… | |
… | |
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> |
|
|
11 | #include <inttypes.h> |
10 | |
12 | |
11 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
13 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
12 | # define snprintf _snprintf // C compilers have this in stdio.h |
14 | # define snprintf _snprintf // C compilers have this in stdio.h |
13 | #endif |
15 | #endif |
14 | |
16 | |
15 | // some old perls do not have this, try to make it work, no |
17 | // 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. |
18 | // guarantees, though. if it breaks, you get to keep the pieces. |
17 | #ifndef UTF8_MAXBYTES |
19 | #ifndef UTF8_MAXBYTES |
18 | # define UTF8_MAXBYTES 13 |
20 | # define UTF8_MAXBYTES 13 |
19 | #endif |
21 | #endif |
|
|
22 | |
|
|
23 | // compatibility with perl <5.18 |
|
|
24 | #ifndef HvNAMELEN_get |
|
|
25 | # define HvNAMELEN_get(hv) strlen (HvNAME (hv)) |
|
|
26 | #endif |
|
|
27 | #ifndef HvNAMELEN |
|
|
28 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
|
|
29 | #endif |
|
|
30 | #ifndef HvNAMEUTF8 |
|
|
31 | # define HvNAMEUTF8(hv) 0 |
|
|
32 | #endif |
|
|
33 | |
|
|
34 | // three extra for rounding, sign, and end of string |
|
|
35 | #define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3) |
20 | |
36 | |
21 | #define F_ASCII 0x00000001UL |
37 | #define F_ASCII 0x00000001UL |
22 | #define F_LATIN1 0x00000002UL |
38 | #define F_LATIN1 0x00000002UL |
23 | #define F_UTF8 0x00000004UL |
39 | #define F_UTF8 0x00000004UL |
24 | #define F_INDENT 0x00000008UL |
40 | #define F_INDENT 0x00000008UL |
… | |
… | |
28 | #define F_ALLOW_NONREF 0x00000100UL |
44 | #define F_ALLOW_NONREF 0x00000100UL |
29 | #define F_SHRINK 0x00000200UL |
45 | #define F_SHRINK 0x00000200UL |
30 | #define F_ALLOW_BLESSED 0x00000400UL |
46 | #define F_ALLOW_BLESSED 0x00000400UL |
31 | #define F_CONV_BLESSED 0x00000800UL |
47 | #define F_CONV_BLESSED 0x00000800UL |
32 | #define F_RELAXED 0x00001000UL |
48 | #define F_RELAXED 0x00001000UL |
33 | |
49 | #define F_ALLOW_UNKNOWN 0x00002000UL |
34 | #define F_MAXDEPTH 0xf8000000UL |
50 | #define F_ALLOW_TAGS 0x00004000UL |
35 | #define S_MAXDEPTH 27 |
|
|
36 | #define F_MAXSIZE 0x01f00000UL |
|
|
37 | #define S_MAXSIZE 20 |
|
|
38 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
51 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
39 | |
52 | |
40 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
|
|
41 | #define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE )) |
|
|
42 | |
|
|
43 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
53 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
44 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
|
|
45 | |
54 | |
46 | #define INIT_SIZE 32 // initial scalar size to be allocated |
55 | #define INIT_SIZE 32 // initial scalar size to be allocated |
47 | #define INDENT_STEP 3 // spaces per indentation level |
56 | #define INDENT_STEP 3 // spaces per indentation level |
48 | |
57 | |
49 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
58 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
50 | |
59 | |
|
|
60 | #define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8) |
|
|
61 | |
51 | #define SB do { |
62 | #define SB do { |
52 | #define SE } while (0) |
63 | #define SE } while (0) |
53 | |
64 | |
54 | #if __GNUC__ >= 3 |
65 | #if __GNUC__ >= 3 |
55 | # define expect(expr,value) __builtin_expect ((expr),(value)) |
66 | # define expect(expr,value) __builtin_expect ((expr), (value)) |
56 | # define inline inline |
67 | # define INLINE static inline |
57 | #else |
68 | #else |
58 | # define expect(expr,value) (expr) |
69 | # define expect(expr,value) (expr) |
59 | # define inline static |
70 | # define INLINE static |
60 | #endif |
71 | #endif |
61 | |
72 | |
62 | #define expect_false(expr) expect ((expr) != 0, 0) |
73 | #define expect_false(expr) expect ((expr) != 0, 0) |
63 | #define expect_true(expr) expect ((expr) != 0, 1) |
74 | #define expect_true(expr) expect ((expr) != 0, 1) |
|
|
75 | |
|
|
76 | #define IN_RANGE_INC(type,val,beg,end) \ |
|
|
77 | ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
|
|
78 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
|
|
79 | |
|
|
80 | #define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)" |
64 | |
81 | |
65 | #ifdef USE_ITHREADS |
82 | #ifdef USE_ITHREADS |
66 | # define JSON_SLOW 1 |
83 | # define JSON_SLOW 1 |
67 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
84 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
|
|
85 | # define BOOL_STASH (bool_stash ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1)) |
68 | #else |
86 | #else |
69 | # define JSON_SLOW 0 |
87 | # define JSON_SLOW 0 |
70 | # define JSON_STASH json_stash |
88 | # define JSON_STASH json_stash |
|
|
89 | # define BOOL_STASH bool_stash |
71 | #endif |
90 | #endif |
72 | |
91 | |
73 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
92 | // the amount of HEs to allocate on the stack, when sorting keys |
74 | static SV *json_true, *json_false; |
93 | #define STACK_HES 64 |
|
|
94 | |
|
|
95 | static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean:: |
|
|
96 | static SV *bool_true, *bool_false, *sv_json; |
|
|
97 | |
|
|
98 | enum { |
|
|
99 | INCR_M_WS = 0, // initial whitespace skipping, must be 0 |
|
|
100 | INCR_M_STR, // inside string |
|
|
101 | INCR_M_BS, // inside backslash |
|
|
102 | INCR_M_C0, // inside comment in initial whitespace sequence |
|
|
103 | INCR_M_C1, // inside comment in other places |
|
|
104 | INCR_M_JSON // outside anything, count nesting |
|
|
105 | }; |
|
|
106 | |
|
|
107 | #define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) |
75 | |
108 | |
76 | typedef struct { |
109 | typedef struct { |
77 | U32 flags; |
110 | U32 flags; |
|
|
111 | U32 max_depth; |
|
|
112 | STRLEN max_size; |
|
|
113 | |
78 | SV *cb_object; |
114 | SV *cb_object; |
79 | HV *cb_sk_object; |
115 | HV *cb_sk_object; |
|
|
116 | |
|
|
117 | // for the incremental parser |
|
|
118 | SV *incr_text; // the source text so far |
|
|
119 | STRLEN incr_pos; // the current offset into the text |
|
|
120 | int incr_nest; // {[]}-nesting level |
|
|
121 | unsigned char incr_mode; |
80 | } JSON; |
122 | } JSON; |
|
|
123 | |
|
|
124 | INLINE void |
|
|
125 | json_init (JSON *json) |
|
|
126 | { |
|
|
127 | Zero (json, 1, JSON); |
|
|
128 | json->max_depth = 512; |
|
|
129 | } |
81 | |
130 | |
82 | ///////////////////////////////////////////////////////////////////////////// |
131 | ///////////////////////////////////////////////////////////////////////////// |
83 | // utility functions |
132 | // utility functions |
84 | |
133 | |
85 | inline void |
134 | INLINE SV * |
|
|
135 | get_bool (const char *name) |
|
|
136 | { |
|
|
137 | SV *sv = get_sv (name, 1); |
|
|
138 | |
|
|
139 | SvREADONLY_on (sv); |
|
|
140 | SvREADONLY_on (SvRV (sv)); |
|
|
141 | |
|
|
142 | return sv; |
|
|
143 | } |
|
|
144 | |
|
|
145 | INLINE void |
86 | shrink (SV *sv) |
146 | shrink (SV *sv) |
87 | { |
147 | { |
88 | sv_utf8_downgrade (sv, 1); |
148 | sv_utf8_downgrade (sv, 1); |
|
|
149 | |
89 | if (SvLEN (sv) > SvCUR (sv) + 1) |
150 | if (SvLEN (sv) > SvCUR (sv) + 1) |
90 | { |
151 | { |
91 | #ifdef SvPV_shrink_to_cur |
152 | #ifdef SvPV_shrink_to_cur |
92 | SvPV_shrink_to_cur (sv); |
153 | SvPV_shrink_to_cur (sv); |
93 | #elif defined (SvPV_renew) |
154 | #elif defined (SvPV_renew) |
94 | SvPV_renew (sv, SvCUR (sv) + 1); |
155 | SvPV_renew (sv, SvCUR (sv) + 1); |
95 | #endif |
156 | #endif |
96 | } |
157 | } |
97 | } |
158 | } |
98 | |
159 | |
|
|
160 | /* adds two STRLENs together, slow, and with paranoia */ |
|
|
161 | STRLEN |
|
|
162 | strlen_sum (STRLEN l1, STRLEN l2) |
|
|
163 | { |
|
|
164 | size_t sum = l1 + l2; |
|
|
165 | |
|
|
166 | if (sum < (size_t)l2 || sum != (size_t)(STRLEN)sum) |
|
|
167 | croak ("JSON::XS: string size overflow"); |
|
|
168 | |
|
|
169 | return sum; |
|
|
170 | } |
|
|
171 | |
|
|
172 | /* similar to SvGROW, but somewhat safer and guarantees exponential realloc strategy */ |
|
|
173 | static char * |
|
|
174 | json_sv_grow (SV *sv, size_t len1, size_t len2) |
|
|
175 | { |
|
|
176 | len1 = strlen_sum (len1, len2); |
|
|
177 | len1 = strlen_sum (len1, len1 >> 1); |
|
|
178 | |
|
|
179 | if (len1 > 4096 - 24) |
|
|
180 | len1 = (len1 | 4095) - 24; |
|
|
181 | |
|
|
182 | return SvGROW (sv, len1); |
|
|
183 | } |
|
|
184 | |
99 | // decode an utf-8 character and return it, or (UV)-1 in |
185 | // decode a utf-8 character and return it, or (UV)-1 in |
100 | // case of an error. |
186 | // case of an error. |
101 | // we special-case "safe" characters from U+80 .. U+7FF, |
187 | // we special-case "safe" characters from U+80 .. U+7FF, |
102 | // but use the very good perl function to parse anything else. |
188 | // but use the very good perl function to parse anything else. |
103 | // note that we never call this function for a ascii codepoints |
189 | // note that we never call this function for a ascii codepoints |
104 | inline UV |
190 | INLINE UV |
105 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
191 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
106 | { |
192 | { |
107 | if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) |
193 | if (expect_true (len >= 2 |
108 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
194 | && IN_RANGE_INC (char, s[0], 0xc2, 0xdf) |
109 | else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
195 | && IN_RANGE_INC (char, s[1], 0x80, 0xbf))) |
110 | { |
196 | { |
111 | *clen = 2; |
197 | *clen = 2; |
112 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
