->json.max_depth) |
305 | croak ("data structure too deep (hit recursion limit)"); |
373 | croak (ERR_NESTING_EXCEEDED); |
306 | |
374 | |
307 | encode_ch (enc, '['); encode_nl (enc); |
375 | encode_ch (enc, '['); |
308 | ++enc->indent; |
376 | |
|
|
377 | if (len >= 0) |
|
|
378 | { |
|
|
379 | encode_nl (enc); ++enc->indent; |
309 | |
380 | |
310 | for (i = 0; i <= len; ++i) |
381 | for (i = 0; i <= len; ++i) |
311 | { |
382 | { |
312 | SV **svp = av_fetch (av, i, 0); |
383 | SV **svp = av_fetch (av, i, 0); |
313 | |
384 | |
314 | encode_indent (enc); |
385 | encode_indent (enc); |
315 | |
386 | |
316 | if (svp) |
387 | if (svp) |
317 | encode_sv (enc, *svp); |
388 | encode_sv (enc, *svp); |
318 | else |
389 | else |
319 | encode_str (enc, "null", 4, 0); |
390 | encode_str (enc, "null", 4, 0); |
320 | |
391 | |
321 | if (i < len) |
392 | if (i < len) |
322 | encode_comma (enc); |
393 | encode_comma (enc); |
323 | } |
394 | } |
324 | |
395 | |
|
|
396 | encode_nl (enc); --enc->indent; encode_indent (enc); |
|
|
397 | } |
|
|
398 | |
325 | encode_nl (enc); |
399 | encode_ch (enc, ']'); |
326 | |
|
|
327 | --enc->indent; |
|
|
328 | encode_indent (enc); encode_ch (enc, ']'); |
|
|
329 | } |
400 | } |
330 | |
401 | |
331 | static void |
402 | static void |
332 | encode_he (enc_t *enc, HE *he) |
403 | encode_hk (enc_t *enc, HE *he) |
333 | { |
404 | { |
334 | encode_ch (enc, '"'); |
405 | encode_ch (enc, '"'); |
335 | |
406 | |
336 | if (HeKLEN (he) == HEf_SVKEY) |
407 | if (HeKLEN (he) == HEf_SVKEY) |
337 | { |
408 | { |
… | |
… | |
350 | encode_ch (enc, '"'); |
421 | encode_ch (enc, '"'); |
351 | |
422 | |
352 | if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); |
423 | if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); |
353 | encode_ch (enc, ':'); |
424 | encode_ch (enc, ':'); |
354 | if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); |
425 | if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); |
355 | encode_sv (enc, HeVAL (he)); |
|
|
356 | } |
426 | } |
357 | |
427 | |
358 | // compare hash entries, used when all keys are bytestrings |
428 | // compare hash entries, used when all keys are bytestrings |
359 | static int |
429 | static int |
360 | he_cmp_fast (const void *a_, const void *b_) |
430 | he_cmp_fast (const void *a_, const void *b_) |
… | |
… | |
365 | HE *b = *(HE **)b_; |
435 | HE *b = *(HE **)b_; |
366 | |
436 | |
367 | STRLEN la = HeKLEN (a); |
437 | STRLEN la = HeKLEN (a); |
368 | STRLEN lb = HeKLEN (b); |
438 | STRLEN lb = HeKLEN (b); |
369 | |
439 | |
370 | if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) |
440 | if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la))) |
371 | cmp = la - lb; |
441 | cmp = lb - la; |
372 | |
442 | |
373 | return cmp; |
443 | return cmp; |
374 | } |
444 | } |
375 | |
445 | |
376 | // compare hash entries, used when some keys are sv's or utf-x |
446 | // compare hash entries, used when some keys are sv's or utf-x |
377 | static int |
447 | static int |
378 | he_cmp_slow (const void *a, const void *b) |
448 | he_cmp_slow (const void *a, const void *b) |
379 | { |
449 | { |
380 | return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); |
450 | return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a)); |
381 | } |
451 | } |
382 | |
452 | |
383 | static void |
453 | static void |
384 | encode_hv (enc_t *enc, HV *hv) |
454 | encode_hv (enc_t *enc, HV *hv) |
385 | { |
455 | { |
386 | int count, i; |
456 | HE *he; |
387 | |
457 | |
388 | if (enc->indent >= enc->maxdepth) |
458 | if (enc->indent >= enc->json.max_depth) |
389 | croak ("data structure too deep (hit recursion limit)"); |
459 | croak (ERR_NESTING_EXCEEDED); |
390 | |
460 | |
391 | encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; |
461 | encode_ch (enc, '{'); |
392 | |
462 | |
393 | if ((count = hv_iterinit (hv))) |
|
|
394 | { |
|
|
395 | // for canonical output we have to sort by keys first |
463 | // for canonical output we have to sort by keys first |
396 | // actually, this is mostly due to the stupid so-called |
464 | // actually, this is mostly due to the stupid so-called |
397 | // security workaround added somewhere in 5.8.x. |
465 | // security workaround added somewhere in 5.8.x. |
398 | // that randomises hash orderings |
466 | // that randomises hash orderings |
399 | if (enc->json.flags & F_CANONICAL) |
467 | if (enc->json.flags & F_CANONICAL) |
|
|
468 | { |
|
|
469 | int count = hv_iterinit (hv); |
|
|
470 | |
|
|
471 | if (SvMAGICAL (hv)) |
400 | { |
472 | { |
|
|
473 | // need to count by iterating. could improve by dynamically building the vector below |
|
|
474 | // but I don't care for the speed of this special case. |
|
|
475 | // note also that we will run into undefined behaviour when the two iterations |
|
|
476 | // do not result in the same count, something I might care for in some later release. |
|
|
477 | |
|
|
478 | count = 0; |
|
|
479 | while (hv_iternext (hv)) |
|
|
480 | ++count; |
|
|
481 | |
|
|
482 | hv_iterinit (hv); |
|
|
483 | } |
|
|
484 | |
|
|
485 | if (count) |
|
|
486 | { |
401 | int fast = 1; |
487 | int i, fast = 1; |
402 | HE *he; |
|
|
403 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
488 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
