1 | #include "EXTERN.h" |
1 | #include "EXTERN.h" |
2 | #include "perl.h" |
2 | #include "perl.h" |
3 | #include "XSUB.h" |
3 | #include "XSUB.h" |
4 | |
4 | |
5 | #include "assert.h" |
5 | #include <assert.h> |
6 | #include "string.h" |
6 | #include <string.h> |
7 | #include "stdlib.h" |
7 | #include <stdlib.h> |
|
|
8 | #include <stdio.h> |
|
|
9 | #include <limits.h> |
|
|
10 | #include <float.h> |
8 | |
11 | |
|
|
12 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
|
|
13 | # define snprintf _snprintf // C compilers have this in stdio.h |
|
|
14 | #endif |
|
|
15 | |
|
|
16 | // some old perls do not have this, try to make it work, no |
|
|
17 | // guarantees, though. if it breaks, you get to keep the pieces. |
|
|
18 | #ifndef UTF8_MAXBYTES |
|
|
19 | # define UTF8_MAXBYTES 13 |
|
|
20 | #endif |
|
|
21 | |
|
|
22 | // compatibility with perl <5.18 |
|
|
23 | #ifndef HvNAMELEN_get |
|
|
24 | # define HvNAMELEN_get(hv) strlen (HvNAME (hv)) |
|
|
25 | #endif |
|
|
26 | #ifndef HvNAMELEN |
|
|
27 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
|
|
28 | #endif |
|
|
29 | #ifndef HvNAMEUTF8 |
|
|
30 | # define HvNAMEUTF8(hv) 0 |
|
|
31 | #endif |
|
|
32 | |
|
|
33 | // three extra for rounding, sign, and end of string |
|
|
34 | #define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3) |
|
|
35 | |
9 | #define F_ASCII 0x00000001 |
36 | #define F_ASCII 0x00000001UL |
|
|
37 | #define F_LATIN1 0x00000002UL |
10 | #define F_UTF8 0x00000002 |
38 | #define F_UTF8 0x00000004UL |
11 | #define F_INDENT 0x00000004 |
39 | #define F_INDENT 0x00000008UL |
12 | #define F_CANONICAL 0x00000008 |
40 | #define F_CANONICAL 0x00000010UL |
13 | #define F_SPACE_BEFORE 0x00000010 |
41 | #define F_SPACE_BEFORE 0x00000020UL |
14 | #define F_SPACE_AFTER 0x00000020 |
42 | #define F_SPACE_AFTER 0x00000040UL |
15 | #define F_JSON_RPC 0x00000040 |
|
|
16 | #define F_ALLOW_NONREF 0x00000080 |
43 | #define F_ALLOW_NONREF 0x00000100UL |
|
|
44 | #define F_SHRINK 0x00000200UL |
|
|
45 | #define F_ALLOW_BLESSED 0x00000400UL |
|
|
46 | #define F_CONV_BLESSED 0x00000800UL |
|
|
47 | #define F_RELAXED 0x00001000UL |
|
|
48 | #define F_ALLOW_UNKNOWN 0x00002000UL |
|
|
49 | #define F_ALLOW_TAGS 0x00004000UL |
|
|
50 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
17 | |
51 | |
18 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
52 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
19 | #define F_DEFAULT 0 |
|
|
20 | |
53 | |
21 | #define INIT_SIZE 32 // initial scalar size to be allocated |
54 | #define INIT_SIZE 32 // initial scalar size to be allocated |
|
|
55 | #define INDENT_STEP 3 // spaces per indentation level |
|
|
56 | |
|
|
57 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
|
|
58 | |
|
|
59 | #define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8) |
22 | |
60 | |
23 | #define SB do { |
61 | #define SB do { |
24 | #define SE } while (0) |
62 | #define SE } while (0) |
25 | |
63 | |
26 | static HV *json_stash; |
64 | #if __GNUC__ >= 3 |
|
|
65 | # define expect(expr,value) __builtin_expect ((expr), (value)) |
|
|
66 | # define INLINE static inline |
|
|
67 | #else |
|
|
68 | # define expect(expr,value) (expr) |
|
|
69 | # define INLINE static |
|
|
70 | #endif |
|
|
71 | |
|
|
72 | #define expect_false(expr) expect ((expr) != 0, 0) |
|
|
73 | #define expect_true(expr) expect ((expr) != 0, 1) |
|
|
74 | |
|
|
75 | #define IN_RANGE_INC(type,val,beg,end) \ |
|
|
76 | ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
|
|
77 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
|
|
78 | |
|
|
79 | #define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)" |
|
|
80 | |
|
|
81 | #ifdef USE_ITHREADS |
|
|
82 | # define JSON_SLOW 1 |
|
|
83 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
|
|
84 | # define BOOL_STASH (bool_stash ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1)) |
|
|
85 | #else |
|
|
86 | # define JSON_SLOW 0 |
|
|
87 | # define JSON_STASH json_stash |
|
|
88 | # define BOOL_STASH bool_stash |
|
|
89 | #endif |
|
|
90 | |
|
|
91 | // the amount of HEs to allocate on the stack, when sorting keys |
|
|
92 | #define STACK_HES 64 |
|
|
93 | |
|
|
94 | static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean:: |
|
|
95 | static SV *bool_true, *bool_false, *sv_json; |
|
|
96 | |
|
|
97 | enum { |
|
|
98 | INCR_M_WS = 0, // initial whitespace skipping, must be 0 |
|
|
99 | INCR_M_STR, // inside string |
|
|
100 | INCR_M_BS, // inside backslash |
|
|
101 | INCR_M_C0, // inside comment in initial whitespace sequence |
|
|
102 | INCR_M_C1, // inside comment in other places |
|
|
103 | INCR_M_JSON // outside anything, count nesting |
|
|
104 | }; |
|
|
105 | |
|
|
106 | #define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) |
|
|
107 | |
|
|
108 | typedef struct { |
|
|
109 | U32 flags; |
|
|
110 | U32 max_depth; |
|
|
111 | STRLEN max_size; |
|
|
112 | |
|
|
113 | SV *cb_object; |
|
|
114 | HV *cb_sk_object; |
|
|
115 | |
|
|
116 | // for the incremental parser |
|
|
117 | SV *incr_text; // the source text so far |
|
|
118 | STRLEN incr_pos; // the current offset into the text |
|
|
119 | int incr_nest; // {[]}-nesting level |
|
|
120 | unsigned char incr_mode; |
|
|
121 | } JSON; |
|
|
122 | |
|
|
123 | INLINE void |
|
|
124 | json_init (JSON *json) |
|
|
125 | { |
|
|
126 | Zero (json, 1, JSON); |
|
|
127 | json->max_depth = 512; |
|
|
128 | } |
|
|
129 | |
|
|
130 | ///////////////////////////////////////////////////////////////////////////// |