198 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
113 | } |
199 | } |
114 | else |
200 | else |
|
|
201 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
|
|
202 | } |
|
|
203 | |
|
|
204 | // likewise for encoding, also never called for ascii codepoints |
|
|
205 | // this function takes advantage of this fact, although current gccs |
|
|
206 | // seem to optimise the check for >= 0x80 away anyways |
|
|
207 | INLINE unsigned char * |
|
|
208 | encode_utf8 (unsigned char *s, UV ch) |
|
|
209 | { |
|
|
210 | if (expect_false (ch < 0x000080)) |
|
|
211 | *s++ = ch; |
|
|
212 | else if (expect_true (ch < 0x000800)) |
|
|
213 | *s++ = 0xc0 | ( ch >> 6), |
|
|
214 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
215 | else if ( ch < 0x010000) |
|
|
216 | *s++ = 0xe0 | ( ch >> 12), |
|
|
217 | *s++ = 0x80 | ((ch >> 6) & 0x3f), |
|
|
218 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
219 | else if ( ch < 0x110000) |
|
|
220 | *s++ = 0xf0 | ( ch >> 18), |
|
|
221 | *s++ = 0x80 | ((ch >> 12) & 0x3f), |
|
|
222 | *s++ = 0x80 | ((ch >> 6) & 0x3f), |
|
|
223 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
224 | |
|
|
225 | return s; |
|
|
226 | } |
|
|
227 | |
|
|
228 | // convert offset pointer to character index, sv must be string |
|
|
229 | static STRLEN |
|
|
230 | ptr_to_index (SV *sv, char *offset) |
|
|
231 | { |
|
|
232 | return SvUTF8 (sv) |
|
|
233 | ? utf8_distance (offset, SvPVX (sv)) |
|
|
234 | : offset - SvPVX (sv); |
|
|
235 | } |
|
|
236 | |
|
|
237 | ///////////////////////////////////////////////////////////////////////////// |
|
|
238 | // fp hell |
|
|
239 | |
|
|
240 | // scan a group of digits, and a trailing exponent |
|
|
241 | static void |
|
|
242 | json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth) |
|
|
243 | { |
|
|
244 | UV uaccum = 0; |
|
|
245 | int eaccum = 0; |
|
|
246 | |
|
|
247 | // if we recurse too deep, skip all remaining digits |
|
|
248 | // to avoid a stack overflow attack |
|
|
249 | if (expect_false (--maxdepth <= 0)) |
|
|
250 | while (((U8)*s - '0') < 10) |
|
|
251 | ++s; |
|
|
252 | |
|
|
253 | for (;;) |
|
|
254 | { |
|
|
255 | U8 dig = (U8)*s - '0'; |
|
|
256 | |
|
|
257 | if (expect_false (dig >= 10)) |
|
|
258 | { |
|
|
259 | if (dig == (U8)((U8)'.' - (U8)'0')) |
|
|
260 | { |
|
|
261 | ++s; |
|
|
262 | json_atof_scan1 (s, accum, expo, 1, maxdepth); |
|
|
263 | } |
|
|
264 | else if ((dig | ' ') == 'e' - '0') |
|
|
265 | { |
|
|
266 | int exp2 = 0; |
|
|
267 | int neg = 0; |
|
|
268 | |
|
|
269 | ++s; |
|
|
270 | |
|
|
271 | if (*s == '-') |
|
|
272 | { |
|
|
273 | ++s; |
|
|
274 | neg = 1; |
|
|
275 | } |
|
|
276 | else if (*s == '+') |
|
|
277 | ++s; |
|
|
278 | |
|
|
279 | while ((dig = (U8)*s - '0') < 10) |
|
|
280 | exp2 = exp2 * 10 + *s++ - '0'; |
|
|
281 | |
|
|
282 | *expo += neg ? -exp2 : exp2; |
|
|
283 | } |
|
|
284 | |
|
|
285 | break; |
|
|
286 | } |
|
|
287 | |
|
|
288 | ++s; |
|
|
289 | |
|
|
290 | uaccum = uaccum * 10 + dig; |
|
|
291 | ++eaccum; |
|
|
292 | |
|
|
293 | // if we have too many digits, then recurse for more |
|
|
294 | // we actually do this for rather few digits |
|
|
295 | if (uaccum >= (UV_MAX - 9) / 10) |
|
|
296 | { |
|
|
297 | if (postdp) *expo -= eaccum; |
|
|
298 | json_atof_scan1 (s, accum, expo, postdp, maxdepth); |
|
|
299 | if (postdp) *expo += eaccum; |
|
|
300 | |
|
|
301 | break; |
|
|
302 | } |
115 | { |
303 | } |
116 | *clen = (STRLEN)-1; |
304 | |
|
|
305 | // this relies greatly on the quality of the pow () |
|
|
306 | // implementation of the platform, but a good |
|
|
307 | // implementation is hard to beat. |
|
|
308 | // (IEEE 754 conformant ones are required to be exact) |
|
|
309 | if (postdp) *expo -= eaccum; |
|
|
310 | *accum += uaccum * Perl_pow (10., *expo); |
|
|
311 | *expo += eaccum; |
|
|
312 | } |
|
|
313 | |
|
|
314 | static NV |
|
|
315 | json_atof (const char *s) |
|
|
316 | { |
|
|
317 | NV accum = 0.; |
|
|
318 | int expo = 0; |
|
|
319 | int neg = 0; |
|
|
320 | |
|
|
321 | if (*s == '-') |
|
|
322 | { |
|
|
323 | ++s; |
|
|
324 | neg = 1; |
|
|
325 | } |
|
|
326 | |
|
|
327 | // a recursion depth of ten gives us >>500 bits |
|
|
328 | json_atof_scan1 (s, &accum, &expo, 0, 10); |
|
|
329 | |
|
|
330 | return neg ? -accum : accum; |
|
|
331 | } |
|
|
332 | |
|
|
333 | // target of scalar reference is bool? -1 == nope, 0 == false, 1 == true |
|
|
334 | static int |
|
|
335 | ref_bool_type (SV *sv) |
|
|
336 | { |
|
|
337 | svtype svt = SvTYPE (sv); |
|
|
338 | |
|
|
339 | if (svt < SVt_PVAV) |
|
|
340 | { |
|
|
341 | STRLEN len = 0; |
|
|
342 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
343 | |
|
|
344 | if (len == 1) |
|
|
345 | if (*pv == '1') |
|
|
346 | return 1; |
|
|
347 | else if (*pv == '0') |
|
|
348 | return 0; |
|
|
349 | } |
|
|
350 | |
|
|
351 | return -1; |
|
|
352 | } |
|
|
353 | |
|
|
354 | // returns whether scalar is not a reference in the sense of allow_nonref |
|
|
355 | static int |
|
|
356 | json_nonref (SV *scalar) |
|
|
357 | { |
|
|
358 | if (!SvROK (scalar)) |
|
|
359 | return 1; |
|
|
360 | |
|
|
361 | scalar = SvRV (scalar); |
|
|
362 | |
|
|
363 | if (SvTYPE (scalar) >= SVt_PVMG) |
|
|
364 | { |
|
|
365 | if (SvSTASH (scalar) == bool_stash) |
117 | return (UV)-1; |
366 | return 1; |
|
|
367 | |
|
|
368 | if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0) |
|
|
369 | return 1; |
118 | } |
370 | } |
|
|
371 | |
|
|
372 | return 0; |
119 | } |
373 | } |
120 | |
374 | |
121 | ///////////////////////////////////////////////////////////////////////////// |
375 | ///////////////////////////////////////////////////////////////////////////// |
122 | // encoder |
376 | // encoder |
123 | |
377 | |
… | |
… | |
127 | char *cur; // SvPVX (sv) + current output position |
381 | char *cur; // SvPVX (sv) + current output position |
128 | char *end; // SvEND (sv) |
382 | char *end; // SvEND (sv) |
129 | SV *sv; // result scalar |
383 | SV *sv; // result scalar |
130 | JSON json; |
384 | JSON json; |
131 | U32 indent; // indentation level |
385 | U32 indent; // indentation level |
132 | U32 maxdepth; // max. indentation/recursion level |
386 | UV limit; // escape character values >= this value when encoding |
133 | } enc_t; |
387 | } enc_t; |
134 | |
388 | |
135 | inline void |
389 | INLINE void |
136 | need (enc_t *enc, STRLEN len) |
390 | need (enc_t *enc, STRLEN len) |
137 | { |
391 | { |
138 | if (expect_false (enc->cur + len >= enc->end)) |
392 | if (expect_false ((uintptr_t)(enc->end - enc->cur) < len)) |
139 | { |
393 | { |
140 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
394 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
141 | SvGROW (enc->sv, cur + len + 1); |
395 | char *buf = json_sv_grow (enc->sv, cur, len); |
142 | enc->cur = SvPVX (enc->sv) + cur; |
396 | enc->cur = buf + cur; |
143 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
397 | enc->end = buf + SvLEN (enc->sv) - 1; |
144 | } |
398 | } |
145 | } |
399 | } |
146 | |
400 | |
147 | inline void |
401 | INLINE void |
148 | encode_ch (enc_t *enc, char ch) |
402 | encode_ch (enc_t *enc, char ch) |
149 | { |
403 | { |
150 | need (enc, 1); |
404 | need (enc, 1); |
151 | *enc->cur++ = ch; |
405 | *enc->cur++ = ch; |
152 | } |
406 | } |
… | |
… | |
164 | |
418 | |
165 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
419 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
166 | { |
420 | { |
167 | if (expect_false (ch == '"')) // but with slow exceptions |
421 | if (expect_false (ch == '"')) // but with slow exceptions |
168 | { |
422 | { |
169 | need (enc, len += 1); |
423 | need (enc, len + 1); |
170 | *enc->cur++ = '\\'; |
424 | *enc->cur++ = '\\'; |
171 | *enc->cur++ = '"'; |
425 | *enc->cur++ = '"'; |
172 | } |
426 | } |
173 | else if (expect_false (ch == '\\')) |
427 | else if (expect_false (ch == '\\')) |
174 | { |
428 | { |
175 | need (enc, len += 1); |
429 | need (enc, len + 1); |
176 | *enc->cur++ = '\\'; |
430 | *enc->cur++ = '\\'; |
177 | *enc->cur++ = '\\'; |
431 | *enc->cur++ = '\\'; |
178 | } |
432 | } |
179 | else |
433 | else |
180 | *enc->cur++ = ch; |
434 | *enc->cur++ = ch; |
… | |
… | |
183 | } |
437 | } |
184 | else |
438 | else |
185 | { |
439 | { |
186 | switch (ch) |
440 | switch (ch) |
187 | { |
441 | { |
188 | case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; |
442 | case '\010': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; |
189 | case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; |
443 | case '\011': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; |
190 | case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; |
444 | case '\012': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; |
191 | case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; |
445 | case '\014': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; |
192 | case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; |
446 | case '\015': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; |
193 | |
447 | |
194 | default: |
448 | default: |
195 | { |
449 | { |
196 | STRLEN clen; |
450 | STRLEN clen; |
197 | UV uch; |
451 | UV uch; |
… | |
… | |
206 | { |
460 | { |
207 | uch = ch; |
461 | uch = ch; |
208 | clen = 1; |
462 | clen = 1; |
209 | } |
463 | } |
210 | |
464 | |
211 | if (uch > 0x10FFFFUL) |
465 | 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 | { |
466 | { |
216 | if (uch > 0xFFFFUL) |
467 | if (uch >= 0x10000UL) |
217 | { |
468 | { |
|
|
469 | if (uch >= 0x110000UL) |
|
|
470 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
|
|
471 | |
218 | need (enc, len += 11); |
472 | need (enc, len + 11); |