404 | HE **hes = _alloca (count * sizeof (HE)); |
489 | HE **hes = _alloca (count * sizeof (HE)); |
405 | #else |
490 | #else |
406 | HE *hes [count]; // if your compiler dies here, you need to enable C99 mode |
491 | HE *hes [count]; // if your compiler dies here, you need to enable C99 mode |
407 | #endif |
492 | #endif |
… | |
… | |
434 | |
519 | |
435 | FREETMPS; |
520 | FREETMPS; |
436 | LEAVE; |
521 | LEAVE; |
437 | } |
522 | } |
438 | |
523 | |
439 | for (i = 0; i < count; ++i) |
524 | encode_nl (enc); ++enc->indent; |
|
|
525 | |
|
|
526 | while (count--) |
440 | { |
527 | { |
441 | encode_indent (enc); |
528 | encode_indent (enc); |
|
|
529 | he = hes [count]; |
442 | encode_he (enc, hes [i]); |
530 | encode_hk (enc, he); |
|
|
531 | encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
443 | |
532 | |
444 | if (i < count - 1) |
533 | if (count) |
445 | encode_comma (enc); |
534 | encode_comma (enc); |
446 | } |
535 | } |
447 | |
536 | |
448 | encode_nl (enc); |
537 | encode_nl (enc); --enc->indent; encode_indent (enc); |
449 | } |
538 | } |
|
|
539 | } |
450 | else |
540 | else |
|
|
541 | { |
|
|
542 | if (hv_iterinit (hv) || SvMAGICAL (hv)) |
|
|
543 | if ((he = hv_iternext (hv))) |
451 | { |
544 | { |
452 | HE *he = hv_iternext (hv); |
545 | encode_nl (enc); ++enc->indent; |
453 | |
546 | |
454 | for (;;) |
547 | for (;;) |
455 | { |
548 | { |
456 | encode_indent (enc); |
549 | encode_indent (enc); |
457 | encode_he (enc, he); |
550 | encode_hk (enc, he); |
|
|
551 | encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
458 | |
552 | |
459 | if (!(he = hv_iternext (hv))) |
553 | if (!(he = hv_iternext (hv))) |
460 | break; |
554 | break; |
461 | |
555 | |
462 | encode_comma (enc); |
556 | encode_comma (enc); |
463 | } |
557 | } |
464 | |
558 | |
465 | encode_nl (enc); |
559 | encode_nl (enc); --enc->indent; encode_indent (enc); |
466 | } |
560 | } |
467 | } |
561 | } |
468 | |
562 | |
469 | --enc->indent; encode_indent (enc); encode_ch (enc, '}'); |
563 | encode_ch (enc, '}'); |
470 | } |
564 | } |
471 | |
565 | |
472 | // encode objects, arrays and special \0=false and \1=true values. |
566 | // encode objects, arrays and special \0=false and \1=true values. |
473 | static void |
567 | static void |
474 | encode_rv (enc_t *enc, SV *sv) |
568 | encode_rv (enc_t *enc, SV *sv) |
… | |
… | |
478 | SvGETMAGIC (sv); |
572 | SvGETMAGIC (sv); |
479 | svt = SvTYPE (sv); |
573 | svt = SvTYPE (sv); |
480 | |
574 | |
481 | if (expect_false (SvOBJECT (sv))) |
575 | if (expect_false (SvOBJECT (sv))) |
482 | { |
576 | { |
|
|
577 | HV *stash = !JSON_SLOW || json_boolean_stash |
|
|
578 | ? json_boolean_stash |
|
|
579 | : gv_stashpv ("JSON::XS::Boolean", 1); |
|
|
580 | |
483 | if (SvSTASH (sv) == json_boolean_stash) |
581 | if (SvSTASH (sv) == stash) |
484 | { |
582 | { |
485 | if (SvIV (sv)) |
583 | if (SvIV (sv)) |
486 | encode_str (enc, "true", 4, 0); |
584 | encode_str (enc, "true", 4, 0); |
487 | else |
585 | else |
488 | encode_str (enc, "false", 5, 0); |
586 | encode_str (enc, "false", 5, 0); |
… | |
… | |
500 | // we re-bless the reference to get overload and other niceties right |
598 | // we re-bless the reference to get overload and other niceties right |
501 | GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0); |
599 | GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0); |
502 | |
600 | |
503 | if (to_json) |
601 | if (to_json) |
504 | { |
602 | { |
505 | int count; |
|
|
506 | dSP; |
603 | dSP; |
507 | |
604 | |
508 | ENTER; SAVETMPS; PUSHMARK (SP); |
605 | ENTER; SAVETMPS; PUSHMARK (SP); |
509 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); |
606 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); |
510 | |
607 | |
… | |
… | |
548 | |
645 | |
549 | if (len == 1 && *pv == '1') |
646 | if (len == 1 && *pv == '1') |
550 | encode_str (enc, "true", 4, 0); |
647 | encode_str (enc, "true", 4, 0); |
551 | else if (len == 1 && *pv == '0') |
648 | else if (len == 1 && *pv == '0') |
552 | encode_str (enc, "false", 5, 0); |
649 | encode_str (enc, "false", 5, 0); |
|
|
650 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
651 | encode_str (enc, "null", 4, 0); |
553 | else |
652 | else |
554 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
653 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
555 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
654 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
556 | } |
655 | } |
|
|
656 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
657 | encode_str (enc, "null", 4, 0); |
557 | else |
658 | else |
558 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
659 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
559 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
660 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
560 | } |
661 | } |
561 | |
662 | |
… | |
… | |
579 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
680 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
580 | enc->cur += strlen (enc->cur); |
681 | enc->cur += strlen (enc->cur); |
581 | } |
682 | } |
582 | else if (SvIOKp (sv)) |
683 | else if (SvIOKp (sv)) |
583 | { |
684 | { |
584 | // we assume we can always read an IV as a UV |
685 | // we assume we can always read an IV as a UV and vice versa |