|
|
131 | // utility functions |
|
|
132 | |
|
|
133 | INLINE SV * |
|
|
134 | get_bool (const char *name) |
|
|
135 | { |
|
|
136 | SV *sv = get_sv (name, 1); |
|
|
137 | |
|
|
138 | SvREADONLY_on (sv); |
|
|
139 | SvREADONLY_on (SvRV (sv)); |
|
|
140 | |
|
|
141 | return sv; |
|
|
142 | } |
|
|
143 | |
|
|
144 | INLINE void |
|
|
145 | shrink (SV *sv) |
|
|
146 | { |
|
|
147 | sv_utf8_downgrade (sv, 1); |
|
|
148 | |
|
|
149 | if (SvLEN (sv) > SvCUR (sv) + 1) |
|
|
150 | { |
|
|
151 | #ifdef SvPV_shrink_to_cur |
|
|
152 | SvPV_shrink_to_cur (sv); |
|
|
153 | #elif defined (SvPV_renew) |
|
|
154 | SvPV_renew (sv, SvCUR (sv) + 1); |
|
|
155 | #endif |
|
|
156 | } |
|
|
157 | } |
|
|
158 | |
|
|
159 | // decode an utf-8 character and return it, or (UV)-1 in |
|
|
160 | // case of an error. |
|
|
161 | // we special-case "safe" characters from U+80 .. U+7FF, |
|
|
162 | // but use the very good perl function to parse anything else. |
|
|
163 | // note that we never call this function for a ascii codepoints |
|
|
164 | INLINE UV |
|
|
165 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
|
|
166 | { |
|
|
167 | if (expect_true (len >= 2 |
|
|
168 | && IN_RANGE_INC (char, s[0], 0xc2, 0xdf) |
|
|
169 | && IN_RANGE_INC (char, s[1], 0x80, 0xbf))) |
|
|
170 | { |
|
|
171 | *clen = 2; |
|
|
172 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
|
|
173 | } |
|
|
174 | else |
|
|
175 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
|
|
176 | } |
|
|
177 | |
|
|
178 | // likewise for encoding, also never called for ascii codepoints |
|
|
179 | // this function takes advantage of this fact, although current gccs |
|
|
180 | // seem to optimise the check for >= 0x80 away anyways |
|
|
181 | INLINE unsigned char * |
|
|
182 | encode_utf8 (unsigned char *s, UV ch) |
|
|
183 | { |
|
|
184 | if (expect_false (ch < 0x000080)) |
|
|
185 | *s++ = ch; |
|
|
186 | else if (expect_true (ch < 0x000800)) |
|
|
187 | *s++ = 0xc0 | ( ch >> 6), |
|
|
188 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
189 | else if ( ch < 0x010000) |
|
|
190 | *s++ = 0xe0 | ( ch >> 12), |
|
|
191 | *s++ = 0x80 | ((ch >> 6) & 0x3f), |
|
|
192 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
193 | else if ( ch < 0x110000) |
|
|
194 | *s++ = 0xf0 | ( ch >> 18), |
|
|
195 | *s++ = 0x80 | ((ch >> 12) & 0x3f), |
|
|
196 | *s++ = 0x80 | ((ch >> 6) & 0x3f), |
|
|
197 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
198 | |
|
|
199 | return s; |
|
|
200 | } |
|
|
201 | |
|
|
202 | // convert offset pointer to character index, sv must be string |
|
|
203 | static STRLEN |
|
|
204 | ptr_to_index (SV *sv, char *offset) |
|
|
205 | { |
|
|
206 | return SvUTF8 (sv) |
|
|
207 | ? utf8_distance (offset, SvPVX (sv)) |
|
|
208 | : offset - SvPVX (sv); |
|
|
209 | } |
|
|
210 | |
|
|
211 | ///////////////////////////////////////////////////////////////////////////// |
|
|
212 | // fp hell |
|
|
213 | |
|
|
214 | // scan a group of digits, and a trailing exponent |
|
|
215 | static void |
|
|
216 | json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth) |
|
|
217 | { |
|
|
218 | UV uaccum = 0; |
|
|
219 | int eaccum = 0; |
|
|
220 | |
|
|
221 | // if we recurse too deep, skip all remaining digits |
|
|
222 | // to avoid a stack overflow attack |
|
|
223 | if (expect_false (--maxdepth <= 0)) |
|
|
224 | while (((U8)*s - '0') < 10) |
|
|
225 | ++s; |
|
|
226 | |
|
|
227 | for (;;) |
|
|
228 | { |
|
|
229 | U8 dig = (U8)*s - '0'; |
|
|
230 | |
|
|
231 | if (expect_false (dig >= 10)) |
|
|
232 | { |
|
|
233 | if (dig == (U8)((U8)'.' - (U8)'0')) |
|
|
234 | { |
|
|
235 | ++s; |
|
|
236 | json_atof_scan1 (s, accum, expo, 1, maxdepth); |
|
|
237 | } |
|
|
238 | else if ((dig | ' ') == 'e' - '0') |
|
|
239 | { |
|
|
240 | int exp2 = 0; |
|
|
241 | int neg = 0; |
|
|
242 | |
|
|
243 | ++s; |
|
|
244 | |
|
|
245 | if (*s == '-') |
|
|
246 | { |
|
|
247 | ++s; |
|
|
248 | neg = 1; |
|
|
249 | } |
|
|
250 | else if (*s == '+') |
|
|
251 | ++s; |
|
|
252 | |
|
|
253 | while ((dig = (U8)*s - '0') < 10) |
|
|
254 | exp2 = exp2 * 10 + *s++ - '0'; |
|
|
255 | |
|
|
256 | *expo += neg ? -exp2 : exp2; |
|
|
257 | } |
|
|
258 | |
|
|
259 | break; |
|
|
260 | } |
|
|
261 | |
|
|
262 | ++s; |
|
|
263 | |
|
|
264 | uaccum = uaccum * 10 + dig; |
|
|
265 | ++eaccum; |
|
|
266 | |
|
|
267 | // if we have too many digits, then recurse for more |
|
|
268 | // we actually do this for rather few digits |
|
|
269 | if (uaccum >= (UV_MAX - 9) / 10) |
|
|
270 | { |
|
|
271 | if (postdp) *expo -= eaccum; |
|
|
272 | json_atof_scan1 (s, accum, expo, postdp, maxdepth); |
|
|
273 | if (postdp) *expo += eaccum; |
|
|
274 | |
|
|
275 | break; |
|
|
276 | } |
|
|
277 | } |
|
|
278 | |
|
|
279 | // this relies greatly on the quality of the pow () |
|
|
280 | // implementation of the platform, but a good |
|
|
281 | // implementation is hard to beat. |
|
|
282 | // (IEEE 754 conformant ones are required to be exact) |
|
|
283 | if (postdp) *expo -= eaccum; |
|
|
284 | *accum += uaccum * Perl_pow (10., *expo); |
|
|
285 | *expo += eaccum; |
|
|
286 | } |
|
|
287 | |
|
|
288 | static NV |
|
|
289 | json_atof (const char *s) |
|
|
290 | { |
|
|
291 | NV accum = 0.; |
|
|
292 | int expo = 0; |
|
|
293 | int neg = 0; |
|
|
294 | |
|
|
295 | if (*s == '-') |
|
|
296 | { |
|
|
297 | ++s; |
|
|
298 | neg = 1; |
|
|
299 | } |
|
|
300 | |
|
|
301 | // a recursion depth of ten gives us >>500 bits |
|
|
302 | json_atof_scan1 (s, &accum, &expo, 0, 10); |