219 | sprintf (enc->cur, "\\u%04x\\u%04x", |
473 | sprintf (enc->cur, "\\u%04x\\u%04x", |
220 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
474 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
221 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
475 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
222 | enc->cur += 12; |
476 | enc->cur += 12; |
223 | } |
477 | } |
224 | else |
478 | else |
225 | { |
479 | { |
226 | static char hexdigit [16] = "0123456789abcdef"; |
|
|
227 | need (enc, len += 5); |
480 | need (enc, len + 5); |
228 | *enc->cur++ = '\\'; |
481 | *enc->cur++ = '\\'; |
229 | *enc->cur++ = 'u'; |
482 | *enc->cur++ = 'u'; |
230 | *enc->cur++ = hexdigit [ uch >> 12 ]; |
483 | *enc->cur++ = PL_hexdigit [ uch >> 12 ]; |
231 | *enc->cur++ = hexdigit [(uch >> 8) & 15]; |
484 | *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; |
232 | *enc->cur++ = hexdigit [(uch >> 4) & 15]; |
485 | *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; |
233 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
486 | *enc->cur++ = PL_hexdigit [(uch >> 0) & 15]; |
234 | } |
487 | } |
235 | |
488 | |
236 | str += clen; |
489 | str += clen; |
237 | } |
490 | } |
238 | else if (enc->json.flags & F_LATIN1) |
491 | else if (enc->json.flags & F_LATIN1) |
… | |
… | |
240 | *enc->cur++ = uch; |
493 | *enc->cur++ = uch; |
241 | str += clen; |
494 | str += clen; |
242 | } |
495 | } |
243 | else if (is_utf8) |
496 | else if (is_utf8) |
244 | { |
497 | { |
245 | need (enc, len += clen); |
498 | need (enc, len + clen); |
246 | do |
499 | do |
247 | { |
500 | { |
248 | *enc->cur++ = *str++; |
501 | *enc->cur++ = *str++; |
249 | } |
502 | } |
250 | while (--clen); |
503 | while (--clen); |
251 | } |
504 | } |
252 | else |
505 | else |
253 | { |
506 | { |
254 | need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
507 | need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
255 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
508 | enc->cur = encode_utf8 (enc->cur, uch); |
256 | ++str; |
509 | ++str; |
257 | } |
510 | } |
258 | } |
511 | } |
259 | } |
512 | } |
260 | } |
513 | } |
261 | |
514 | |
262 | --len; |
515 | --len; |
263 | } |
516 | } |
264 | } |
517 | } |
265 | |
518 | |
266 | inline void |
519 | INLINE void |
267 | encode_indent (enc_t *enc) |
520 | encode_indent (enc_t *enc) |
268 | { |
521 | { |
269 | if (enc->json.flags & F_INDENT) |
522 | if (enc->json.flags & F_INDENT) |
270 | { |
523 | { |
271 | int spaces = enc->indent * INDENT_STEP; |
524 | int spaces = enc->indent * INDENT_STEP; |
… | |
… | |
274 | memset (enc->cur, ' ', spaces); |
527 | memset (enc->cur, ' ', spaces); |
275 | enc->cur += spaces; |
528 | enc->cur += spaces; |
276 | } |
529 | } |
277 | } |
530 | } |
278 | |
531 | |
279 | inline void |
532 | INLINE void |
280 | encode_space (enc_t *enc) |
533 | encode_space (enc_t *enc) |
281 | { |
534 | { |
282 | need (enc, 1); |
535 | need (enc, 1); |
283 | encode_ch (enc, ' '); |
536 | encode_ch (enc, ' '); |
284 | } |
537 | } |
285 | |
538 | |
286 | inline void |
539 | INLINE void |
287 | encode_nl (enc_t *enc) |
540 | encode_nl (enc_t *enc) |
288 | { |
541 | { |
289 | if (enc->json.flags & F_INDENT) |
542 | if (enc->json.flags & F_INDENT) |
290 | { |
543 | { |
291 | need (enc, 1); |
544 | need (enc, 1); |
292 | encode_ch (enc, '\n'); |
545 | encode_ch (enc, '\n'); |
293 | } |
546 | } |
294 | } |
547 | } |
295 | |
548 | |
296 | inline void |
549 | INLINE void |
297 | encode_comma (enc_t *enc) |
550 | encode_comma (enc_t *enc) |
298 | { |
551 | { |
299 | encode_ch (enc, ','); |
552 | encode_ch (enc, ','); |
300 | |
553 | |
301 | if (enc->json.flags & F_INDENT) |
554 | if (enc->json.flags & F_INDENT) |
… | |
… | |
309 | static void |
562 | static void |
310 | encode_av (enc_t *enc, AV *av) |
563 | encode_av (enc_t *enc, AV *av) |
311 | { |
564 | { |
312 | int i, len = av_len (av); |
565 | int i, len = av_len (av); |
313 | |
566 | |
314 | if (enc->indent >= enc->maxdepth) |
567 | if (enc->indent >= enc->json.max_depth) |
315 | croak ("data structure too deep (hit recursion limit)"); |
568 | croak (ERR_NESTING_EXCEEDED); |
316 | |
569 | |
317 | encode_ch (enc, '['); |
570 | encode_ch (enc, '['); |
318 | |
571 | |
319 | if (len >= 0) |
572 | if (len >= 0) |
320 | { |
573 | { |
321 | encode_nl (enc); ++enc->indent; |
574 | encode_nl (enc); ++enc->indent; |
322 | |
575 | |
323 | for (i = 0; i <= len; ++i) |
576 | for (i = 0; i <= len; ++i) |
… | |
… | |
335 | encode_comma (enc); |
588 | encode_comma (enc); |
336 | } |
589 | } |
337 | |
590 | |
338 | encode_nl (enc); --enc->indent; encode_indent (enc); |
591 | encode_nl (enc); --enc->indent; encode_indent (enc); |
339 | } |
592 | } |
340 | |
593 | |
341 | encode_ch (enc, ']'); |
594 | encode_ch (enc, ']'); |
342 | } |
595 | } |
343 | |
596 | |
344 | static void |
597 | static void |
345 | encode_hk (enc_t *enc, HE *he) |
598 | encode_hk (enc_t *enc, HE *he) |
… | |
… | |
349 | if (HeKLEN (he) == HEf_SVKEY) |
602 | if (HeKLEN (he) == HEf_SVKEY) |
350 | { |
603 | { |
351 | SV *sv = HeSVKEY (he); |
604 | SV *sv = HeSVKEY (he); |
352 | STRLEN len; |
605 | STRLEN len; |
353 | char *str; |
606 | char *str; |
354 | |
607 | |
355 | SvGETMAGIC (sv); |
608 | SvGETMAGIC (sv); |
356 | str = SvPV (sv, len); |
609 | str = SvPV (sv, len); |
357 | |
610 | |
358 | encode_str (enc, str, len, SvUTF8 (sv)); |
611 | encode_str (enc, str, len, SvUTF8 (sv)); |
359 | } |
612 | } |
… | |
… | |
394 | |
647 | |
395 | static void |
648 | static void |
396 | encode_hv (enc_t *enc, HV *hv) |
649 | encode_hv (enc_t *enc, HV *hv) |
397 | { |
650 | { |
398 | HE *he; |
651 | HE *he; |
399 | int count; |
|
|
400 | |
652 | |
401 | if (enc->indent >= enc->maxdepth) |
653 | if (enc->indent >= enc->json.max_depth) |
402 | croak ("data structure too deep (hit recursion limit)"); |
654 | croak (ERR_NESTING_EXCEEDED); |
403 | |
655 | |
404 | encode_ch (enc, '{'); |
656 | encode_ch (enc, '{'); |
405 | |
657 | |
406 | // for canonical output we have to sort by keys first |
658 | // for canonical output we have to sort by keys first |
407 | // actually, this is mostly due to the stupid so-called |
659 | // actually, this is mostly due to the stupid so-called |
408 | // security workaround added somewhere in 5.8.x. |
660 | // security workaround added somewhere in 5.8.x |
409 | // that randomises hash orderings |
661 | // that randomises hash orderings |
410 | if (enc->json.flags & F_CANONICAL) |
662 | if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv)) |
411 | { |
663 | { |
412 | int count = hv_iterinit (hv); |
664 | int count = hv_iterinit (hv); |
413 | |
665 | |
414 | if (SvMAGICAL (hv)) |
666 | if (SvMAGICAL (hv)) |
415 | { |
667 | { |
… | |
… | |
426 | } |
678 | } |
427 | |
679 | |
428 | if (count) |
680 | if (count) |
429 | { |
681 | { |
430 | int i, fast = 1; |
682 | int i, fast = 1; |
431 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
683 | HE *hes_stack [STACK_HES]; |
432 | HE **hes = _alloca (count * sizeof (HE)); |
684 | HE **hes = hes_stack; |
433 | #else |
685 | |
434 | HE *hes [count]; // if your compiler dies here, you need to enable C99 mode |
686 | // allocate larger arrays on the heap |
435 | #endif |
687 | if (count > STACK_HES) |
|
|
688 | { |
|
|
689 | SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes))); |
|
|
690 | hes = (HE **)SvPVX (sv); |
|
|
691 | } |
436 | |
692 | |
437 | i = 0; |
693 | i = 0; |
438 | while ((he = hv_iternext (hv))) |
694 | while ((he = hv_iternext (hv))) |
439 | { |
695 | { |
440 | hes [i++] = he; |
696 | hes [i++] = he; |
… | |
… | |
509 | // encode objects, arrays and special \0=false and \1=true values. |
765 | // encode objects, arrays and special \0=false and \1=true values. |
510 | static void |
766 | static void |
511 | encode_rv (enc_t *enc, SV *sv) |
767 | encode_rv (enc_t *enc, SV *sv) |
512 | { |
768 | { |
513 | svtype svt; |
769 | svtype svt; |
|
|
770 | GV *method; |
514 | |
771 | |
515 | SvGETMAGIC (sv); |
772 | SvGETMAGIC (sv); |
516 | svt = SvTYPE (sv); |
773 | svt = SvTYPE (sv); |
517 | |
774 | |
518 | if (expect_false (SvOBJECT (sv))) |
775 | if (expect_false (SvOBJECT (sv))) |
519 | { |
776 | { |
520 | HV *stash = !JSON_SLOW || json_boolean_stash |
777 | HV *stash = SvSTASH (sv); |
521 | ? json_boolean_stash |
|
|
522 | : gv_stashpv ("JSON::XS::Boolean", 1); |
|
|
523 | |
778 | |
524 | if (SvSTASH (sv) == stash) |
779 | if (stash == bool_stash) |
525 | { |
780 | { |
526 | if (SvIV (sv)) |
781 | if (SvIV (sv)) |
527 | encode_str (enc, "true", 4, 0); |
782 | encode_str (enc, "true", 4, 0); |
528 | else |
783 | else |
529 | encode_str (enc, "false", 5, 0); |
784 | encode_str (enc, "false", 5, 0); |
530 | } |
785 | } |
|
|
786 | else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0))) |
|
|
787 | { |
|
|
788 | int count; |
|
|
789 | dSP; |
|
|
790 | |
|
|
791 | ENTER; SAVETMPS; |
|
|
792 | SAVESTACK_POS (); |
|
|
793 | PUSHMARK (SP); |
|
|
794 | EXTEND (SP, 2); |
|
|
795 | // we re-bless the reference to get overload and other niceties right |
|
|
796 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
|
|
797 | PUSHs (sv_json); |
|
|
798 | |
|
|
799 | PUTBACK; |
|
|
800 | count = call_sv ((SV *)GvCV (method), G_ARRAY); |
|
|
801 | SPAGAIN; |
|
|
802 | |
|
|
803 | // catch this surprisingly common error |
|
|
804 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
805 | croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
806 | |
|
|
807 | encode_ch (enc, '('); |
|
|
808 | encode_ch (enc, '"'); |
|
|
809 | encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash)); |
|
|
810 | encode_ch (enc, '"'); |
|
|