585 | if (SvUV (sv) & ~(UV)0x7fff) |
686 | // we assume two's complement |
586 | { |
687 | // we assume no aliasing issues in the union |
587 | // large integer, use the (rather slow) snprintf way. |
688 | if (SvIsUV (sv) ? SvUVX (sv) <= 59000 |
588 | need (enc, sizeof (UV) * 3); |
689 | : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000) |
589 | enc->cur += |
|
|
590 | SvIsUV(sv) |
|
|
591 | ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv)) |
|
|
592 | : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv)); |
|
|
593 | } |
|
|
594 | else |
|
|
595 | { |
690 | { |
596 | // optimise the "small number case" |
691 | // optimise the "small number case" |
597 | // code will likely be branchless and use only a single multiplication |
692 | // code will likely be branchless and use only a single multiplication |
|
|
693 | // works for numbers up to 59074 |
598 | I32 i = SvIV (sv); |
694 | I32 i = SvIVX (sv); |
599 | U32 u; |
695 | U32 u; |
600 | char digit, nz = 0; |
696 | char digit, nz = 0; |
601 | |
697 | |
602 | need (enc, 6); |
698 | need (enc, 6); |
603 | |
699 | |
… | |
… | |
609 | |
705 | |
610 | // now output digit by digit, each time masking out the integer part |
706 | // now output digit by digit, each time masking out the integer part |
611 | // and multiplying by 5 while moving the decimal point one to the right, |
707 | // and multiplying by 5 while moving the decimal point one to the right, |
612 | // resulting in a net multiplication by 10. |
708 | // resulting in a net multiplication by 10. |
613 | // we always write the digit to memory but conditionally increment |
709 | // we always write the digit to memory but conditionally increment |
614 | // the pointer, to ease the usage of conditional move instructions. |
710 | // the pointer, to enable the use of conditional move instructions. |
615 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; |
711 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5; |
616 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; |
712 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5; |
617 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; |
713 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5; |
618 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; |
714 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5; |
619 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
715 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
620 | } |
716 | } |
|
|
717 | else |
|
|
718 | { |
|
|
719 | // large integer, use the (rather slow) snprintf way. |
|
|
720 | need (enc, IVUV_MAXCHARS); |
|
|
721 | enc->cur += |
|
|
722 | SvIsUV(sv) |
|
|
723 | ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) |
|
|
724 | : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); |
|
|
725 | } |
621 | } |
726 | } |
622 | else if (SvROK (sv)) |
727 | else if (SvROK (sv)) |
623 | encode_rv (enc, SvRV (sv)); |
728 | encode_rv (enc, SvRV (sv)); |
624 | else if (!SvOK (sv)) |
729 | else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) |
625 | encode_str (enc, "null", 4, 0); |
730 | encode_str (enc, "null", 4, 0); |
626 | else |
731 | else |
627 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
732 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
628 | SvPV_nolen (sv), SvFLAGS (sv)); |
733 | SvPV_nolen (sv), SvFLAGS (sv)); |
629 | } |
734 | } |
… | |
… | |
639 | enc.json = *json; |
744 | enc.json = *json; |
640 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
745 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
641 | enc.cur = SvPVX (enc.sv); |
746 | enc.cur = SvPVX (enc.sv); |
642 | enc.end = SvEND (enc.sv); |
747 | enc.end = SvEND (enc.sv); |
643 | enc.indent = 0; |
748 | enc.indent = 0; |
644 | enc.maxdepth = DEC_DEPTH (enc.json.flags); |
749 | enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
|
|
750 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
|
|
751 | : 0x110000UL; |
645 | |
752 | |
646 | SvPOK_only (enc.sv); |
753 | SvPOK_only (enc.sv); |
647 | encode_sv (&enc, scalar); |
754 | encode_sv (&enc, scalar); |
648 | |
755 | |
649 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
756 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
… | |
… | |
670 | JSON json; |
777 | JSON json; |
671 | U32 depth; // recursion depth |
778 | U32 depth; // recursion depth |
672 | U32 maxdepth; // recursion depth limit |
779 | U32 maxdepth; // recursion depth limit |
673 | } dec_t; |
780 | } dec_t; |
674 | |
781 | |
675 | inline void |
782 | INLINE void |
|
|
783 | decode_comment (dec_t *dec) |
|
|
784 | { |
|
|
785 | // only '#'-style comments allowed a.t.m. |
|
|
786 | |
|
|
787 | while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) |
|
|
788 | ++dec->cur; |
|
|
789 | } |
|
|
790 | |
|
|
791 | INLINE void |
676 | decode_ws (dec_t *dec) |
792 | decode_ws (dec_t *dec) |
677 | { |
793 | { |
678 | for (;;) |
794 | for (;;) |
679 | { |
795 | { |
680 | char ch = *dec->cur; |
796 | char ch = *dec->cur; |
681 | |
797 | |
682 | if (ch > 0x20 |
798 | if (ch > 0x20) |
|
|
799 | { |
|
|
800 | if (expect_false (ch == '#')) |
|
|
801 | { |
|
|
802 | if (dec->json.flags & F_RELAXED) |
|
|
803 | decode_comment (dec); |
|
|
804 | else |
|
|
805 | break; |
|
|
806 | } |
|
|
807 | else |
|
|
808 | break; |