|
|
303 | |
|
|
304 | return neg ? -accum : accum; |
|
|
305 | } |
|
|
306 | |
|
|
307 | // target of scalar reference is bool? -1 == nope, 0 == false, 1 == true |
|
|
308 | static int |
|
|
309 | ref_bool_type (SV *sv) |
|
|
310 | { |
|
|
311 | svtype svt = SvTYPE (sv); |
|
|
312 | |
|
|
313 | if (svt < SVt_PVAV) |
|
|
314 | { |
|
|
315 | STRLEN len = 0; |
|
|
316 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
317 | |
|
|
318 | if (len == 1) |
|
|
319 | if (*pv == '1') |
|
|
320 | return 1; |
|
|
321 | else if (*pv == '0') |
|
|
322 | return 0; |
|
|
323 | } |
|
|
324 | |
|
|
325 | return -1; |
|
|
326 | } |
|
|
327 | |
|
|
328 | // returns whether scalar is not a reference in the sense of allow_nonref |
|
|
329 | static int |
|
|
330 | json_nonref (SV *scalar) |
|
|
331 | { |
|
|
332 | if (!SvROK (scalar)) |
|
|
333 | return 1; |
|
|
334 | |
|
|
335 | scalar = SvRV (scalar); |
|
|
336 | |
|
|
337 | if (SvTYPE (scalar) >= SVt_PVMG) |
|
|
338 | { |
|
|
339 | if (SvSTASH (scalar) == bool_stash) |
|
|
340 | return 1; |
|
|
341 | |
|
|
342 | if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0) |
|
|
343 | return 1; |
|
|
344 | } |
|
|
345 | |
|
|
346 | return 0; |
|
|
347 | } |
|
|
348 | |
|
|
349 | ///////////////////////////////////////////////////////////////////////////// |
|
|
350 | // encoder |
27 | |
351 | |
28 | // structure used for encoding JSON |
352 | // structure used for encoding JSON |
29 | typedef struct |
353 | typedef struct |
30 | { |
354 | { |
31 | char *cur; |
355 | char *cur; // SvPVX (sv) + current output position |
32 | STRLEN len; // SvLEN (sv) |
|
|
33 | char *end; // SvEND (sv) |
356 | char *end; // SvEND (sv) |
34 | SV *sv; |
357 | SV *sv; // result scalar |
35 | UV flags; |
358 | JSON json; |
36 | int max_recurse; |
359 | U32 indent; // indentation level |
37 | int indent; |
360 | UV limit; // escape character values >= this value when encoding |
38 | } enc_t; |
361 | } enc_t; |
39 | |
362 | |
40 | // structure used for decoding JSON |
363 | INLINE void |
41 | typedef struct |
|
|
42 | { |
|
|
43 | char *cur; |
|
|
44 | char *end; |
|
|
45 | const char *err; |
|
|
46 | UV flags; |
|
|
47 | } dec_t; |
|
|
48 | |
|
|
49 | static UV * |
|
|
50 | SvJSON (SV *sv) |
|
|
51 | { |
|
|
52 | if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) |
|
|
53 | croak ("object is not of type JSON::XS"); |
|
|
54 | |
|
|
55 | return &SvUVX (SvRV (sv)); |
|
|
56 | } |
|
|
57 | |
|
|
58 | ///////////////////////////////////////////////////////////////////////////// |
|
|
59 | |
|
|
60 | static void |
|
|
61 | need (enc_t *enc, STRLEN len) |
364 | need (enc_t *enc, STRLEN len) |
62 | { |
365 | { |
63 | if (enc->cur + len >= enc->end) |
366 | if (expect_false (enc->cur + len >= enc->end)) |
64 | { |
367 | { |
65 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
368 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
66 | SvGROW (enc->sv, cur + len + 1); |
369 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
67 | enc->cur = SvPVX (enc->sv) + cur; |
370 | enc->cur = SvPVX (enc->sv) + cur; |
68 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); |
371 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
69 | } |
372 | } |
70 | } |
373 | } |
71 | |
374 | |
72 | static void |
375 | INLINE void |
73 | encode_ch (enc_t *enc, char ch) |
376 | encode_ch (enc_t *enc, char ch) |
74 | { |
377 | { |
75 | need (enc, 1); |
378 | need (enc, 1); |
76 | *enc->cur++ = ch; |
379 | *enc->cur++ = ch; |
77 | } |
380 | } |
… | |
… | |
85 | |
388 | |
86 | while (str < end) |
389 | while (str < end) |
87 | { |
390 | { |
88 | unsigned char ch = *(unsigned char *)str; |
391 | unsigned char ch = *(unsigned char *)str; |
89 | |
392 | |
90 | if (ch == '"') |
393 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
91 | { |
394 | { |
|
|
395 | if (expect_false (ch == '"')) // but with slow exceptions |
|
|
396 | { |
92 | need (enc, len += 1); |
397 | need (enc, len += 1); |
93 | *enc->cur++ = '\\'; |
398 | *enc->cur++ = '\\'; |
94 | *enc->cur++ = '"'; |
399 | *enc->cur++ = '"'; |
95 | ++str; |
|
|
96 | } |
400 | } |
97 | else if (ch == '\\') |
401 | else if (expect_false (ch == '\\')) |
98 | { |
402 | { |
99 | need (enc, len += 1); |
403 | need (enc, len += 1); |
100 | *enc->cur++ = '\\'; |
404 | *enc->cur++ = '\\'; |
101 | *enc->cur++ = '\\'; |
405 | *enc->cur++ = '\\'; |
102 | ++str; |
|
|
103 | } |
406 | } |
104 | else if (ch >= 0x20 && ch < 0x80) // most common case |
407 | else |
105 | { |
|
|
106 | *enc->cur++ = ch; |
408 | *enc->cur++ = ch; |
107 | ++str; |
409 | |
108 | } |
|
|
109 | else if (ch == '\015') |
|
|
110 | { |
|
|
111 | need (enc, len += 1); |
|
|
112 | *enc->cur++ = '\\'; |
|
|
113 | *enc->cur++ = 'r'; |
|
|
114 | ++str; |
|
|
115 | } |
|
|
116 | else if (ch == '\012') |
|
|
117 | { |
|
|
118 | need (enc, len += 1); |
|
|
119 | *enc->cur++ = '\\'; |
|
|
120 | *enc->cur++ = 'n'; |
|
|
121 | ++str; |
410 | ++str; |
122 | } |
411 | } |
123 | else |
412 | else |
124 | { |
413 | { |
125 | STRLEN clen; |
414 | switch (ch) |
126 | UV uch; |
|
|
127 | |
|
|
128 | if (is_utf8) |
|
|
129 | { |
415 | { |
130 | uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); |
416 | case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; |
131 | if (clen < 0) |
417 | case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; |
132 | croak ("malformed UTF-8 character in string, cannot convert to JSON"); |
418 | case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; |