811 | encode_ch (enc, ')'); |
|
|
812 | encode_ch (enc, '['); |
|
|
813 | |
|
|
814 | while (count) |
|
|
815 | { |
|
|
816 | encode_sv (enc, SP[1 - count--]); |
|
|
817 | |
|
|
818 | if (count) |
|
|
819 | encode_ch (enc, ','); |
|
|
820 | } |
|
|
821 | |
|
|
822 | encode_ch (enc, ']'); |
|
|
823 | |
|
|
824 | FREETMPS; LEAVE; |
|
|
825 | } |
|
|
826 | else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0))) |
|
|
827 | { |
|
|
828 | dSP; |
|
|
829 | |
|
|
830 | ENTER; SAVETMPS; |
|
|
831 | PUSHMARK (SP); |
|
|
832 | // we re-bless the reference to get overload and other niceties right |
|
|
833 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
|
|
834 | |
|
|
835 | // calling with G_SCALAR ensures that we always get a 1 return value |
|
|
836 | PUTBACK; |
|
|
837 | call_sv ((SV *)GvCV (method), G_SCALAR); |
|
|
838 | SPAGAIN; |
|
|
839 | |
|
|
840 | // catch this surprisingly common error |
|
|
841 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
842 | croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
843 | |
|
|
844 | sv = POPs; |
|
|
845 | PUTBACK; |
|
|
846 | |
|
|
847 | encode_sv (enc, sv); |
|
|
848 | |
|
|
849 | FREETMPS; LEAVE; |
|
|
850 | } |
|
|
851 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
852 | encode_str (enc, "null", 4, 0); |
531 | else |
853 | else |
532 | { |
|
|
533 | #if 0 |
|
|
534 | if (0 && sv_derived_from (rv, "JSON::Literal")) |
|
|
535 | { |
|
|
536 | // not yet |
|
|
537 | } |
|
|
538 | #endif |
|
|
539 | if (enc->json.flags & F_CONV_BLESSED) |
|
|
540 | { |
|
|
541 | // we re-bless the reference to get overload and other niceties right |
|
|
542 | GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0); |
|
|
543 | |
|
|
544 | if (to_json) |
|
|
545 | { |
|
|
546 | dSP; |
|
|
547 | |
|
|
548 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
549 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); |
|
|
550 | |
|
|
551 | // calling with G_SCALAR ensures that we always get a 1 return value |
|
|
552 | PUTBACK; |
|
|
553 | call_sv ((SV *)GvCV (to_json), G_SCALAR); |
|
|
554 | SPAGAIN; |
|
|
555 | |
|
|
556 | // catch this surprisingly common error |
|
|
557 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
558 | croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
559 | |
|
|
560 | sv = POPs; |
|
|
561 | PUTBACK; |
|
|
562 | |
|
|
563 | encode_sv (enc, sv); |
|
|
564 | |
|
|
565 | FREETMPS; LEAVE; |
|
|
566 | } |
|
|
567 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
568 | encode_str (enc, "null", 4, 0); |
|
|
569 | else |
|
|
570 | croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it", |
|
|
571 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
572 | } |
|
|
573 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
574 | encode_str (enc, "null", 4, 0); |
|
|
575 | else |
|
|
576 | croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", |
854 | croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)", |
577 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
855 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
578 | } |
|
|
579 | } |
856 | } |
580 | else if (svt == SVt_PVHV) |
857 | else if (svt == SVt_PVHV) |
581 | encode_hv (enc, (HV *)sv); |
858 | encode_hv (enc, (HV *)sv); |
582 | else if (svt == SVt_PVAV) |
859 | else if (svt == SVt_PVAV) |
583 | encode_av (enc, (AV *)sv); |
860 | encode_av (enc, (AV *)sv); |
584 | else if (svt < SVt_PVAV) |
861 | else if (svt < SVt_PVAV) |
585 | { |
862 | { |
586 | STRLEN len = 0; |
863 | int bool_type = ref_bool_type (sv); |
587 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
588 | |
864 | |
589 | if (len == 1 && *pv == '1') |
865 | if (bool_type == 1) |
590 | encode_str (enc, "true", 4, 0); |
866 | encode_str (enc, "true", 4, 0); |
591 | else if (len == 1 && *pv == '0') |
867 | else if (bool_type == 0) |
592 | encode_str (enc, "false", 5, 0); |
868 | encode_str (enc, "false", 5, 0); |
|
|
869 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
870 | encode_str (enc, "null", 4, 0); |
593 | else |
871 | else |
594 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
872 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
595 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
873 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
596 | } |
874 | } |
|
|
875 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
876 | encode_str (enc, "null", 4, 0); |
597 | else |
877 | else |
598 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
878 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
599 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
879 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
600 | } |
880 | } |
601 | |
881 | |
… | |
… | |
619 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
899 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
620 | enc->cur += strlen (enc->cur); |
900 | enc->cur += strlen (enc->cur); |
621 | } |
901 | } |
622 | else if (SvIOKp (sv)) |
902 | else if (SvIOKp (sv)) |
623 | { |
903 | { |
624 | // we assume we can always read an IV as a UV |
904 | // we assume we can always read an IV as a UV and vice versa |
625 | if (SvUV (sv) & ~(UV)0x7fff) |
905 | // we assume two's complement |
626 | { |
906 | // we assume no aliasing issues in the union |
627 | // large integer, use the (rather slow) snprintf way. |
907 | if (SvIsUV (sv) ? SvUVX (sv) <= 59000 |
628 | need (enc, sizeof (UV) * 3); |
908 | : 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 | { |
909 | { |
636 | // optimise the "small number case" |
910 | // optimise the "small number case" |
637 | // code will likely be branchless and use only a single multiplication |
911 | // code will likely be branchless and use only a single multiplication |
|
|
912 | // works for numbers up to 59074 |
638 | I32 i = SvIV (sv); |
913 | I32 i = SvIVX (sv); |
639 | U32 u; |
914 | U32 u; |
640 | char digit, nz = 0; |
915 | char digit, nz = 0; |
641 | |
916 | |
642 | need (enc, 6); |
917 | need (enc, 6); |
643 | |
918 | |
… | |
… | |
649 | |
924 | |
650 | // now output digit by digit, each time masking out the integer part |
925 | // 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, |
926 | // and multiplying by 5 while moving the decimal point one to the right, |
652 | // resulting in a net multiplication by 10. |
927 | // resulting in a net multiplication by 10. |
653 | // we always write the digit to memory but conditionally increment |
928 | // we always write the digit to memory but conditionally increment |
654 | // the pointer, to ease the usage of conditional move instructions. |
929 | // 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; |
930 | 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; |
931 | 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; |
932 | 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; |
933 | 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' |
934 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
660 | } |
935 | } |
|
|
936 | else |
|
|
937 | { |
|
|
938 | // large integer, use the (rather slow) snprintf way. |
|
|
939 | need (enc, IVUV_MAXCHARS); |
|
|
940 | enc->cur += |
|
|
941 | SvIsUV(sv) |
|
|
942 | ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) |
|
|
943 | : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); |
|
|
944 | } |
661 | } |
945 | } |
662 | else if (SvROK (sv)) |
946 | else if (SvROK (sv)) |
663 | encode_rv (enc, SvRV (sv)); |
947 | encode_rv (enc, SvRV (sv)); |
664 | else if (!SvOK (sv)) |
948 | else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) |
665 | encode_str (enc, "null", 4, 0); |
949 | encode_str (enc, "null", 4, 0); |
666 | else |
950 | else |
667 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
951 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data", |
668 | SvPV_nolen (sv), SvFLAGS (sv)); |
952 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
669 | } |
953 | } |
670 | |
954 | |
671 | static SV * |
955 | static SV * |
672 | encode_json (SV *scalar, JSON *json) |
956 | encode_json (SV *scalar, JSON *json) |
673 | { |
957 | { |
674 | enc_t enc; |
958 | enc_t enc; |
675 | |
959 | |
676 | if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
960 | if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar)) |
677 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
961 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
678 | |
962 | |
679 | enc.json = *json; |
963 | enc.json = *json; |
680 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
964 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
681 | enc.cur = SvPVX (enc.sv); |
965 | enc.cur = SvPVX (enc.sv); |
682 | enc.end = SvEND (enc.sv); |
966 | enc.end = SvEND (enc.sv); |
683 | enc.indent = 0; |
967 | enc.indent = 0; |
684 | enc.maxdepth = DEC_DEPTH (enc.json.flags); |
968 | enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
|
|
969 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
|
|
970 | : 0x110000UL; |
685 | |
971 | |
686 | SvPOK_only (enc.sv); |
972 | SvPOK_only (enc.sv); |
687 | encode_sv (&enc, scalar); |
973 | encode_sv (&enc, scalar); |
|
|
974 | encode_nl (&enc); |
688 | |
975 | |
689 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
976 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
690 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
977 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
691 | |
978 | |
692 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
979 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
… | |
… | |
710 | JSON json; |
997 | JSON json; |
711 | U32 depth; // recursion depth |
998 | U32 depth; // recursion depth |