|
|
809 | } |
683 | || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) |
810 | else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09) |
684 | break; |
811 | break; // parse error, but let higher level handle it, gives better error messages |
685 | |
812 | |
686 | ++dec->cur; |
813 | ++dec->cur; |
687 | } |
814 | } |
688 | } |
815 | } |
689 | |
816 | |
… | |
… | |
693 | if (*dec->cur != ch) \ |
820 | if (*dec->cur != ch) \ |
694 | ERR (# ch " expected"); \ |
821 | ERR (# ch " expected"); \ |
695 | ++dec->cur; \ |
822 | ++dec->cur; \ |
696 | SE |
823 | SE |
697 | |
824 | |
698 | #define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") |
825 | #define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED) |
699 | #define DEC_DEC_DEPTH --dec->depth |
826 | #define DEC_DEC_DEPTH --dec->depth |
700 | |
827 | |
701 | static SV *decode_sv (dec_t *dec); |
828 | static SV *decode_sv (dec_t *dec); |
702 | |
829 | |
703 | static signed char decode_hexdigit[256]; |
830 | static signed char decode_hexdigit[256]; |
… | |
… | |
795 | |
922 | |
796 | if (hi >= 0x80) |
923 | if (hi >= 0x80) |
797 | { |
924 | { |
798 | utf8 = 1; |
925 | utf8 = 1; |
799 | |
926 | |
800 | cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); |
927 | cur = encode_utf8 (cur, hi); |
801 | } |
928 | } |
802 | else |
929 | else |
803 | *cur++ = hi; |
930 | *cur++ = hi; |
804 | } |
931 | } |
805 | break; |
932 | break; |
… | |
… | |
807 | default: |
934 | default: |
808 | --dec_cur; |
935 | --dec_cur; |
809 | ERR ("illegal backslash escape sequence in string"); |
936 | ERR ("illegal backslash escape sequence in string"); |
810 | } |
937 | } |
811 | } |
938 | } |
812 | else if (expect_true (ch >= 0x20 && ch <= 0x7f)) |
939 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
813 | *cur++ = ch; |
940 | *cur++ = ch; |
814 | else if (ch >= 0x80) |
941 | else if (ch >= 0x80) |
815 | { |
942 | { |
816 | STRLEN clen; |
943 | STRLEN clen; |
817 | UV uch; |
944 | UV uch; |
… | |
… | |
940 | |
1067 | |
941 | if (!is_nv) |
1068 | if (!is_nv) |
942 | { |
1069 | { |
943 | int len = dec->cur - start; |
1070 | int len = dec->cur - start; |
944 | |
1071 | |
945 | // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so |
1072 | // special case the rather common 1..5-digit-int case |
946 | if (*start == '-') |
1073 | if (*start == '-') |
947 | switch (len) |
1074 | switch (len) |
948 | { |
1075 | { |
949 | case 2: return newSViv (-( start [1] - '0' * 1)); |
1076 | case 2: return newSViv (-( start [1] - '0' * 1)); |
950 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
1077 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
951 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
1078 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
952 | case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
1079 | case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
|
|
1080 | case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
953 | } |
1081 | } |
954 | else |
1082 | else |
955 | switch (len) |
1083 | switch (len) |
956 | { |
1084 | { |
957 | case 1: return newSViv ( start [0] - '0' * 1); |
1085 | case 1: return newSViv ( start [0] - '0' * 1); |
958 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1086 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
959 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
1087 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
960 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
1088 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
|
|
1089 | case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
961 | } |
1090 | } |
962 | |
1091 | |
963 | { |
1092 | { |
964 | UV uv; |
1093 | UV uv; |
965 | int numtype = grok_number (start, len, &uv); |
1094 | int numtype = grok_number (start, len, &uv); |
… | |
… | |
1026 | |
1155 | |
1027 | if (*dec->cur != ',') |
1156 | if (*dec->cur != ',') |
1028 | ERR (", or ] expected while parsing array"); |
1157 | ERR (", or ] expected while parsing array"); |
1029 | |
1158 | |
1030 | ++dec->cur; |
1159 | ++dec->cur; |
|
|
1160 | |
|
|
1161 | decode_ws (dec); |
|
|
1162 | |
|
|
1163 | if (*dec->cur == ']' && dec->json.flags & F_RELAXED) |
|
|
1164 | { |
|
|
1165 | ++dec->cur; |
|
|
1166 | break; |
|
|
1167 | } |
1031 | } |
1168 | } |
1032 | |
1169 | |
1033 | DEC_DEC_DEPTH; |
1170 | DEC_DEC_DEPTH; |
1034 | return newRV_noinc ((SV *)av); |
1171 | return newRV_noinc ((SV *)av); |
1035 | |
1172 | |
… | |
… | |
1051 | if (*dec->cur == '}') |
1188 | if (*dec->cur == '}') |
1052 | ++dec->cur; |
1189 | ++dec->cur; |
1053 | else |
1190 | else |
1054 | for (;;) |
1191 | for (;;) |
1055 | { |
1192 | { |
1056 | decode_ws (dec); EXPECT_CH ('"'); |
1193 | EXPECT_CH ('"'); |
1057 | |
1194 | |
1058 | // heuristic: assume that |
1195 | // heuristic: assume that |
1059 | // a) decode_str + hv_store_ent are abysmally slow. |
1196 | // a) decode_str + hv_store_ent are abysmally slow. |
1060 | // b) most hash keys are short, simple ascii text. |
1197 | // b) most hash keys are short, simple ascii text. |
1061 | // => try to "fast-match" such strings to avoid |
1198 | // => try to "fast-match" such strings to avoid |
… | |
… | |
1065 | char *p = dec->cur; |
1202 | char *p = dec->cur; |
1066 | char *e = p + 24; // only try up to 24 bytes |