133 | } |
419 | case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; |
134 | else |
420 | case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; |
135 | { |
|
|
136 | uch = ch; |
|
|
137 | clen = 1; |
|
|
138 | } |
|
|
139 | |
421 | |
140 | if (uch < 0x80 || enc->flags & F_ASCII) |
422 | default: |
141 | { |
|
|
142 | if (uch > 0xFFFFUL) |
|
|
143 | { |
423 | { |
|
|
424 | STRLEN clen; |
|
|
425 | UV uch; |
|
|
426 | |
|
|
427 | if (is_utf8) |
|
|
428 | { |
|
|
429 | uch = decode_utf8 (str, end - str, &clen); |
|
|
430 | if (clen == (STRLEN)-1) |
|
|
431 | croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
|
|
432 | } |
|
|
433 | else |
|
|
434 | { |
|
|
435 | uch = ch; |
|
|
436 | clen = 1; |
|
|
437 | } |
|
|
438 | |
|
|
439 | if (uch < 0x80/*0x20*/ || uch >= enc->limit) |
|
|
440 | { |
|
|
441 | if (uch >= 0x10000UL) |
|
|
442 | { |
|
|
443 | if (uch >= 0x110000UL) |
|
|
444 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
|
|
445 | |
144 | need (enc, len += 11); |
446 | need (enc, len += 11); |
145 | sprintf (enc->cur, "\\u%04x\\u%04x", |
447 | sprintf (enc->cur, "\\u%04x\\u%04x", |
146 | (uch - 0x10000) / 0x400 + 0xD800, |
448 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
147 | (uch - 0x10000) % 0x400 + 0xDC00); |
449 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
148 | enc->cur += 12; |
450 | enc->cur += 12; |
|
|
451 | } |
|
|
452 | else |
|
|
453 | { |
|
|
454 | need (enc, len += 5); |
|
|
455 | *enc->cur++ = '\\'; |
|
|
456 | *enc->cur++ = 'u'; |
|
|
457 | *enc->cur++ = PL_hexdigit [ uch >> 12 ]; |
|
|
458 | *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; |
|
|
459 | *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; |
|
|
460 | *enc->cur++ = PL_hexdigit [(uch >> 0) & 15]; |
|
|
461 | } |
|
|
462 | |
|
|
463 | str += clen; |
|
|
464 | } |
|
|
465 | else if (enc->json.flags & F_LATIN1) |
|
|
466 | { |
|
|
467 | *enc->cur++ = uch; |
|
|
468 | str += clen; |
|
|
469 | } |
|
|
470 | else if (is_utf8) |
|
|
471 | { |
|
|
472 | need (enc, len += clen); |
|
|
473 | do |
|
|
474 | { |
|
|
475 | *enc->cur++ = *str++; |
|
|
476 | } |
|
|
477 | while (--clen); |
|
|
478 | } |
|
|
479 | else |
|
|
480 | { |
|
|
481 | need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
|
|
482 | enc->cur = encode_utf8 (enc->cur, uch); |
|
|
483 | ++str; |
|
|
484 | } |
149 | } |
485 | } |
150 | else |
|
|
151 | { |
|
|
152 | static char hexdigit [16] = "0123456789abcdef"; |
|
|
153 | need (enc, len += 5); |
|
|
154 | *enc->cur++ = '\\'; |
|
|
155 | *enc->cur++ = 'u'; |
|
|
156 | *enc->cur++ = hexdigit [ uch >> 12 ]; |
|
|
157 | *enc->cur++ = hexdigit [(uch >> 8) & 15]; |
|
|
158 | *enc->cur++ = hexdigit [(uch >> 4) & 15]; |
|
|
159 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
|
|
160 | } |
|
|
161 | |
|
|
162 | str += clen; |
|
|
163 | } |
|
|
164 | else if (is_utf8) |
|
|
165 | { |
|
|
166 | need (enc, len += clen); |
|
|
167 | while (clen--) |
|
|
168 | *enc->cur++ = *str++; |
|
|
169 | } |
|
|
170 | else |
|
|
171 | { |
|
|
172 | need (enc, 10); // never more than 11 bytes needed |
|
|
173 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
|
|
174 | ++str; |
|
|
175 | } |
486 | } |
176 | } |
487 | } |
177 | |
488 | |
178 | --len; |
489 | --len; |
179 | } |
490 | } |
180 | } |
491 | } |
181 | |
492 | |
182 | #define INDENT SB \ |
493 | INLINE void |
|
|
494 | encode_indent (enc_t *enc) |
|
|
495 | { |
183 | if (enc->flags & F_INDENT) \ |
496 | if (enc->json.flags & F_INDENT) |
184 | { \ |
497 | { |
185 | int i_; \ |
498 | int spaces = enc->indent * INDENT_STEP; |
186 | need (enc, enc->indent); \ |
499 | |
187 | for (i_ = enc->indent * 3; i_--; )\ |
500 | need (enc, spaces); |
|
|
501 | memset (enc->cur, ' ', spaces); |
|
|
502 | enc->cur += spaces; |
|
|
503 | } |
|
|
504 | } |
|
|
505 | |
|
|
506 | INLINE void |
|
|
507 | encode_space (enc_t *enc) |
|
|
508 | { |
|
|
509 | need (enc, 1); |
|
|
510 | encode_ch (enc, ' '); |
|
|
511 | } |
|
|
512 | |
|
|
513 | INLINE void |
|
|
514 | encode_nl (enc_t *enc) |
|
|
515 | { |
|
|
516 | if (enc->json.flags & F_INDENT) |
|
|
517 | { |
|
|
518 | need (enc, 1); |
188 | encode_ch (enc, ' '); \ |
519 | encode_ch (enc, '\n'); |
189 | } \ |
520 | } |
190 | SE |
521 | } |
191 | |
522 | |
192 | #define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE |
523 | INLINE void |
193 | #define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE |
524 | encode_comma (enc_t *enc) |
194 | #define COMMA SB \ |
525 | { |
195 | encode_ch (enc, ','); \ |
526 | encode_ch (enc, ','); |
|
|
527 | |
196 | if (enc->flags & F_INDENT) \ |
528 | if (enc->json.flags & F_INDENT) |
197 | NL; \ |
529 | encode_nl (enc); |
198 | else if (enc->flags & F_SPACE_AFTER) \ |
530 | else if (enc->json.flags & F_SPACE_AFTER) |
199 | SPACE; \ |
531 | encode_space (enc); |
200 | SE |
532 | } |
201 | |
533 | |
202 | static void encode_sv (enc_t *enc, SV *sv); |
534 | static void encode_sv (enc_t *enc, SV *sv); |
203 | |
535 | |
204 | static void |
536 | static void |
205 | encode_av (enc_t *enc, AV *av) |
537 | encode_av (enc_t *enc, AV *av) |
206 | { |
538 | { |
207 | int i, len = av_len (av); |
539 | int i, len = av_len (av); |
208 | |
540 | |
|
|
541 | if (enc->indent >= enc->json.max_depth) |
|
|
542 | croak (ERR_NESTING_EXCEEDED); |
|
|
543 | |
209 | encode_ch (enc, '['); NL; |
544 | encode_ch (enc, '['); |
210 | ++enc->indent; |
|
|
211 | |
545 | |
|
|
546 | if (len >= 0) |
|
|
547 | { |
|
|
548 | encode_nl (enc); ++enc->indent; |