712 | U32 maxdepth; // recursion depth limit |
999 | U32 maxdepth; // recursion depth limit |
713 | } dec_t; |
1000 | } dec_t; |
714 | |
1001 | |
715 | inline void |
1002 | INLINE void |
716 | decode_comment (dec_t *dec) |
1003 | decode_comment (dec_t *dec) |
717 | { |
1004 | { |
718 | // only '#'-style comments allowed a.t.m. |
1005 | // only '#'-style comments allowed a.t.m. |
719 | |
1006 | |
720 | while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) |
1007 | while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) |
721 | ++dec->cur; |
1008 | ++dec->cur; |
722 | } |
1009 | } |
723 | |
1010 | |
724 | inline void |
1011 | INLINE void |
725 | decode_ws (dec_t *dec) |
1012 | decode_ws (dec_t *dec) |
726 | { |
1013 | { |
727 | for (;;) |
1014 | for (;;) |
728 | { |
1015 | { |
729 | char ch = *dec->cur; |
1016 | char ch = *dec->cur; |
… | |
… | |
753 | if (*dec->cur != ch) \ |
1040 | if (*dec->cur != ch) \ |
754 | ERR (# ch " expected"); \ |
1041 | ERR (# ch " expected"); \ |
755 | ++dec->cur; \ |
1042 | ++dec->cur; \ |
756 | SE |
1043 | SE |
757 | |
1044 | |
758 | #define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") |
1045 | #define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED) |
759 | #define DEC_DEC_DEPTH --dec->depth |
1046 | #define DEC_DEC_DEPTH --dec->depth |
760 | |
1047 | |
761 | static SV *decode_sv (dec_t *dec); |
1048 | static SV *decode_sv (dec_t *dec); |
762 | |
1049 | |
763 | static signed char decode_hexdigit[256]; |
1050 | static signed char decode_hexdigit[256]; |
… | |
… | |
855 | |
1142 | |
856 | if (hi >= 0x80) |
1143 | if (hi >= 0x80) |
857 | { |
1144 | { |
858 | utf8 = 1; |
1145 | utf8 = 1; |
859 | |
1146 | |
860 | cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); |
1147 | cur = encode_utf8 (cur, hi); |
861 | } |
1148 | } |
862 | else |
1149 | else |
863 | *cur++ = hi; |
1150 | *cur++ = hi; |
864 | } |
1151 | } |
865 | break; |
1152 | break; |
… | |
… | |
867 | default: |
1154 | default: |
868 | --dec_cur; |
1155 | --dec_cur; |
869 | ERR ("illegal backslash escape sequence in string"); |
1156 | ERR ("illegal backslash escape sequence in string"); |
870 | } |
1157 | } |
871 | } |
1158 | } |
872 | else if (expect_true (ch >= 0x20 && ch <= 0x7f)) |
1159 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
873 | *cur++ = ch; |
1160 | *cur++ = ch; |
874 | else if (ch >= 0x80) |
1161 | else if (ch >= 0x80) |
875 | { |
1162 | { |
876 | STRLEN clen; |
1163 | STRLEN clen; |
877 | UV uch; |
|
|
878 | |
1164 | |
879 | --dec_cur; |
1165 | --dec_cur; |
880 | |
1166 | |
881 | uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
1167 | decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
882 | if (clen == (STRLEN)-1) |
1168 | if (clen == (STRLEN)-1) |
883 | ERR ("malformed UTF-8 character in JSON string"); |
1169 | ERR ("malformed UTF-8 character in JSON string"); |
884 | |
1170 | |
885 | do |
1171 | do |
886 | *cur++ = *dec_cur++; |
1172 | *cur++ = *dec_cur++; |
887 | while (--clen); |
1173 | while (--clen); |
888 | |
1174 | |
889 | utf8 = 1; |
1175 | utf8 = 1; |
890 | } |
1176 | } |
|
|
1177 | else if (ch == '\t' && dec->json.flags & F_RELAXED) |
|
|
1178 | *cur++ = ch; |
891 | else |
1179 | else |
892 | { |
1180 | { |
893 | --dec_cur; |
1181 | --dec_cur; |
894 | |
1182 | |
895 | if (!ch) |
1183 | if (!ch) |
… | |
… | |
903 | { |
1191 | { |
904 | STRLEN len = cur - buf; |
1192 | STRLEN len = cur - buf; |
905 | |
1193 | |
906 | if (sv) |
1194 | if (sv) |
907 | { |
1195 | { |
908 | SvGROW (sv, SvCUR (sv) + len + 1); |
1196 | STRLEN cur = SvCUR (sv); |
|
|
1197 | |
|
|
1198 | if (SvLEN (sv) - cur <= len) |
|
|
1199 | json_sv_grow (sv, cur, len); |
|
|
1200 | |
909 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
1201 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
910 | SvCUR_set (sv, SvCUR (sv) + len); |
1202 | SvCUR_set (sv, SvCUR (sv) + len); |
911 | } |
1203 | } |
912 | else |
1204 | else |
913 | sv = newSVpvn (buf, len); |
1205 | sv = newSVpvn (buf, len); |
… | |
… | |
1000 | |
1292 | |
1001 | if (!is_nv) |
1293 | if (!is_nv) |
1002 | { |
1294 | { |
1003 | int len = dec->cur - start; |
1295 | int len = dec->cur - start; |
1004 | |
1296 | |
1005 | // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so |
1297 | // special case the rather common 1..5-digit-int case |
1006 | if (*start == '-') |
1298 | if (*start == '-') |
1007 | switch (len) |
1299 | switch (len) |
1008 | { |
1300 | { |
1009 | case 2: return newSViv (-( start [1] - '0' * 1)); |
1301 | case 2: return newSViv (-(IV)( start [1] - '0' * 1)); |
1010 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
1302 | case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11)); |
1011 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
1303 | case 4: return newSViv (-(IV)( 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)); |
1304 | case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
|
|
1305 | case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
1013 | } |
1306 | } |
1014 | else |
1307 | else |
1015 | switch (len) |
1308 | switch (len) |
1016 | { |
1309 | { |
1017 | case 1: return newSViv ( start [0] - '0' * 1); |
1310 | case 1: return newSViv ( start [0] - '0' * 1); |
1018 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
1311 | 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); |
1312 | 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); |
1313 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
|
|
1314 | case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
1021 | } |
1315 | } |
1022 | |
1316 | |
1023 | { |
1317 | { |
1024 | UV uv; |
1318 | UV uv; |
1025 | int numtype = grok_number (start, len, &uv); |
1319 | int numtype = grok_number (start, len, &uv); |
… | |
… | |
1034 | } |
1328 | } |
1035 | |
1329 | |
1036 | len -= *start == '-' ? 1 : 0; |
1330 | len -= *start == '-' ? 1 : 0; |
1037 | |
1331 | |
1038 | // does not fit into IV or UV, try NV |
1332 | // does not fit into IV or UV, try NV |
1039 | if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) |
1333 | if (len <= NV_DIG) |
1040 | #if defined (LDBL_DIG) |
|
|
1041 | || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len) |
|
|
1042 | #endif |
|
|
1043 | ) |
|
|
1044 | // fits into NV without loss of precision |
1334 | // fits into NV without loss of precision |
1045 | return newSVnv (Atof (start)); |
1335 | return newSVnv (json_atof (start)); |
1046 | |
1336 | |
1047 | // everything else fails, convert it to a string |
1337 | // everything else fails, convert it to a string |
1048 | return newSVpvn (start, dec->cur - start); |
1338 | return newSVpvn (start, dec->cur - start); |
1049 | } |
1339 | } |
1050 | |
1340 | |
1051 | // loss of precision here |
1341 | // loss of precision here |
1052 | return newSVnv (Atof (start)); |
1342 | return newSVnv (json_atof (start)); |
1053 | |
1343 | |
1054 | fail: |
1344 | fail: |
1055 | return 0; |
1345 | return 0; |
1056 | } |
1346 | } |
1057 | |
1347 | |
… | |
… | |
1081 | if (*dec->cur == ']') |
1371 | if (*dec->cur == ']') |
1082 | { |
1372 | { |
1083 | ++dec->cur; |
1373 | ++dec->cur; |
1084 | break; |
1374 | break; |
1085 | } |
1375 | } |
1086 | |
1376 | |
1087 | if (*dec->cur != ',') |
1377 | if (*dec->cur != ',') |
1088 | ERR (", or ] expected while parsing array"); |
1378 | ERR (", or ] expected while parsing array"); |
1089 | |
1379 | |
1090 | ++dec->cur; |
1380 | ++dec->cur; |
1091 | |
1381 | |
… | |
… | |
1133 | char *p = dec->cur; |
1423 | char *p = dec->cur; |
1134 | char *e = p + 24; // only try up to 24 bytes |
1424 | char *e = p + 24; // only try up to 24 bytes |
1135 | |
1425 | |
1136 | for (;;) |
1426 | for (;;) |
1137 | { |
1427 | { |
1138 | // the >= 0x80 is true on most architectures |
1428 | // the >= 0x80 is false on most architectures |
1139 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1429 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1140 | { |
1430 | { |
1141 | // slow path, back up and use decode_str |
1431 | // slow path, back up and use decode_str |
1142 | SV *key = decode_str (dec); |
1432 | SV *key = decode_str (dec); |
1143 | if (!key) |
1433 | if (!key) |
… | |
… | |
1215 | |
1505 | |
1216 | hv_iterinit (hv); |
1506 | hv_iterinit (hv); |
1217 | he = hv_iternext (hv); |
1507 | he = hv_iternext (hv); |
1218 | hv_iterinit (hv); |
1508 | hv_iterinit (hv); |
1219 | |
1509 | |
1220 | // the next line creates a mortal sv each time its called. |
1510 | // the next line creates a mortal sv each time it's called. |
1221 | // might want to optimise this for common cases. |
1511 | // might want to optimise this for common cases. |
1222 | cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); |
1512 | cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); |
1223 | |
1513 | |
1224 | if (cb) |
1514 | if (cb) |
1225 | { |
1515 | { |
1226 | dSP; |
1516 | dSP; |
1227 | int count; |
1517 | int count; |
1228 | |
1518 | |
1229 | ENTER; SAVETMPS; PUSHMARK (SP); |
1519 | ENTER; SAVETMPS; |
|
|
1520 | SAVESTACK_POS (); |
|
|
1521 | PUSHMARK (SP); |
1230 | XPUSHs (HeVAL (he)); |
1522 | XPUSHs (HeVAL (he)); |
|
|
1523 | sv_2mortal (sv); |
1231 | |
1524 | |
1232 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1525 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1233 | |
1526 | |
1234 | if (count == 1) |
1527 | if (count == 1) |
1235 | { |
1528 | { |
1236 | sv = newSVsv (POPs); |
1529 | sv = newSVsv (POPs); |
1237 | FREETMPS; LEAVE; |
1530 | FREETMPS; LEAVE; |
1238 | return sv; |
1531 | return sv; |
1239 | } |
1532 | } |
1240 | |
1533 | |
|
|
1534 | SvREFCNT_inc (sv); |
1241 | FREETMPS; LEAVE; |
1535 | FREETMPS; LEAVE; |
1242 | } |