1203 | char *e = p + 24; // only try up to 24 bytes |
1067 | |
1204 | |
1068 | for (;;) |
1205 | for (;;) |
1069 | { |
1206 | { |
1070 | // the >= 0x80 is true on most architectures |
1207 | // the >= 0x80 is false on most architectures |
1071 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1208 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1072 | { |
1209 | { |
1073 | // slow path, back up and use decode_str |
1210 | // slow path, back up and use decode_str |
1074 | SV *key = decode_str (dec); |
1211 | SV *key = decode_str (dec); |
1075 | if (!key) |
1212 | if (!key) |
1076 | goto fail; |
1213 | goto fail; |
1077 | |
1214 | |
1078 | decode_ws (dec); EXPECT_CH (':'); |
1215 | decode_ws (dec); EXPECT_CH (':'); |
1079 | |
1216 | |
|
|
1217 | decode_ws (dec); |
1080 | value = decode_sv (dec); |
1218 | value = decode_sv (dec); |
1081 | if (!value) |
1219 | if (!value) |
1082 | { |
1220 | { |
1083 | SvREFCNT_dec (key); |
1221 | SvREFCNT_dec (key); |
1084 | goto fail; |
1222 | goto fail; |
… | |
… | |
1096 | int len = p - key; |
1234 | int len = p - key; |
1097 | dec->cur = p + 1; |
1235 | dec->cur = p + 1; |
1098 | |
1236 | |
1099 | decode_ws (dec); EXPECT_CH (':'); |
1237 | decode_ws (dec); EXPECT_CH (':'); |
1100 | |
1238 | |
|
|
1239 | decode_ws (dec); |
1101 | value = decode_sv (dec); |
1240 | value = decode_sv (dec); |
1102 | if (!value) |
1241 | if (!value) |
1103 | goto fail; |
1242 | goto fail; |
1104 | |
1243 | |
1105 | hv_store (hv, key, len, value, 0); |
1244 | hv_store (hv, key, len, value, 0); |
… | |
… | |
1121 | |
1260 | |
1122 | if (*dec->cur != ',') |
1261 | if (*dec->cur != ',') |
1123 | ERR (", or } expected while parsing object/hash"); |
1262 | ERR (", or } expected while parsing object/hash"); |
1124 | |
1263 | |
1125 | ++dec->cur; |
1264 | ++dec->cur; |
|
|
1265 | |
|
|
1266 | decode_ws (dec); |
|
|
1267 | |
|
|
1268 | if (*dec->cur == '}' && dec->json.flags & F_RELAXED) |
|
|
1269 | { |
|
|
1270 | ++dec->cur; |
|
|
1271 | break; |
|
|
1272 | } |
1126 | } |
1273 | } |
1127 | |
1274 | |
1128 | DEC_DEC_DEPTH; |
1275 | DEC_DEC_DEPTH; |
1129 | sv = newRV_noinc ((SV *)hv); |
1276 | sv = newRV_noinc ((SV *)hv); |
1130 | |
1277 | |
… | |
… | |
1195 | } |
1342 | } |
1196 | |
1343 | |
1197 | static SV * |
1344 | static SV * |
1198 | decode_sv (dec_t *dec) |
1345 | decode_sv (dec_t *dec) |
1199 | { |
1346 | { |
1200 | decode_ws (dec); |
|
|
1201 | |
|
|
1202 | // the beauty of JSON: you need exactly one character lookahead |
1347 | // the beauty of JSON: you need exactly one character lookahead |
1203 | // to parse anything. |
1348 | // to parse everything. |
1204 | switch (*dec->cur) |
1349 | switch (*dec->cur) |
1205 | { |
1350 | { |
1206 | case '"': ++dec->cur; return decode_str (dec); |
1351 | case '"': ++dec->cur; return decode_str (dec); |
1207 | case '[': ++dec->cur; return decode_av (dec); |
1352 | case '[': ++dec->cur; return decode_av (dec); |
1208 | case '{': ++dec->cur; return decode_hv (dec); |
1353 | case '{': ++dec->cur; return decode_hv (dec); |
1209 | |
1354 | |
1210 | case '-': |
1355 | case '-': |
1211 | case '0': case '1': case '2': case '3': case '4': |
1356 | case '0': case '1': case '2': case '3': case '4': |
1212 | case '5': case '6': case '7': case '8': case '9': |
1357 | case '5': case '6': case '7': case '8': case '9': |
1213 | return decode_num (dec); |
1358 | return decode_num (dec); |
1214 | |
1359 | |
1215 | case 't': |
1360 | case 't': |
1216 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1361 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1217 | { |
1362 | { |
1218 | dec->cur += 4; |
1363 | dec->cur += 4; |
|
|
1364 | #if JSON_SLOW |
|
|
1365 | json_true = get_bool ("JSON::XS::true"); |
|
|
1366 | #endif |
1219 | return SvREFCNT_inc (json_true); |
1367 | return newSVsv (json_true); |
1220 | } |
1368 | } |
1221 | else |
1369 | else |
1222 | ERR ("'true' expected"); |
1370 | ERR ("'true' expected"); |
1223 | |
1371 | |
1224 | break; |
1372 | break; |
1225 | |
1373 | |
1226 | case 'f': |
1374 | case 'f': |
1227 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1375 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1228 | { |
1376 | { |
1229 | dec->cur += 5; |
1377 | dec->cur += 5; |
|
|
1378 | #if JSON_SLOW |
|
|
1379 | json_false = get_bool ("JSON::XS::false"); |
|
|
1380 | #endif |
1230 | return SvREFCNT_inc (json_false); |
1381 | return newSVsv (json_false); |
1231 | } |
1382 | } |
1232 | else |
1383 | else |
1233 | ERR ("'false' expected"); |
1384 | ERR ("'false' expected"); |
1234 | |
1385 | |
1235 | break; |
1386 | break; |
… | |
… | |
1253 | fail: |
1404 | fail: |
1254 | return 0; |
1405 | return 0; |
1255 | } |
1406 | } |
1256 | |
1407 | |
1257 | static SV * |
1408 | static SV * |
1258 | decode_json (SV *string, JSON *json, UV *offset_return) |
1409 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1259 | { |
1410 | { |
1260 | dec_t dec; |
1411 | dec_t dec; |
1261 | UV offset; |
1412 | STRLEN offset; |
1262 | SV *sv; |
1413 | SV *sv; |
1263 | |
1414 | |
1264 | SvGETMAGIC (string); |
1415 | SvGETMAGIC (string); |
1265 | SvUPGRADE (string, SVt_PV); |
1416 | SvUPGRADE (string, SVt_PV); |
1266 | |
1417 | |
1267 | if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) |