|
|
549 | |
212 | for (i = 0; i <= len; ++i) |
550 | for (i = 0; i <= len; ++i) |
213 | { |
551 | { |
214 | INDENT; |
552 | SV **svp = av_fetch (av, i, 0); |
215 | encode_sv (enc, *av_fetch (av, i, 0)); |
|
|
216 | |
553 | |
|
|
554 | encode_indent (enc); |
|
|
555 | |
|
|
556 | if (svp) |
|
|
557 | encode_sv (enc, *svp); |
|
|
558 | else |
|
|
559 | encode_str (enc, "null", 4, 0); |
|
|
560 | |
217 | if (i < len) |
561 | if (i < len) |
218 | COMMA; |
562 | encode_comma (enc); |
219 | } |
563 | } |
220 | |
564 | |
221 | NL; |
565 | encode_nl (enc); --enc->indent; encode_indent (enc); |
|
|
566 | } |
222 | |
567 | |
223 | --enc->indent; |
|
|
224 | INDENT; encode_ch (enc, ']'); |
568 | encode_ch (enc, ']'); |
225 | } |
569 | } |
226 | |
570 | |
227 | static void |
571 | static void |
228 | encode_he (enc_t *enc, HE *he) |
572 | encode_hk (enc_t *enc, HE *he) |
229 | { |
573 | { |
230 | encode_ch (enc, '"'); |
574 | encode_ch (enc, '"'); |
231 | |
575 | |
232 | if (HeKLEN (he) == HEf_SVKEY) |
576 | if (HeKLEN (he) == HEf_SVKEY) |
233 | { |
577 | { |
234 | SV *sv = HeSVKEY (he); |
578 | SV *sv = HeSVKEY (he); |
235 | STRLEN len; |
579 | STRLEN len; |
236 | char *str; |
580 | char *str; |
237 | |
581 | |
238 | SvGETMAGIC (sv); |
582 | SvGETMAGIC (sv); |
239 | str = SvPV (sv, len); |
583 | str = SvPV (sv, len); |
240 | |
584 | |
241 | encode_str (enc, str, len, SvUTF8 (sv)); |
585 | encode_str (enc, str, len, SvUTF8 (sv)); |
242 | } |
586 | } |
243 | else |
587 | else |
244 | encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
588 | encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
245 | |
589 | |
246 | encode_ch (enc, '"'); |
590 | encode_ch (enc, '"'); |
247 | |
591 | |
248 | if (enc->flags & F_SPACE_BEFORE) SPACE; |
592 | if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); |
249 | encode_ch (enc, ':'); |
593 | encode_ch (enc, ':'); |
250 | if (enc->flags & F_SPACE_AFTER ) SPACE; |
594 | if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); |
251 | encode_sv (enc, HeVAL (he)); |
|
|
252 | } |
595 | } |
253 | |
596 | |
254 | // compare hash entries, used when all keys are bytestrings |
597 | // compare hash entries, used when all keys are bytestrings |
255 | static int |
598 | static int |
256 | he_cmp_fast (const void *a_, const void *b_) |
599 | he_cmp_fast (const void *a_, const void *b_) |
… | |
… | |
261 | HE *b = *(HE **)b_; |
604 | HE *b = *(HE **)b_; |
262 | |
605 | |
263 | STRLEN la = HeKLEN (a); |
606 | STRLEN la = HeKLEN (a); |
264 | STRLEN lb = HeKLEN (b); |
607 | STRLEN lb = HeKLEN (b); |
265 | |
608 | |
266 | if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) |
609 | if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la))) |
267 | cmp = la < lb ? -1 : la == lb ? 0 : 1; |
610 | cmp = lb - la; |
268 | |
611 | |
269 | return cmp; |
612 | return cmp; |
270 | } |
613 | } |
271 | |
614 | |
272 | // compare hash entries, used when some keys are sv's or utf-x |
615 | // compare hash entries, used when some keys are sv's or utf-x |
273 | static int |
616 | static int |
274 | he_cmp_slow (const void *a, const void *b) |
617 | he_cmp_slow (const void *a, const void *b) |
275 | { |
618 | { |
276 | return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); |
619 | return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a)); |
277 | } |
620 | } |
278 | |
621 | |
279 | static void |
622 | static void |
280 | encode_hv (enc_t *enc, HV *hv) |
623 | encode_hv (enc_t *enc, HV *hv) |
281 | { |
624 | { |
282 | int count, i; |
625 | HE *he; |
283 | |
626 | |
284 | encode_ch (enc, '{'); NL; ++enc->indent; |
627 | if (enc->indent >= enc->json.max_depth) |
|
|
628 | croak (ERR_NESTING_EXCEEDED); |
285 | |
629 | |
286 | if ((count = hv_iterinit (hv))) |
630 | encode_ch (enc, '{'); |
287 | { |
631 | |
288 | // for canonical output we have to sort by keys first |
632 | // for canonical output we have to sort by keys first |
289 | // actually, this is mostly due to the stupid so-called |
633 | // actually, this is mostly due to the stupid so-called |
290 | // security workaround added somewhere in 5.8.x. |
634 | // security workaround added somewhere in 5.8.x |
291 | // that randomises hash orderings |
635 | // that randomises hash orderings |
292 | if (enc->flags & F_CANONICAL) |
636 | if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv)) |
|
|
637 | { |
|
|
638 | int count = hv_iterinit (hv); |
|
|
639 | |
|
|
640 | if (SvMAGICAL (hv)) |
293 | { |
641 | { |
294 | HE *he, *hes [count]; |
642 | // need to count by iterating. could improve by dynamically building the vector below |
|
|
643 | // but I don't care for the speed of this special case. |
|
|
644 | // note also that we will run into undefined behaviour when the two iterations |
|
|
645 | // do not result in the same count, something I might care for in some later release. |
|
|
646 | |
|
|
647 | count = 0; |
|
|
648 | while (hv_iternext (hv)) |
|
|
649 | ++count; |
|
|
650 | |
|
|
651 | hv_iterinit (hv); |
|
|
652 | } |
|
|
653 | |
|
|
654 | if (count) |
|
|
655 | { |
295 | int fast = 1; |
656 | int i, fast = 1; |
|
|
657 | HE *hes_stack [STACK_HES]; |
|
|
658 | HE **hes = hes_stack; |
|
|
659 | |
|
|
660 | // allocate larger arrays on the heap |
|
|
661 | if (count > STACK_HES) |
|
|
662 | { |
|
|
663 | SV *sv = sv_2mortal (NEWSV (0, count * sizeof (*hes))); |
|
|
664 | hes = (HE **)SvPVX (sv); |
|
|
665 | } |
296 | |
666 | |
297 | i = 0; |
667 | i = 0; |
298 | while ((he = hv_iternext (hv))) |