1536 | } |
1243 | } |
1537 | } |
1244 | |
1538 | |
1245 | if (dec->json.cb_object) |
1539 | if (dec->json.cb_object) |
1246 | { |
1540 | { |
1247 | dSP; |
1541 | dSP; |
1248 | int count; |
1542 | int count; |
1249 | |
1543 | |
1250 | ENTER; SAVETMPS; PUSHMARK (SP); |
1544 | ENTER; SAVETMPS; |
|
|
1545 | SAVESTACK_POS (); |
|
|
1546 | PUSHMARK (SP); |
1251 | XPUSHs (sv_2mortal (sv)); |
1547 | XPUSHs (sv_2mortal (sv)); |
1252 | |
1548 | |
1253 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
1549 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
1254 | |
1550 | |
1255 | if (count == 1) |
1551 | if (count == 1) |
… | |
… | |
1271 | DEC_DEC_DEPTH; |
1567 | DEC_DEC_DEPTH; |
1272 | return 0; |
1568 | return 0; |
1273 | } |
1569 | } |
1274 | |
1570 | |
1275 | static SV * |
1571 | static SV * |
|
|
1572 | decode_tag (dec_t *dec) |
|
|
1573 | { |
|
|
1574 | SV *tag = 0; |
|
|
1575 | SV *val = 0; |
|
|
1576 | |
|
|
1577 | if (!(dec->json.flags & F_ALLOW_TAGS)) |
|
|
1578 | ERR ("malformed JSON string, neither array, object, number, string or atom"); |
|
|
1579 | |
|
|
1580 | ++dec->cur; |
|
|
1581 | |
|
|
1582 | decode_ws (dec); |
|
|
1583 | |
|
|
1584 | tag = decode_sv (dec); |
|
|
1585 | if (!tag) |
|
|
1586 | goto fail; |
|
|
1587 | |
|
|
1588 | if (!SvPOK (tag)) |
|
|
1589 | ERR ("malformed JSON string, (tag) must be a string"); |
|
|
1590 | |
|
|
1591 | decode_ws (dec); |
|
|
1592 | |
|
|
1593 | if (*dec->cur != ')') |
|
|
1594 | ERR (") expected after tag"); |
|
|
1595 | |
|
|
1596 | ++dec->cur; |
|
|
1597 | |
|
|
1598 | decode_ws (dec); |
|
|
1599 | |
|
|
1600 | val = decode_sv (dec); |
|
|
1601 | if (!val) |
|
|
1602 | goto fail; |
|
|
1603 | |
|
|
1604 | if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV) |
|
|
1605 | ERR ("malformed JSON string, tag value must be an array"); |
|
|
1606 | |
|
|
1607 | { |
|
|
1608 | AV *av = (AV *)SvRV (val); |
|
|
1609 | int i, len = av_len (av) + 1; |
|
|
1610 | HV *stash = gv_stashsv (tag, 0); |
|
|
1611 | SV *sv; |
|
|
1612 | |
|
|
1613 | if (!stash) |
|
|
1614 | ERR ("cannot decode perl-object (package does not exist)"); |
|
|
1615 | |
|
|
1616 | GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); |
|
|
1617 | |
|
|
1618 | if (!method) |
|
|
1619 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
|
|
1620 | |
|
|
1621 | dSP; |
|
|
1622 | |
|
|
1623 | ENTER; SAVETMPS; |
|
|
1624 | PUSHMARK (SP); |
|
|
1625 | EXTEND (SP, len + 2); |
|
|
1626 | // we re-bless the reference to get overload and other niceties right |
|
|
1627 | PUSHs (tag); |
|
|
1628 | PUSHs (sv_json); |
|
|
1629 | |
|
|
1630 | for (i = 0; i < len; ++i) |
|
|
1631 | PUSHs (*av_fetch (av, i, 1)); |
|
|
1632 | |
|
|
1633 | PUTBACK; |
|
|
1634 | call_sv ((SV *)GvCV (method), G_SCALAR); |
|
|
1635 | SPAGAIN; |
|
|
1636 | |
|
|
1637 | SvREFCNT_dec (tag); |
|
|
1638 | SvREFCNT_dec (val); |
|
|
1639 | sv = SvREFCNT_inc (POPs); |
|
|
1640 | |
|
|
1641 | PUTBACK; |
|
|
1642 | |
|
|
1643 | FREETMPS; LEAVE; |
|
|
1644 | |
|
|
1645 | return sv; |
|
|
1646 | } |
|
|
1647 | |
|
|
1648 | fail: |
|
|
1649 | SvREFCNT_dec (tag); |
|
|
1650 | SvREFCNT_dec (val); |
|
|
1651 | return 0; |
|
|
1652 | } |
|
|
1653 | |
|
|
1654 | static SV * |
1276 | decode_sv (dec_t *dec) |
1655 | decode_sv (dec_t *dec) |
1277 | { |
1656 | { |
1278 | // the beauty of JSON: you need exactly one character lookahead |
1657 | // the beauty of JSON: you need exactly one character lookahead |
1279 | // to parse anything. |
1658 | // to parse everything. |
1280 | switch (*dec->cur) |
1659 | switch (*dec->cur) |
1281 | { |
1660 | { |
1282 | case '"': ++dec->cur; return decode_str (dec); |
1661 | case '"': ++dec->cur; return decode_str (dec); |
1283 | case '[': ++dec->cur; return decode_av (dec); |
1662 | case '[': ++dec->cur; return decode_av (dec); |
1284 | case '{': ++dec->cur; return decode_hv (dec); |
1663 | case '{': ++dec->cur; return decode_hv (dec); |
|
|
1664 | case '(': return decode_tag (dec); |
1285 | |
1665 | |
1286 | case '-': |
1666 | case '-': |
1287 | case '0': case '1': case '2': case '3': case '4': |
1667 | case '0': case '1': case '2': case '3': case '4': |
1288 | case '5': case '6': case '7': case '8': case '9': |
1668 | case '5': case '6': case '7': case '8': case '9': |
1289 | return decode_num (dec); |
1669 | return decode_num (dec); |
… | |
… | |
1291 | case 't': |
1671 | case 't': |
1292 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1672 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1293 | { |
1673 | { |
1294 | dec->cur += 4; |
1674 | dec->cur += 4; |
1295 | #if JSON_SLOW |
1675 | #if JSON_SLOW |
1296 | json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); |
1676 | bool_true = get_bool ("Types::Serialiser::true"); |
1297 | #endif |
1677 | #endif |
1298 | return SvREFCNT_inc (json_true); |
1678 | return newSVsv (bool_true); |
1299 | } |
1679 | } |
1300 | else |
1680 | else |
1301 | ERR ("'true' expected"); |
1681 | ERR ("'true' expected"); |
1302 | |
1682 | |
1303 | break; |
1683 | break; |
… | |
… | |
1305 | case 'f': |
1685 | case 'f': |
1306 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1686 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1307 | { |
1687 | { |
1308 | dec->cur += 5; |
1688 | dec->cur += 5; |
1309 | #if JSON_SLOW |
1689 | #if JSON_SLOW |
1310 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1690 | bool_false = get_bool ("Types::Serialiser::false"); |
1311 | #endif |
1691 | #endif |
1312 | return SvREFCNT_inc (json_false); |
1692 | return newSVsv (bool_false); |
1313 | } |
1693 | } |
1314 | else |
1694 | else |
1315 | ERR ("'false' expected"); |
1695 | ERR ("'false' expected"); |
1316 | |
1696 | |
1317 | break; |
1697 | break; |
… | |
… | |
1326 | ERR ("'null' expected"); |
1706 | ERR ("'null' expected"); |
1327 | |
1707 | |
1328 | break; |
1708 | break; |
1329 | |
1709 | |
1330 | default: |
1710 | default: |
1331 | ERR ("malformed JSON string, neither array, object, number, string or atom"); |
1711 | ERR ("malformed JSON string, neither tag, array, object, number, string or atom"); |
1332 | break; |
1712 | break; |
1333 | } |
1713 | } |
1334 | |
1714 | |
1335 | fail: |
1715 | fail: |
1336 | return 0; |
1716 | return 0; |
1337 | } |
1717 | } |
1338 | |
1718 | |
1339 | static SV * |
1719 | static SV * |
1340 | decode_json (SV *string, JSON *json, UV *offset_return) |
1720 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1341 | { |
1721 | { |
1342 | dec_t dec; |
1722 | dec_t dec; |
1343 | UV offset; |
|
|
1344 | SV *sv; |
1723 | SV *sv; |
1345 | |
1724 | |
|
|
1725 | /* work around bugs in 5.10 where manipulating magic values |
|
|
1726 | * makes perl ignore the magic in subsequent accesses. |
|
|
1727 | * also make a copy of non-PV values, to get them into a clean |
|
|
1728 | * state (SvPV should do that, but it's buggy, see below). |
|
|
1729 | * |
|
|
1730 | * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16), |
|
|
1731 | * as reported by Reini Urban. |
|
|
1732 | */ |
1346 | SvGETMAGIC (string); |
1733 | /*SvGETMAGIC (string);*/ |
|
|
1734 | if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string)) |
|
|
1735 | string = sv_2mortal (newSVsv (string)); |
|
|
1736 | |
1347 | SvUPGRADE (string, SVt_PV); |
1737 | SvUPGRADE (string, SVt_PV); |
1348 | |
1738 | |
1349 | if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) |
1739 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
|
|
1740 | * assertion with -DDEBUGGING, although SvCUR is documented to |
|
|
1741 | * return the xpv_cur field which certainly exists after upgrading. |
|
|
1742 | * according to nicholas clark, calling SvPOK fixes this. |
|
|
1743 | * But it doesn't fix it, so try another workaround, call SvPV_nolen |
|
|
1744 | * and hope for the best. |
|
|
1745 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
|
|
1746 | * assertion business is seriously broken, try yet another workaround |
|
|
1747 | * for the broken -DDEBUGGING. |
|
|
1748 | */ |
|
|
1749 | { |
|
|
1750 | #ifdef DEBUGGING |
|
|
1751 | STRLEN offset = SvOK (string) ? sv_len (string) : 0; |
|
|
1752 | #else |
|
|
1753 | STRLEN offset = SvCUR (string); |
|
|
1754 | #endif |
|
|
1755 | |
|
|
1756 | if (offset > json->max_size && json->max_size) |
1350 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1757 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
1351 | (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); |
1758 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
|
|
1759 | } |
1352 | |
1760 | |
1353 | if (json->flags & F_UTF8) |
1761 | if (DECODE_WANTS_OCTETS (json)) |
1354 | sv_utf8_downgrade (string, 0); |
1762 | sv_utf8_downgrade (string, 0); |
1355 | else |
1763 | else |
1356 | sv_utf8_upgrade (string); |
1764 | sv_utf8_upgrade (string); |
1357 | |
1765 | |
1358 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1766 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1359 | |
1767 | |
1360 | dec.json = *json; |
1768 | dec.json = *json; |
1361 | dec.cur = SvPVX (string); |
1769 | dec.cur = SvPVX (string); |
1362 | dec.end = SvEND (string); |
1770 | dec.end = SvEND (string); |
1363 | dec.err = 0; |
1771 | dec.err = 0; |
1364 | dec.depth = 0; |
1772 | dec.depth = 0; |
1365 | dec.maxdepth = DEC_DEPTH (dec.json.flags); |
|
|
1366 | |
1773 | |
1367 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1774 | if (dec.json.cb_object || dec.json.cb_sk_object) |
1368 | dec.json.flags |= F_HOOK; |
1775 | dec.json.flags |= F_HOOK; |
1369 | |
1776 | |
1370 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1777 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
1371 | |
1778 | |
1372 | decode_ws (&dec); |