1418 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
|
|
1419 | * assertion with -DDEBUGGING, although SvCUR is documented to |
|
|
1420 | * return the xpv_cur field which certainly exists after upgrading. |
|
|
1421 | * according to nicholas clark, calling SvPOK fixes this. |
|
|
1422 | * But it doesn't fix it, so try another workaround, call SvPV_nolen |
|
|
1423 | * and hope for the best. |
|
|
1424 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
|
|
1425 | * assertion business is seriously broken, try yet another workaround |
|
|
1426 | * for the broken -DDEBUGGING. |
|
|
1427 | */ |
|
|
1428 | #ifdef DEBUGGING |
|
|
1429 | offset = SvOK (string) ? sv_len (string) : 0; |
|
|
1430 | #else |
|
|
1431 | offset = SvCUR (string); |
|
|
1432 | #endif |
|
|
1433 | |
|
|
1434 | if (offset > json->max_size && json->max_size) |
1268 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1435 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1269 | (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); |
1436 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
1270 | |
1437 | |
1271 | if (json->flags & F_UTF8) |
1438 | if (json->flags & F_UTF8) |
1272 | sv_utf8_downgrade (string, 0); |
1439 | sv_utf8_downgrade (string, 0); |
1273 | else |
1440 | else |
1274 | sv_utf8_upgrade (string); |
1441 | sv_utf8_upgrade (string); |
1275 | |
1442 | |
1276 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1443 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1277 | |
1444 | |
1278 | dec.json = *json; |
1445 | dec.json = *json; |
1279 | dec.cur = SvPVX (string); |
1446 | dec.cur = SvPVX (string); |
1280 | dec.end = SvEND (string); |
1447 | dec.end = SvEND (string); |
1281 | dec.err = 0; |
1448 | dec.err = 0; |
1282 | dec.depth = 0; |
1449 | dec.depth = 0; |
1283 | dec.maxdepth = DEC_DEPTH (dec.json.flags); |
|
|
1284 | |
1450 | |
1285 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1451 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1286 | dec.json.flags |= F_HOOK; |
1452 | dec.json.flags |= F_HOOK; |
1287 | |
1453 | |
1288 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1454 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
|
|
1455 | |
|
|
1456 | decode_ws (&dec); |
1289 | sv = decode_sv (&dec); |
1457 | sv = decode_sv (&dec); |
1290 | |
1458 | |
1291 | if (!(offset_return || !sv)) |
1459 | if (!(offset_return || !sv)) |
1292 | { |
1460 | { |
1293 | // check for trailing garbage |
1461 | // check for trailing garbage |
… | |
… | |
1337 | |
1505 | |
1338 | return sv; |
1506 | return sv; |
1339 | } |
1507 | } |
1340 | |
1508 | |
1341 | ///////////////////////////////////////////////////////////////////////////// |
1509 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1510 | // incremental parser |
|
|
1511 | |
|
|
1512 | static void |
|
|
1513 | incr_parse (JSON *self) |
|
|
1514 | { |
|
|
1515 | const char *p = SvPVX (self->incr_text) + self->incr_pos; |
|
|
1516 | |
|
|
1517 | for (;;) |
|
|
1518 | { |
|
|
1519 | //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 |
|
|
1520 | switch (self->incr_mode) |
|
|
1521 | { |
|
|
1522 | // only used for intiial whitespace skipping |
|
|
1523 | case INCR_M_WS: |
|
|
1524 | for (;;) |
|
|
1525 | { |
|
|
1526 | if (*p > 0x20) |
|
|
1527 | { |
|
|
1528 | self->incr_mode = INCR_M_JSON; |
|
|
1529 | goto incr_m_json; |
|
|
1530 | } |
|
|
1531 | else if (!*p) |
|
|
1532 | goto interrupt; |
|
|
1533 | |
|
|
1534 | ++p; |
|
|
1535 | } |
|
|
1536 | |
|
|
1537 | // skip a single char inside a string (for \\-processing) |
|
|
1538 | case INCR_M_BS: |
|
|
1539 | if (!*p) |
|
|
1540 | goto interrupt; |
|
|
1541 | |
|
|
1542 | ++p; |
|
|
1543 | self->incr_mode = INCR_M_STR; |
|
|
1544 | goto incr_m_str; |
|
|
1545 | |
|
|
1546 | // inside a string |
|
|
1547 | case INCR_M_STR: |
|
|
1548 | incr_m_str: |
|
|
1549 | for (;;) |
|
|
1550 | { |
|
|
1551 | if (*p == '"') |
|
|
1552 | { |
|
|
1553 | ++p; |
|
|
1554 | self->incr_mode = INCR_M_JSON; |
|
|
1555 | |
|
|
1556 | if (!self->incr_nest) |
|
|
1557 | goto interrupt; |
|
|
1558 | |
|
|
1559 | goto incr_m_json; |
|
|
1560 | } |
|
|
1561 | else if (*p == '\\') |
|
|
1562 | { |
|
|
1563 | ++p; // "virtually" consumes character after \ |
|
|
1564 | |
|
|
1565 | if (!*p) // if at end of string we have to switch modes |
|
|
1566 | { |
|
|
1567 | self->incr_mode = INCR_M_BS; |
|
|
1568 | goto interrupt; |
|
|
1569 | } |
|
|
1570 | } |
|
|
1571 | else if (!*p) |
|
|
1572 | goto interrupt; |
|
|
1573 | |
|
|
1574 | ++p; |
|
|
1575 | } |
|
|
1576 | |
|
|
1577 | // after initial ws, outside string |
|
|
1578 | case INCR_M_JSON: |
|
|
1579 | incr_m_json: |
|
|
1580 | for (;;) |
|
|
1581 | { |
|
|
1582 | switch (*p++) |
|
|
1583 | { |
|
|
1584 | case 0: |
|
|
1585 | --p; |
|
|
1586 | goto interrupt; |
|
|
1587 | |
|
|
1588 | case 0x09: |
|
|
1589 | case 0x0a: |
|
|
1590 | case 0x0d: |
|
|
1591 | case 0x20: |
|
|
1592 | if (!self->incr_nest) |
|
|
1593 | { |
|
|
1594 | --p; // do not eat the whitespace, let the next round do it |
|
|
1595 | goto interrupt; |
|
|
1596 | } |
|
|
1597 | break; |
|
|
1598 | |
|
|
1599 | case '"': |
|
|
1600 | self->incr_mode = INCR_M_STR; |
|
|
1601 | goto incr_m_str; |
|
|
1602 | |
|
|
1603 | case '[': |
|
|
1604 | case '{': |
|
|
1605 | if (++self->incr_nest > self->max_depth) |