668 | while ((he = hv_iternext (hv))) |
299 | { |
669 | { |
300 | hes [i++] = he; |
670 | hes [i++] = he; |
… | |
… | |
306 | |
676 | |
307 | if (fast) |
677 | if (fast) |
308 | qsort (hes, count, sizeof (HE *), he_cmp_fast); |
678 | qsort (hes, count, sizeof (HE *), he_cmp_fast); |
309 | else |
679 | else |
310 | { |
680 | { |
311 | // hack to disable "use bytes" |
681 | // hack to forcefully disable "use bytes" |
312 | COP *oldcop = PL_curcop, cop; |
682 | COP cop = *PL_curcop; |
313 | cop.op_private = 0; |
683 | cop.op_private = 0; |
|
|
684 | |
|
|
685 | ENTER; |
|
|
686 | SAVETMPS; |
|
|
687 | |
|
|
688 | SAVEVPTR (PL_curcop); |
314 | PL_curcop = &cop; |
689 | PL_curcop = &cop; |
315 | |
690 | |
316 | SAVETMPS; |
|
|
317 | qsort (hes, count, sizeof (HE *), he_cmp_slow); |
691 | qsort (hes, count, sizeof (HE *), he_cmp_slow); |
|
|
692 | |
318 | FREETMPS; |
693 | FREETMPS; |
319 | |
694 | LEAVE; |
320 | PL_curcop = oldcop; |
|
|
321 | } |
695 | } |
322 | |
696 | |
323 | for (i = 0; i < count; ++i) |
697 | encode_nl (enc); ++enc->indent; |
|
|
698 | |
|
|
699 | while (count--) |
324 | { |
700 | { |
325 | INDENT; |
701 | encode_indent (enc); |
|
|
702 | he = hes [count]; |
326 | encode_he (enc, hes [i]); |
703 | encode_hk (enc, he); |
|
|
704 | encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
327 | |
705 | |
328 | if (i < count - 1) |
706 | if (count) |
329 | COMMA; |
707 | encode_comma (enc); |
330 | } |
708 | } |
331 | |
709 | |
332 | NL; |
710 | encode_nl (enc); --enc->indent; encode_indent (enc); |
333 | } |
711 | } |
|
|
712 | } |
|
|
713 | else |
|
|
714 | { |
|
|
715 | if (hv_iterinit (hv) || SvMAGICAL (hv)) |
|
|
716 | if ((he = hv_iternext (hv))) |
|
|
717 | { |
|
|
718 | encode_nl (enc); ++enc->indent; |
|
|
719 | |
|
|
720 | for (;;) |
|
|
721 | { |
|
|
722 | encode_indent (enc); |
|
|
723 | encode_hk (enc, he); |
|
|
724 | encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
|
|
725 | |
|
|
726 | if (!(he = hv_iternext (hv))) |
|
|
727 | break; |
|
|
728 | |
|
|
729 | encode_comma (enc); |
|
|
730 | } |
|
|
731 | |
|
|
732 | encode_nl (enc); --enc->indent; encode_indent (enc); |
|
|
733 | } |
|
|
734 | } |
|
|
735 | |
|
|
736 | encode_ch (enc, '}'); |
|
|
737 | } |
|
|
738 | |
|
|
739 | // encode objects, arrays and special \0=false and \1=true values. |
|
|
740 | static void |
|
|
741 | encode_rv (enc_t *enc, SV *sv) |
|
|
742 | { |
|
|
743 | svtype svt; |
|
|
744 | GV *method; |
|
|
745 | |
|
|
746 | SvGETMAGIC (sv); |
|
|
747 | svt = SvTYPE (sv); |
|
|
748 | |
|
|
749 | if (expect_false (SvOBJECT (sv))) |
|
|
750 | { |
|
|
751 | HV *stash = SvSTASH (sv); |
|
|
752 | |
|
|
753 | if (stash == bool_stash) |
|
|
754 | { |
|
|
755 | if (SvIV (sv)) |
|
|
756 | encode_str (enc, "true", 4, 0); |
|
|
757 | else |
|
|
758 | encode_str (enc, "false", 5, 0); |
|
|
759 | } |
|
|
760 | else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0))) |
|
|
761 | { |
|
|
762 | int count; |
|
|
763 | dSP; |
|
|
764 | |
|
|
765 | ENTER; SAVETMPS; |
|
|
766 | SAVESTACK_POS (); |
|
|
767 | PUSHMARK (SP); |
|
|
768 | EXTEND (SP, 2); |
|
|
769 | // we re-bless the reference to get overload and other niceties right |
|
|
770 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
|
|
771 | PUSHs (sv_json); |
|
|
772 | |
|
|
773 | PUTBACK; |
|
|
774 | count = call_sv ((SV *)GvCV (method), G_ARRAY); |
|
|
775 | SPAGAIN; |
|
|
776 | |
|
|
777 | // catch this surprisingly common error |
|
|
778 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
779 | croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
780 | |
|
|
781 | encode_ch (enc, '('); |
|
|
782 | encode_ch (enc, '"'); |
|
|
783 | encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash)); |
|
|
784 | encode_ch (enc, '"'); |
|
|
785 | encode_ch (enc, ')'); |
|
|
786 | encode_ch (enc, '['); |
|
|
787 | |
|
|
788 | while (count) |
|
|
789 | { |
|
|
790 | encode_sv (enc, SP[1 - count--]); |
|
|
791 | |
|
|
792 | if (count) |
|
|
793 | encode_ch (enc, ','); |
|
|
794 | } |
|
|
795 | |
|
|
796 | encode_ch (enc, ']'); |
|
|
797 | |
|
|
798 | FREETMPS; LEAVE; |
|
|
799 | } |
|
|
800 | else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0))) |
|
|
801 | { |
|
|
802 | dSP; |
|
|
803 | |
|
|
804 | ENTER; SAVETMPS; |
|
|
805 | PUSHMARK (SP); |
|
|
806 | // we re-bless the reference to get overload and other niceties right |
|
|
807 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
|
|
808 | |
|
|
809 | // calling with G_SCALAR ensures that we always get a 1 return value |
|
|
810 | PUTBACK; |
|
|
811 | call_sv ((SV *)GvCV (method), G_SCALAR); |
|
|
812 | SPAGAIN; |
|
|
813 | |
|
|
814 | // catch this surprisingly common error |
|
|
815 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
816 | croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
817 | |
|
|
818 | sv = POPs; |
|
|
819 | PUTBACK; |
|
|
820 | |
|
|
821 | encode_sv (enc, sv); |
|
|
822 | |
|
|
823 | FREETMPS; LEAVE; |
|
|
824 | } |
|
|
825 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
826 | encode_str (enc, "null", 4, 0); |
334 | else |
827 | else |
335 | { |
828 | croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)", |
336 | SV *sv; |
829 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
337 | HE *he = hv_iternext (hv); |
830 | } |
338 | |
831 | else if (svt == SVt_PVHV) |
339 | for (;;) |
832 | encode_hv (enc, (HV *)sv); |
340 | { |
833 | else if (svt == SVt_PVAV) |
341 | INDENT; |
834 | encode_av (enc, (AV *)sv); |