1779 | decode_ws (&dec); |
1373 | sv = decode_sv (&dec); |
1780 | sv = decode_sv (&dec); |
1374 | |
1781 | |
1375 | if (!(offset_return || !sv)) |
1782 | if (offset_return) |
|
|
1783 | *offset_return = dec.cur - SvPVX (string); |
|
|
1784 | else if (sv) |
1376 | { |
1785 | { |
1377 | // check for trailing garbage |
1786 | // check for trailing garbage |
1378 | decode_ws (&dec); |
1787 | decode_ws (&dec); |
1379 | |
1788 | |
1380 | if (*dec.cur) |
1789 | if (*dec.cur) |
1381 | { |
1790 | { |
1382 | dec.err = "garbage after JSON object"; |
1791 | dec.err = "garbage after JSON object"; |
1383 | SvREFCNT_dec (sv); |
1792 | SvREFCNT_dec (sv); |
1384 | sv = 0; |
1793 | sv = 0; |
1385 | } |
1794 | } |
1386 | } |
|
|
1387 | |
|
|
1388 | if (offset_return || !sv) |
|
|
1389 | { |
|
|
1390 | offset = dec.json.flags & F_UTF8 |
|
|
1391 | ? dec.cur - SvPVX (string) |
|
|
1392 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
1393 | |
|
|
1394 | if (offset_return) |
|
|
1395 | *offset_return = offset; |
|
|
1396 | } |
1795 | } |
1397 | |
1796 | |
1398 | if (!sv) |
1797 | if (!sv) |
1399 | { |
1798 | { |
1400 | SV *uni = sv_newmortal (); |
1799 | SV *uni = sv_newmortal (); |
… | |
… | |
1406 | SAVEVPTR (PL_curcop); |
1805 | SAVEVPTR (PL_curcop); |
1407 | PL_curcop = &cop; |
1806 | PL_curcop = &cop; |
1408 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1807 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1409 | LEAVE; |
1808 | LEAVE; |
1410 | |
1809 | |
1411 | croak ("%s, at character offset %d [\"%s\"]", |
1810 | croak ("%s, at character offset %d (before \"%s\")", |
1412 | dec.err, |
1811 | dec.err, |
1413 | (int)offset, |
1812 | (int)ptr_to_index (string, dec.cur), |
1414 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1813 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1415 | } |
1814 | } |
1416 | |
1815 | |
1417 | sv = sv_2mortal (sv); |
1816 | sv = sv_2mortal (sv); |
1418 | |
1817 | |
1419 | if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1818 | if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv)) |
1420 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1819 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1421 | |
1820 | |
1422 | return sv; |
1821 | return sv; |
|
|
1822 | } |
|
|
1823 | |
|
|
1824 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1825 | // incremental parser |
|
|
1826 | |
|
|
1827 | static void |
|
|
1828 | incr_parse (JSON *self) |
|
|
1829 | { |
|
|
1830 | const char *p = SvPVX (self->incr_text) + self->incr_pos; |
|
|
1831 | |
|
|
1832 | // the state machine here is a bit convoluted and could be simplified a lot |
|
|
1833 | // but this would make it slower, so... |
|
|
1834 | |
|
|
1835 | for (;;) |
|
|
1836 | { |
|
|
1837 | //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 |
|
|
1838 | switch (self->incr_mode) |
|
|
1839 | { |
|
|
1840 | // only used for initial whitespace skipping |
|
|
1841 | case INCR_M_WS: |
|
|
1842 | for (;;) |
|
|
1843 | { |
|
|
1844 | if (*p > 0x20) |
|
|
1845 | { |
|
|
1846 | if (*p == '#') |
|
|
1847 | { |
|
|
1848 | self->incr_mode = INCR_M_C0; |
|
|
1849 | goto incr_m_c; |
|
|
1850 | } |
|
|
1851 | else |
|
|
1852 | { |
|
|
1853 | self->incr_mode = INCR_M_JSON; |
|
|
1854 | goto incr_m_json; |
|
|
1855 | } |
|
|
1856 | } |
|
|
1857 | else if (!*p) |
|
|
1858 | goto interrupt; |
|
|
1859 | |
|
|
1860 | ++p; |
|
|
1861 | } |
|
|
1862 | |
|
|
1863 | // skip a single char inside a string (for \\-processing) |
|
|
1864 | case INCR_M_BS: |
|
|
1865 | if (!*p) |
|
|
1866 | goto interrupt; |
|
|
1867 | |
|
|
1868 | ++p; |
|
|
1869 | self->incr_mode = INCR_M_STR; |
|
|
1870 | goto incr_m_str; |
|
|
1871 | |
|
|
1872 | // inside #-style comments |
|
|
1873 | case INCR_M_C0: |
|
|
1874 | case INCR_M_C1: |
|
|
1875 | incr_m_c: |
|
|
1876 | for (;;) |
|
|
1877 | { |
|
|
1878 | if (*p == '\n') |
|
|
1879 | { |
|
|
1880 | self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON; |
|
|
1881 | break; |
|
|
1882 | } |
|
|
1883 | else if (!*p) |
|
|
1884 | goto interrupt; |
|
|
1885 | |
|
|
1886 | ++p; |
|
|
1887 | } |
|
|
1888 | |
|
|
1889 | break; |
|
|
1890 | |
|
|
1891 | // inside a string |
|
|
1892 | case INCR_M_STR: |
|
|
1893 | incr_m_str: |
|
|
1894 | for (;;) |
|
|
1895 | { |
|
|
1896 | if (*p == '"') |
|
|
1897 | { |
|
|
1898 | ++p; |
|
|
1899 | self->incr_mode = INCR_M_JSON; |
|
|
1900 | |
|
|
1901 | if (!self->incr_nest) |
|
|
1902 | goto interrupt; |
|
|
1903 | |
|
|
1904 | goto incr_m_json; |
|
|
1905 | } |
|
|
1906 | else if (*p == '\\') |
|
|
1907 | { |
|
|
1908 | ++p; // "virtually" consumes character after \ |
|
|
1909 | |
|
|
1910 | if (!*p) // if at end of string we have to switch modes |
|
|
1911 | { |
|
|
1912 | self->incr_mode = INCR_M_BS; |
|
|
1913 | goto interrupt; |
|
|
1914 | } |
|
|
1915 | } |
|
|
1916 | else if (!*p) |
|
|
1917 | goto interrupt; |
|
|
1918 | |
|
|
1919 | ++p; |
|
|
1920 | } |
|
|
1921 | |
|
|
1922 | // after initial ws, outside string |
|
|
1923 | case INCR_M_JSON: |
|
|
1924 | incr_m_json: |
|
|
1925 | for (;;) |
|
|
1926 | { |
|
|
1927 | switch (*p++) |
|
|
1928 | { |
|
|
1929 | case 0: |
|
|
1930 | --p; |
|
|
1931 | goto interrupt; |
|
|
1932 | |
|
|
1933 | case 0x09: |
|
|
1934 | case 0x0a: |
|
|
1935 | case 0x0d: |
|
|
1936 | case 0x20: |
|
|
1937 | if (!self->incr_nest) |
|
|
1938 | { |
|
|
1939 | --p; // do not eat the whitespace, let the next round do it |
|
|
1940 | goto interrupt; |
|
|
1941 | } |
|
|
1942 | break; |
|
|
1943 | |
|
|
1944 | case '"': |
|
|
1945 | self->incr_mode = INCR_M_STR; |
|
|
1946 | goto incr_m_str; |
|
|
1947 | |
|
|
1948 | case '[': |
|
|
1949 | case '{': |
|
|
1950 | case '(': |
|
|
1951 | if (++self->incr_nest > self->max_depth) |
|
|
1952 | croak (ERR_NESTING_EXCEEDED); |
|
|
1953 | break; |
|
|
1954 | |
|
|
1955 | case ']': |
|
|
1956 | case '}': |
|
|
1957 | if (--self->incr_nest <= 0) |
|
|
1958 | goto interrupt; |
|
|
1959 | break; |
|
|
1960 | |
|
|
1961 | case ')': |
|
|
1962 | --self->incr_nest; |
|
|
1963 | break; |
|
|
1964 | |
|
|
1965 | case '#': |
|
|
1966 | self->incr_mode = INCR_M_C1; |
|
|
1967 | goto incr_m_c; |
|
|
1968 | } |
|
|
1969 | } |
|
|
1970 | } |
|
|
1971 | |
|
|
1972 | modechange: |
|
|
1973 | ; |
|
|
1974 | } |
|
|
1975 | |
|
|
1976 | interrupt: |
|
|
1977 | self->incr_pos = p - SvPVX (self->incr_text); |
|
|
1978 | //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D |
|
|
1979 | //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D |
1423 | } |
1980 | } |
1424 | |
1981 | |
1425 | ///////////////////////////////////////////////////////////////////////////// |
1982 | ///////////////////////////////////////////////////////////////////////////// |
1426 | // XS interface functions |
1983 | // XS interface functions |
1427 | |
1984 | |
… | |
… | |
1436 | i >= '0' && i <= '9' ? i - '0' |
1993 | i >= '0' && i <= '9' ? i - '0' |
1437 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1994 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1438 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1995 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1439 | : -1; |
1996 | : -1; |
1440 | |
1997 | |
1441 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1998 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1442 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
1999 | bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); |
|
|
2000 | bool_true = get_bool ("Types::Serialiser::true"); |
|
|
2001 | bool_false = get_bool ("Types::Serialiser::false"); |
1443 | |
2002 | |
1444 | json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); |
2003 | sv_json = newSVpv ("JSON", 0); |
1445 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
2004 | SvREADONLY_on (sv_json); |
|
|
2005 | |
|
|
2006 | CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ |
1446 | } |
2007 | } |
1447 | |
2008 | |
1448 | PROTOTYPES: DISABLE |
2009 | PROTOTYPES: DISABLE |
1449 | |
2010 | |
1450 | void CLONE (...) |
2011 | void CLONE (...) |
1451 | CODE: |
2012 | CODE: |
1452 | json_stash = 0; |
2013 | json_stash = 0; |
1453 | json_boolean_stash = 0; |
2014 | bool_stash = 0; |
1454 | |
2015 | |
1455 | void new (char *klass) |
2016 | void new (char *klass) |
1456 | PPCODE: |
2017 | PPCODE: |
1457 | { |
2018 | { |
1458 | SV *pv = NEWSV (0, sizeof (JSON)); |
2019 | SV *pv = NEWSV (0, sizeof (JSON)); |
1459 | SvPOK_only (pv); |
2020 | SvPOK_only (pv); |
1460 | Zero (SvPVX (pv), 1, JSON); |
2021 | json_init ((JSON *)SvPVX (pv)); |
1461 | ((JSON *)SvPVX (pv))->flags = F_DEFAULT; |
|
|
1462 | XPUSHs (sv_2mortal (sv_bless ( |
2022 | XPUSHs (sv_2mortal (sv_bless ( |
1463 | newRV_noinc (pv), |
2023 | newRV_noinc (pv), |
1464 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
2024 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
1465 | ))); |
2025 | ))); |
1466 | } |
2026 | } |
… | |
… | |
1478 | allow_nonref = F_ALLOW_NONREF |
2038 | allow_nonref = F_ALLOW_NONREF |
1479 | shrink = F_SHRINK |
2039 | shrink = F_SHRINK |
1480 | allow_blessed = F_ALLOW_BLESSED |
2040 | allow_blessed = F_ALLOW_BLESSED |
1481 | convert_blessed = F_CONV_BLESSED |
2041 | convert_blessed = F_CONV_BLESSED |
1482 | relaxed = F_RELAXED |
2042 | relaxed = F_RELAXED |
|
|
2043 | allow_unknown = F_ALLOW_UNKNOWN |
|
|
2044 | allow_tags = F_ALLOW_TAGS |
1483 | PPCODE: |
2045 | PPCODE: |
1484 | { |
2046 | { |
1485 | if (enable) |
2047 | if (enable) |
1486 | self->flags |= ix; |
2048 | self->flags |= ix; |
1487 | else |
2049 | else |
… | |
… | |
1502 | get_allow_nonref = F_ALLOW_NONREF |
2064 | get_allow_nonref = F_ALLOW_NONREF |
1503 | get_shrink = F_SHRINK |
2065 | get_shrink = F_SHRINK |