|
|
1606 | croak (ERR_NESTING_EXCEEDED); |
|
|
1607 | break; |
|
|
1608 | |
|
|
1609 | case ']': |
|
|
1610 | case '}': |
|
|
1611 | if (--self->incr_nest <= 0) |
|
|
1612 | goto interrupt; |
|
|
1613 | } |
|
|
1614 | } |
|
|
1615 | } |
|
|
1616 | |
|
|
1617 | modechange: |
|
|
1618 | ; |
|
|
1619 | } |
|
|
1620 | |
|
|
1621 | interrupt: |
|
|
1622 | self->incr_pos = p - SvPVX (self->incr_text); |
|
|
1623 | //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D |
|
|
1624 | } |
|
|
1625 | |
|
|
1626 | ///////////////////////////////////////////////////////////////////////////// |
1342 | // XS interface functions |
1627 | // XS interface functions |
1343 | |
1628 | |
1344 | MODULE = JSON::XS PACKAGE = JSON::XS |
1629 | MODULE = JSON::XS PACKAGE = JSON::XS |
1345 | |
1630 | |
1346 | BOOT: |
1631 | BOOT: |
… | |
… | |
1355 | : -1; |
1640 | : -1; |
1356 | |
1641 | |
1357 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1642 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1358 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
1643 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
1359 | |
1644 | |
1360 | json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); |
1645 | json_true = get_bool ("JSON::XS::true"); |
1361 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1646 | json_false = get_bool ("JSON::XS::false"); |
1362 | } |
1647 | } |
1363 | |
1648 | |
1364 | PROTOTYPES: DISABLE |
1649 | PROTOTYPES: DISABLE |
|
|
1650 | |
|
|
1651 | void CLONE (...) |
|
|
1652 | CODE: |
|
|
1653 | json_stash = 0; |
|
|
1654 | json_boolean_stash = 0; |
1365 | |
1655 | |
1366 | void new (char *klass) |
1656 | void new (char *klass) |
1367 | PPCODE: |
1657 | PPCODE: |
1368 | { |
1658 | { |
1369 | SV *pv = NEWSV (0, sizeof (JSON)); |
1659 | SV *pv = NEWSV (0, sizeof (JSON)); |
1370 | SvPOK_only (pv); |
1660 | SvPOK_only (pv); |
1371 | Zero (SvPVX (pv), 1, JSON); |
1661 | json_init ((JSON *)SvPVX (pv)); |
1372 | ((JSON *)SvPVX (pv))->flags = F_DEFAULT; |
1662 | XPUSHs (sv_2mortal (sv_bless ( |
1373 | XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash))); |
1663 | newRV_noinc (pv), |
|
|
1664 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
|
|
1665 | ))); |
1374 | } |
1666 | } |
1375 | |
1667 | |
1376 | void ascii (JSON *self, int enable = 1) |
1668 | void ascii (JSON *self, int enable = 1) |
1377 | ALIAS: |
1669 | ALIAS: |
1378 | ascii = F_ASCII |
1670 | ascii = F_ASCII |
… | |
… | |
1385 | pretty = F_PRETTY |
1677 | pretty = F_PRETTY |
1386 | allow_nonref = F_ALLOW_NONREF |
1678 | allow_nonref = F_ALLOW_NONREF |
1387 | shrink = F_SHRINK |
1679 | shrink = F_SHRINK |
1388 | allow_blessed = F_ALLOW_BLESSED |
1680 | allow_blessed = F_ALLOW_BLESSED |
1389 | convert_blessed = F_CONV_BLESSED |
1681 | convert_blessed = F_CONV_BLESSED |
|
|
1682 | relaxed = F_RELAXED |
|
|
1683 | allow_unknown = F_ALLOW_UNKNOWN |
1390 | PPCODE: |
1684 | PPCODE: |
1391 | { |
1685 | { |
1392 | if (enable) |
1686 | if (enable) |
1393 | self->flags |= ix; |
1687 | self->flags |= ix; |
1394 | else |
1688 | else |
1395 | self->flags &= ~ix; |
1689 | self->flags &= ~ix; |
1396 | |
1690 | |
1397 | XPUSHs (ST (0)); |
1691 | XPUSHs (ST (0)); |
1398 | } |
1692 | } |
1399 | |
1693 | |
1400 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
1694 | void get_ascii (JSON *self) |
|
|
1695 | ALIAS: |
|
|
1696 | get_ascii = F_ASCII |
|
|
1697 | get_latin1 = F_LATIN1 |
|
|
1698 | get_utf8 = F_UTF8 |
|
|
1699 | get_indent = F_INDENT |
|
|
1700 | get_canonical = F_CANONICAL |
|
|
1701 | get_space_before = F_SPACE_BEFORE |
|
|
1702 | get_space_after = F_SPACE_AFTER |
|
|
1703 | get_allow_nonref = F_ALLOW_NONREF |
|
|
1704 | get_shrink = F_SHRINK |
|
|
1705 | get_allow_blessed = F_ALLOW_BLESSED |
|
|
1706 | get_convert_blessed = F_CONV_BLESSED |
|
|
1707 | get_relaxed = F_RELAXED |
|
|
1708 | get_allow_unknown = F_ALLOW_UNKNOWN |
1401 | PPCODE: |
1709 | PPCODE: |
1402 | { |
1710 | XPUSHs (boolSV (self->flags & ix)); |
1403 | UV log2 = 0; |
|
|
1404 | |
1711 | |
1405 | if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
1712 | void max_depth (JSON *self, U32 max_depth = 0x80000000UL) |
1406 | |
1713 | PPCODE: |
1407 | while ((1UL << log2) < max_depth) |
1714 | self->max_depth = max_depth; |
1408 | ++log2; |
|
|
1409 | |
|
|
1410 | self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
|
|
1411 | |
|
|
1412 | XPUSHs (ST (0)); |
1715 | XPUSHs (ST (0)); |
1413 | } |
|
|
1414 | |
1716 | |
|
|
1717 | U32 get_max_depth (JSON *self) |
|
|
1718 | CODE: |
|
|
1719 | RETVAL = self->max_depth; |
|
|
1720 | OUTPUT: |
|
|
1721 | RETVAL |
|
|
1722 | |
1415 | void max_size (JSON *self, UV max_size = 0) |
1723 | void max_size (JSON *self, U32 max_size = 0) |
1416 | PPCODE: |
1724 | PPCODE: |
1417 | { |
1725 | self->max_size = max_size; |
1418 | UV log2 = 0; |
|
|
1419 | |
|
|
1420 | if (max_size > 0x80000000UL) max_size = 0x80000000UL; |
|
|
1421 | if (max_size == 1) max_size = 2; |
|
|
1422 | |
|
|
1423 | while ((1UL << log2) < max_size) |
|
|
1424 | ++log2; |
|
|
1425 | |
|
|
1426 | self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); |
|
|
1427 | |
|
|
1428 | XPUSHs (ST (0)); |
1726 | XPUSHs (ST (0)); |
1429 | } |
1727 | |
|
|
1728 | int get_max_size (JSON *self) |
|
|
1729 | CODE: |