342 | encode_he (enc, he); |
835 | else if (svt < SVt_PVAV) |
343 | |
|
|
344 | if (!(he = hv_iternext (hv))) |
|
|
345 | break; |
|
|
346 | |
|
|
347 | COMMA; |
|
|
348 | } |
|
|
349 | |
|
|
350 | NL; |
|
|
351 | } |
|
|
352 | } |
836 | { |
|
|
837 | int bool_type = ref_bool_type (sv); |
353 | |
838 | |
354 | --enc->indent; INDENT; encode_ch (enc, '}'); |
839 | if (bool_type == 1) |
|
|
840 | encode_str (enc, "true", 4, 0); |
|
|
841 | else if (bool_type == 0) |
|
|
842 | encode_str (enc, "false", 5, 0); |
|
|
843 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
844 | encode_str (enc, "null", 4, 0); |
|
|
845 | else |
|
|
846 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
|
|
847 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
848 | } |
|
|
849 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
850 | encode_str (enc, "null", 4, 0); |
|
|
851 | else |
|
|
852 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
|
|
853 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
355 | } |
854 | } |
356 | |
855 | |
357 | static void |
856 | static void |
358 | encode_sv (enc_t *enc, SV *sv) |
857 | encode_sv (enc_t *enc, SV *sv) |
359 | { |
858 | { |
… | |
… | |
367 | encode_str (enc, str, len, SvUTF8 (sv)); |
866 | encode_str (enc, str, len, SvUTF8 (sv)); |
368 | encode_ch (enc, '"'); |
867 | encode_ch (enc, '"'); |
369 | } |
868 | } |
370 | else if (SvNOKp (sv)) |
869 | else if (SvNOKp (sv)) |
371 | { |
870 | { |
|
|
871 | // trust that perl will do the right thing w.r.t. JSON syntax. |
372 | need (enc, NV_DIG + 32); |
872 | need (enc, NV_DIG + 32); |
373 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
873 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
374 | enc->cur += strlen (enc->cur); |
874 | enc->cur += strlen (enc->cur); |
375 | } |
875 | } |
376 | else if (SvIOKp (sv)) |
876 | else if (SvIOKp (sv)) |
377 | { |
877 | { |
|
|
878 | // we assume we can always read an IV as a UV and vice versa |
|
|
879 | // we assume two's complement |
|
|
880 | // we assume no aliasing issues in the union |
|
|
881 | if (SvIsUV (sv) ? SvUVX (sv) <= 59000 |
|
|
882 | : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000) |
|
|
883 | { |
|
|
884 | // optimise the "small number case" |
|
|
885 | // code will likely be branchless and use only a single multiplication |
|
|
886 | // works for numbers up to 59074 |
|
|
887 | I32 i = SvIVX (sv); |
|
|
888 | U32 u; |
|
|
889 | char digit, nz = 0; |
|
|
890 | |
378 | need (enc, 64); |
891 | need (enc, 6); |
|
|
892 | |
|
|
893 | *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0; |
|
|
894 | u = i < 0 ? -i : i; |
|
|
895 | |
|
|
896 | // convert to 4.28 fixed-point representation |
|
|
897 | u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits |
|
|
898 | |
|
|
899 | // now output digit by digit, each time masking out the integer part |
|
|
900 | // and multiplying by 5 while moving the decimal point one to the right, |
|
|
901 | // resulting in a net multiplication by 10. |
|
|
902 | // we always write the digit to memory but conditionally increment |
|
|
903 | // the pointer, to enable the use of conditional move instructions. |
|
|
904 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5; |
|
|
905 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5; |
|
|
906 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5; |
|
|
907 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5; |
|
|
908 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
|
|
909 | } |
|
|
910 | else |
|
|
911 | { |
|
|
912 | // large integer, use the (rather slow) snprintf way. |
|
|
913 | need (enc, IVUV_MAXCHARS); |
379 | enc->cur += |
914 | enc->cur += |
380 | SvIsUV(sv) |
915 | SvIsUV(sv) |
381 | ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) |
916 | ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) |
382 | : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); |
917 | : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); |
|
|
918 | } |
383 | } |
919 | } |
384 | else if (SvROK (sv)) |
920 | else if (SvROK (sv)) |
385 | { |
921 | encode_rv (enc, SvRV (sv)); |
386 | if (!--enc->max_recurse) |
922 | else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) |
387 | croak ("data structure too deep (hit recursion limit)"); |
|
|
388 | |
|
|
389 | sv = SvRV (sv); |
|
|
390 | |
|
|
391 | switch (SvTYPE (sv)) |
|
|
392 | { |
|
|
393 | case SVt_PVAV: encode_av (enc, (AV *)sv); break; |
|
|
394 | case SVt_PVHV: encode_hv (enc, (HV *)sv); break; |
|
|
395 | |
|
|
396 | default: |
|
|
397 | croak ("JSON can only represent references to arrays or hashes"); |
|
|
398 | } |
|
|
399 | } |
|
|
400 | else if (!SvOK (sv)) |
|
|
401 | encode_str (enc, "null", 4, 0); |
923 | encode_str (enc, "null", 4, 0); |
402 | else |
924 | else |
403 | croak ("encountered perl type that JSON cannot handle"); |
925 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data", |
|
|
926 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
404 | } |
927 | } |
405 | |
928 | |
406 | static SV * |
929 | static SV * |
407 | encode_json (SV *scalar, UV flags) |
930 | encode_json (SV *scalar, JSON *json) |
408 | { |
931 | { |
409 | if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
|
|
410 | croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); |
|
|
411 | |
|
|
412 | enc_t enc; |
932 | enc_t enc; |
413 | enc.flags = flags; |
933 | |
|
|
934 | if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar)) |
|
|
935 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