1504 | get_allow_blessed = F_ALLOW_BLESSED |
2066 | get_allow_blessed = F_ALLOW_BLESSED |
1505 | get_convert_blessed = F_CONV_BLESSED |
2067 | get_convert_blessed = F_CONV_BLESSED |
1506 | get_relaxed = F_RELAXED |
2068 | get_relaxed = F_RELAXED |
|
|
2069 | get_allow_unknown = F_ALLOW_UNKNOWN |
|
|
2070 | get_allow_tags = F_ALLOW_TAGS |
1507 | PPCODE: |
2071 | PPCODE: |
1508 | XPUSHs (boolSV (self->flags & ix)); |
2072 | XPUSHs (boolSV (self->flags & ix)); |
1509 | |
2073 | |
1510 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
2074 | void max_depth (JSON *self, U32 max_depth = 0x80000000UL) |
1511 | PPCODE: |
2075 | PPCODE: |
1512 | { |
2076 | self->max_depth = max_depth; |
1513 | UV log2 = 0; |
|
|
1514 | |
|
|
1515 | if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
|
|
1516 | |
|
|
1517 | while ((1UL << log2) < max_depth) |
|
|
1518 | ++log2; |
|
|
1519 | |
|
|
1520 | self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
|
|
1521 | |
|
|
1522 | XPUSHs (ST (0)); |
2077 | XPUSHs (ST (0)); |
1523 | } |
|
|
1524 | |
2078 | |
1525 | U32 get_max_depth (JSON *self) |
2079 | U32 get_max_depth (JSON *self) |
1526 | CODE: |
2080 | CODE: |
1527 | RETVAL = DEC_DEPTH (self->flags); |
2081 | RETVAL = self->max_depth; |
1528 | OUTPUT: |
2082 | OUTPUT: |
1529 | RETVAL |
2083 | RETVAL |
1530 | |
2084 | |
1531 | void max_size (JSON *self, UV max_size = 0) |
2085 | void max_size (JSON *self, U32 max_size = 0) |
1532 | PPCODE: |
2086 | PPCODE: |
1533 | { |
2087 | self->max_size = max_size; |
1534 | UV log2 = 0; |
|
|
1535 | |
|
|
1536 | if (max_size > 0x80000000UL) max_size = 0x80000000UL; |
|
|
1537 | if (max_size == 1) max_size = 2; |
|
|
1538 | |
|
|
1539 | while ((1UL << log2) < max_size) |
|
|
1540 | ++log2; |
|
|
1541 | |
|
|
1542 | self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); |
|
|
1543 | |
|
|
1544 | XPUSHs (ST (0)); |
2088 | XPUSHs (ST (0)); |
1545 | } |
|
|
1546 | |
2089 | |
1547 | int get_max_size (JSON *self) |
2090 | int get_max_size (JSON *self) |
1548 | CODE: |
2091 | CODE: |
1549 | RETVAL = DEC_SIZE (self->flags); |
2092 | RETVAL = self->max_size; |
1550 | OUTPUT: |
2093 | OUTPUT: |
1551 | RETVAL |
2094 | RETVAL |
1552 | |
2095 | |
1553 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
2096 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1554 | PPCODE: |
2097 | PPCODE: |
… | |
… | |
1560 | } |
2103 | } |
1561 | |
2104 | |
1562 | void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) |
2105 | void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) |
1563 | PPCODE: |
2106 | PPCODE: |
1564 | { |
2107 | { |
1565 | if (!self->cb_sk_object) |
2108 | if (!self->cb_sk_object) |
1566 | self->cb_sk_object = newHV (); |
2109 | self->cb_sk_object = newHV (); |
1567 | |
2110 | |
1568 | if (SvOK (cb)) |
2111 | if (SvOK (cb)) |
1569 | hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
2112 | hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
1570 | else |
2113 | else |
… | |
… | |
1581 | XPUSHs (ST (0)); |
2124 | XPUSHs (ST (0)); |
1582 | } |
2125 | } |
1583 | |
2126 | |
1584 | void encode (JSON *self, SV *scalar) |
2127 | void encode (JSON *self, SV *scalar) |
1585 | PPCODE: |
2128 | PPCODE: |
1586 | XPUSHs (encode_json (scalar, self)); |
2129 | PUTBACK; scalar = encode_json (scalar, self); SPAGAIN; |
|
|
2130 | XPUSHs (scalar); |
1587 | |
2131 | |
1588 | void decode (JSON *self, SV *jsonstr) |
2132 | void decode (JSON *self, SV *jsonstr) |
1589 | PPCODE: |
2133 | PPCODE: |
1590 | XPUSHs (decode_json (jsonstr, self, 0)); |
2134 | PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN; |
|
|
2135 | XPUSHs (jsonstr); |
1591 | |
2136 | |
1592 | void decode_prefix (JSON *self, SV *jsonstr) |
2137 | void decode_prefix (JSON *self, SV *jsonstr) |
1593 | PPCODE: |
2138 | PPCODE: |
1594 | { |
2139 | { |
1595 | UV offset; |
2140 | SV *sv; |
|
|
2141 | STRLEN offset; |
|
|
2142 | PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; |
1596 | EXTEND (SP, 2); |
2143 | EXTEND (SP, 2); |
1597 | PUSHs (decode_json (jsonstr, self, &offset)); |
2144 | PUSHs (sv); |
1598 | PUSHs (sv_2mortal (newSVuv (offset))); |
2145 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset)))); |
|
|
2146 | } |
|
|
2147 | |
|
|
2148 | void incr_parse (JSON *self, SV *jsonstr = 0) |
|
|
2149 | PPCODE: |
|
|
2150 | { |
|
|
2151 | if (!self->incr_text) |
|
|
2152 | self->incr_text = newSVpvn ("", 0); |
|
|
2153 | |
|
|
2154 | /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */ |
|
|
2155 | if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text)) |
|
|
2156 | if (DECODE_WANTS_OCTETS (self)) |
|
|
2157 | { |
|
|
2158 | if (self->incr_pos) |
|
|
2159 | self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text), |
|
|
2160 | (U8 *)SvPVX (self->incr_text) + self->incr_pos); |
|
|
2161 | |
|
|
2162 | sv_utf8_downgrade (self->incr_text, 0); |
|
|
2163 | } |
|
|
2164 | else |
|
|
2165 | { |
|
|
2166 | sv_utf8_upgrade (self->incr_text); |
|
|
2167 | |
|
|
2168 | if (self->incr_pos) |
|
|
2169 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
|
|
2170 | - (U8 *)SvPVX (self->incr_text); |
|
|
2171 | } |
|
|
2172 | |
|
|
2173 | // append data, if any |
|
|
2174 | if (jsonstr) |
|
|
2175 | { |
|
|
2176 | /* make sure both strings have same encoding */ |
|
|
2177 | if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text)) |
|
|
2178 | if (SvUTF8 (jsonstr)) |
|
|
2179 | sv_utf8_downgrade (jsonstr, 0); |
|
|
2180 | else |
|
|
2181 | sv_utf8_upgrade (jsonstr); |
|
|
2182 | |
|
|
2183 | /* and then just blindly append */ |
|
|
2184 | { |
|
|
2185 | STRLEN len; |
|
|
2186 | const char *str = SvPV (jsonstr, len); |
|
|
2187 | STRLEN cur = SvCUR (self->incr_text); |
|
|
2188 | |
|
|
2189 | if (SvLEN (self->incr_text) - cur <= len) |
|
|
2190 | json_sv_grow (self->incr_text, cur, len); |
|
|
2191 | |
|
|
2192 | Move (str, SvEND (self->incr_text), len, char); |
|
|
2193 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
|
|
2194 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
|
|
2195 | } |
|
|
2196 | } |
|
|
2197 | |
|
|
2198 | if (GIMME_V != G_VOID) |
|
|
2199 | do |
|
|
2200 | { |
|
|
2201 | SV *sv; |
|
|
2202 | STRLEN offset; |
|
|
2203 | |
|
|
2204 | if (!INCR_DONE (self)) |
|
|
2205 | { |
|
|
2206 | incr_parse (self); |
|
|
2207 | |
|
|
2208 | if (self->incr_pos > self->max_size && self->max_size) |
|
|
2209 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
|
|
2210 | (unsigned long)self->incr_pos, (unsigned long)self->max_size); |
|
|
2211 | |
|
|
2212 | if (!INCR_DONE (self)) |
|
|
2213 | { |
|
|
2214 | // as an optimisation, do not accumulate white space in the incr buffer |
|
|
2215 | if (self->incr_mode == INCR_M_WS && self->incr_pos) |
|
|
2216 | { |
|
|
2217 | self->incr_pos = 0; |
|
|
2218 | SvCUR_set (self->incr_text, 0); |
|
|
2219 | } |
|
|
2220 | |
|
|
2221 | break; |
|
|
2222 | } |
|
|
2223 | } |
|
|
2224 | |
|
|
2225 | PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; |
|
|
2226 | XPUSHs (sv); |
|
|
2227 | |
|
|
2228 | self->incr_pos -= offset; |
|
|
2229 | self->incr_nest = 0; |
|
|
2230 | self->incr_mode = 0; |
|
|
2231 | |
|
|
2232 | sv_chop (self->incr_text, SvPVX (self->incr_text) + offset); |
|
|
2233 | } |
|
|
2234 | while (GIMME_V == G_ARRAY); |
|
|
2235 | } |
|
|
2236 | |
|
|
2237 | SV *incr_text (JSON *self) |
|
|
2238 | ATTRS: lvalue |
|
|
2239 | CODE: |
|
|
2240 | { |
|
|
2241 | if (self->incr_pos) |
|
|
2242 | croak ("incr_text can not be called when the incremental parser already started parsing"); |
|
|
2243 | |
|
|
2244 | RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef; |
|
|
2245 | } |
|
|
2246 | OUTPUT: |
|
|
2247 | RETVAL |
|
|
2248 | |
|
|
2249 | void incr_skip (JSON *self) |
|
|
2250 | CODE: |
|
|
2251 | { |
|
|
2252 | if (self->incr_pos) |
|
|
2253 | { |
|
|
2254 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos); |
|
|
2255 | self->incr_pos = 0; |
|
|
2256 | self->incr_nest = 0; |
|
|
2257 | self->incr_mode = 0; |
|
|
2258 | } |
|
|
2259 | } |
|
|
2260 | |
|
|
2261 | void incr_reset (JSON *self) |
|
|
2262 | CODE: |
|
|
2263 | { |
|
|
2264 | SvREFCNT_dec (self->incr_text); |
|
|
2265 | self->incr_text = 0; |
|
|
2266 | self->incr_pos = 0; |
|
|
2267 | self->incr_nest = 0; |
|
|
2268 | self->incr_mode = 0; |
1599 | } |
2269 | } |
1600 | |
2270 | |
1601 | void DESTROY (JSON *self) |
2271 | void DESTROY (JSON *self) |
1602 | CODE: |
2272 | CODE: |
1603 | SvREFCNT_dec (self->cb_sk_object); |
2273 | SvREFCNT_dec (self->cb_sk_object); |
1604 | SvREFCNT_dec (self->cb_object); |
2274 | SvREFCNT_dec (self->cb_object); |
|
|
2275 | SvREFCNT_dec (self->incr_text); |
1605 | |
2276 | |
1606 | PROTOTYPES: ENABLE |
2277 | PROTOTYPES: ENABLE |
1607 | |
2278 | |
1608 | void to_json (SV *scalar) |
2279 | void encode_json (SV *scalar) |
1609 | PPCODE: |
2280 | PPCODE: |
1610 | { |
2281 | { |
1611 | JSON json = { F_DEFAULT | F_UTF8 }; |
2282 | JSON json; |
1612 | XPUSHs (encode_json (scalar, &json)); |
2283 | json_init (&json); |
|
|
2284 | json.flags |= F_UTF8; |
|
|
2285 | PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN; |
|
|
2286 | XPUSHs (scalar); |
1613 | } |
2287 | } |
1614 | |
2288 | |
1615 | void from_json (SV *jsonstr) |
2289 | void decode_json (SV *jsonstr) |
1616 | PPCODE: |
2290 | PPCODE: |
1617 | { |
2291 | { |
1618 | JSON json = { F_DEFAULT | F_UTF8 }; |
2292 | JSON json; |
1619 | XPUSHs (decode_json (jsonstr, &json, 0)); |
2293 | json_init (&json); |
|
|
2294 | json.flags |= F_UTF8; |
|
|
2295 | PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN; |
|
|
2296 | XPUSHs (jsonstr); |
1620 | } |
2297 | } |
1621 | |
2298 | |