|
|
1730 | RETVAL = self->max_size; |
|
|
1731 | OUTPUT: |
|
|
1732 | RETVAL |
1430 | |
1733 | |
1431 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1734 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1432 | PPCODE: |
1735 | PPCODE: |
1433 | { |
1736 | { |
1434 | SvREFCNT_dec (self->cb_object); |
1737 | SvREFCNT_dec (self->cb_object); |
… | |
… | |
1468 | XPUSHs (decode_json (jsonstr, self, 0)); |
1771 | XPUSHs (decode_json (jsonstr, self, 0)); |
1469 | |
1772 | |
1470 | void decode_prefix (JSON *self, SV *jsonstr) |
1773 | void decode_prefix (JSON *self, SV *jsonstr) |
1471 | PPCODE: |
1774 | PPCODE: |
1472 | { |
1775 | { |
1473 | UV offset; |
1776 | STRLEN offset; |
1474 | EXTEND (SP, 2); |
1777 | EXTEND (SP, 2); |
1475 | PUSHs (decode_json (jsonstr, self, &offset)); |
1778 | PUSHs (decode_json (jsonstr, self, &offset)); |
1476 | PUSHs (sv_2mortal (newSVuv (offset))); |
1779 | PUSHs (sv_2mortal (newSVuv (offset))); |
|
|
1780 | } |
|
|
1781 | |
|
|
1782 | void incr_parse (JSON *self, SV *jsonstr = 0) |
|
|
1783 | PPCODE: |
|
|
1784 | { |
|
|
1785 | if (!self->incr_text) |
|
|
1786 | self->incr_text = newSVpvn ("", 0); |
|
|
1787 | |
|
|
1788 | // append data, if any |
|
|
1789 | if (jsonstr) |
|
|
1790 | { |
|
|
1791 | if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) |
|
|
1792 | { |
|
|
1793 | /* utf-8-ness differs, need to upgrade */ |
|
|
1794 | sv_utf8_upgrade (self->incr_text); |
|
|
1795 | |
|
|
1796 | if (self->incr_pos) |
|
|
1797 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
|
|
1798 | - (U8 *)SvPVX (self->incr_text); |
|
|
1799 | } |
|
|
1800 | |
|
|
1801 | { |
|
|
1802 | STRLEN len; |
|
|
1803 | const char *str = SvPV (jsonstr, len); |
|
|
1804 | SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); |
|
|
1805 | Move (str, SvEND (self->incr_text), len, char); |
|
|
1806 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
|
|
1807 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
|
|
1808 | } |
|
|
1809 | } |
|
|
1810 | |
|
|
1811 | if (GIMME_V != G_VOID) |
|
|
1812 | do |
|
|
1813 | { |
|
|
1814 | STRLEN offset; |
|
|
1815 | |
|
|
1816 | if (!INCR_DONE (self)) |
|
|
1817 | { |
|
|
1818 | incr_parse (self); |
|
|
1819 | |
|
|
1820 | if (self->incr_pos > self->max_size && self->max_size) |
|
|
1821 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
|
|
1822 | (unsigned long)self->incr_pos, (unsigned long)self->max_size); |
|
|
1823 | |
|
|
1824 | if (!INCR_DONE (self)) |
|
|
1825 | break; |
|
|
1826 | } |
|
|
1827 | |
|
|
1828 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
|
|
1829 | |
|
|
1830 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset); |
|
|
1831 | self->incr_pos -= offset; |
|
|
1832 | self->incr_nest = 0; |
|
|
1833 | self->incr_mode = 0; |
|
|
1834 | } |
|
|
1835 | while (GIMME_V == G_ARRAY); |
|
|
1836 | } |
|
|
1837 | |
|
|
1838 | SV *incr_text (JSON *self) |
|
|
1839 | ATTRS: lvalue |
|
|
1840 | CODE: |
|
|
1841 | { |
|
|
1842 | if (self->incr_pos) |
|
|
1843 | croak ("incr_text can not be called when the incremental parser already started parsing"); |
|
|
1844 | |
|
|
1845 | RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef; |
|
|
1846 | } |
|
|
1847 | OUTPUT: |
|
|
1848 | RETVAL |
|
|
1849 | |
|
|
1850 | void incr_skip (JSON *self) |
|
|
1851 | CODE: |
|
|
1852 | { |
|
|
1853 | if (self->incr_pos) |
|
|
1854 | { |
|
|
1855 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos); |
|
|
1856 | self->incr_pos = 0; |
|
|
1857 | self->incr_nest = 0; |
|
|
1858 | self->incr_mode = 0; |
|
|
1859 | } |
|
|
1860 | } |
|
|
1861 | |
|
|
1862 | void incr_reset (JSON *self) |
|
|
1863 | CODE: |
|
|
1864 | { |
|
|
1865 | SvREFCNT_dec (self->incr_text); |
|
|
1866 | self->incr_text = 0; |
|
|
1867 | self->incr_pos = 0; |
|
|
1868 | self->incr_nest = 0; |
|
|
1869 | self->incr_mode = 0; |
1477 | } |
1870 | } |
1478 | |
1871 | |
1479 | void DESTROY (JSON *self) |
1872 | void DESTROY (JSON *self) |
1480 | CODE: |
1873 | CODE: |
1481 | SvREFCNT_dec (self->cb_sk_object); |
1874 | SvREFCNT_dec (self->cb_sk_object); |
1482 | SvREFCNT_dec (self->cb_object); |
1875 | SvREFCNT_dec (self->cb_object); |
|
|
1876 | SvREFCNT_dec (self->incr_text); |
1483 | |
1877 | |
1484 | PROTOTYPES: ENABLE |
1878 | PROTOTYPES: ENABLE |
1485 | |
1879 | |
1486 | void to_json (SV *scalar) |
1880 | void encode_json (SV *scalar) |
|
|
1881 | ALIAS: |
|
|
1882 | to_json_ = 0 |
|
|
1883 | encode_json = F_UTF8 |
1487 | PPCODE: |
1884 | PPCODE: |
1488 | { |
1885 | { |
1489 | JSON json = { F_DEFAULT | F_UTF8 }; |
1886 | JSON json; |
|
|
1887 | json_init (&json); |
|
|
1888 | json.flags |= ix; |
1490 | XPUSHs (encode_json (scalar, &json)); |
1889 | XPUSHs (encode_json (scalar, &json)); |
1491 | } |
1890 | } |
1492 | |
1891 | |
1493 | void from_json (SV *jsonstr) |
1892 | void decode_json (SV *jsonstr) |
|
|
1893 | ALIAS: |
|
|
1894 | from_json_ = 0 |
|
|
1895 | decode_json = F_UTF8 |
1494 | PPCODE: |
1896 | PPCODE: |
1495 | { |
1897 | { |
1496 | JSON json = { F_DEFAULT | F_UTF8 }; |
1898 | JSON json; |
|
|
1899 | json_init (&json); |
|
|
1900 | json.flags |= ix; |
1497 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1901 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1498 | } |
1902 | } |
1499 | |
1903 | |
|
|
1904 | |