|
|
936 | |
|
|
937 | enc.json = *json; |
414 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
938 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
415 | enc.cur = SvPVX (enc.sv); |
939 | enc.cur = SvPVX (enc.sv); |
416 | enc.end = SvEND (enc.sv); |
940 | enc.end = SvEND (enc.sv); |
417 | enc.max_recurse = 0; |
|
|
418 | enc.indent = 0; |
941 | enc.indent = 0; |
|
|
942 | enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
|
|
943 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
|
|
944 | : 0x110000UL; |
419 | |
945 | |
420 | SvPOK_only (enc.sv); |
946 | SvPOK_only (enc.sv); |
421 | encode_sv (&enc, scalar); |
947 | encode_sv (&enc, scalar); |
|
|
948 | encode_nl (&enc); |
422 | |
949 | |
|
|
950 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
|
|
951 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
|
|
952 | |
423 | if (!(flags & (F_ASCII | F_UTF8))) |
953 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
424 | SvUTF8_on (enc.sv); |
954 | SvUTF8_on (enc.sv); |
425 | |
955 | |
426 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
956 | if (enc.json.flags & F_SHRINK) |
|
|
957 | shrink (enc.sv); |
|
|
958 | |
427 | return enc.sv; |
959 | return enc.sv; |
428 | } |
960 | } |
429 | |
961 | |
430 | ///////////////////////////////////////////////////////////////////////////// |
962 | ///////////////////////////////////////////////////////////////////////////// |
|
|
963 | // decoder |
431 | |
964 | |
432 | #define WS \ |
965 | // structure used for decoding JSON |
|
|
966 | typedef struct |
|
|
967 | { |
|
|
968 | char *cur; // current parser pointer |
|
|
969 | char *end; // end of input string |
|
|
970 | const char *err; // parse error, if != 0 |
|
|
971 | JSON json; |
|
|
972 | U32 depth; // recursion depth |
|
|
973 | U32 maxdepth; // recursion depth limit |
|
|
974 | } dec_t; |
|
|
975 | |
|
|
976 | INLINE void |
|
|
977 | decode_comment (dec_t *dec) |
|
|
978 | { |
|
|
979 | // only '#'-style comments allowed a.t.m. |
|
|
980 | |
|
|
981 | while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) |
|
|
982 | ++dec->cur; |
|
|
983 | } |
|
|
984 | |
|
|
985 | INLINE void |
|
|
986 | decode_ws (dec_t *dec) |
|
|
987 | { |
433 | for (;;) \ |
988 | for (;;) |
434 | { \ |
989 | { |
435 | char ch = *dec->cur; \ |
990 | char ch = *dec->cur; |
|
|
991 | |
436 | if (ch > 0x20 \ |
992 | if (ch > 0x20) |
|
|
993 | { |
|
|
994 | if (expect_false (ch == '#')) |
|
|
995 | { |
|
|
996 | if (dec->json.flags & F_RELAXED) |
|
|
997 | decode_comment (dec); |
|
|
998 | else |
|
|
999 | break; |
|
|
1000 | } |
|
|
1001 | else |
|
|
1002 | break; |
|
|
1003 | } |
437 | || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ |
1004 | else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09) |
438 | break; \ |
1005 | break; // parse error, but let higher level handle it, gives better error messages |
|
|
1006 | |
439 | ++dec->cur; \ |
1007 | ++dec->cur; |
440 | } |
1008 | } |
|
|
1009 | } |
441 | |
1010 | |
442 | #define ERR(reason) SB dec->err = reason; goto fail; SE |
1011 | #define ERR(reason) SB dec->err = reason; goto fail; SE |
|
|
1012 | |
443 | #define EXPECT_CH(ch) SB \ |
1013 | #define EXPECT_CH(ch) SB \ |
444 | if (*dec->cur != ch) \ |
1014 | if (*dec->cur != ch) \ |
445 | ERR (# ch " expected"); \ |
1015 | ERR (# ch " expected"); \ |
446 | ++dec->cur; \ |
1016 | ++dec->cur; \ |
447 | SE |
1017 | SE |
448 | |
1018 | |
|
|
1019 | #define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED) |
|
|
1020 | #define DEC_DEC_DEPTH --dec->depth |
|
|
1021 | |
449 | static SV *decode_sv (dec_t *dec); |
1022 | static SV *decode_sv (dec_t *dec); |
450 | |
1023 | |
451 | static signed char decode_hexdigit[256]; |
1024 | static signed char decode_hexdigit[256]; |
452 | |
1025 | |
453 | static UV |
1026 | static UV |
454 | decode_4hex (dec_t *dec) |
1027 | decode_4hex (dec_t *dec) |
455 | { |
1028 | { |
456 | signed char d1, d2, d3, d4; |
1029 | signed char d1, d2, d3, d4; |
|
|
1030 | unsigned char *cur = (unsigned char *)dec->cur; |
457 | |
1031 | |
458 | d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; |
1032 | d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected"); |
459 | if (d1 < 0) ERR ("four hexadecimal digits expected"); |
1033 | d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected"); |
460 | d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; |
1034 | d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected"); |
461 | if (d2 < 0) ERR ("four hexadecimal digits expected"); |
1035 | d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected"); |
462 | d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]]; |
|
|
463 | if (d3 < 0) ERR ("four hexadecimal digits expected"); |
|
|
464 | d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]]; |
|
|
465 | if (d4 < 0) ERR ("four hexadecimal digits expected"); |
|
|
466 | |
1036 | |
467 | dec->cur += 4; |
1037 | dec->cur += 4; |
468 | |
1038 | |
469 | return ((UV)d1) << 12 |
1039 | return ((UV)d1) << 12 |
470 | | ((UV)d2) << 8 |
1040 | | ((UV)d2) << 8 |
… | |
… | |
473 | |
1043 | |
474 | fail: |
1044 | fail: |
475 | return (UV)-1; |
1045 | return (UV)-1; |
476 | } |
1046 | } |
477 | |
1047 | |
478 | #define APPEND_GROW(n) SB \ |
|
|
479 | if (cur + (n) >= end) \ |
|
|
480 | { \ |
|
|
481 | STRLEN ofs = cur - SvPVX (sv); \ |
|
|
482 | SvGROW (sv, ofs + (n) + 1); \ |
|
|
483 | cur = SvPVX (sv) + ofs; \ |
|
|
484 | end = SvEND (sv); \ |
|
|
485 | } \ |
|
|
486 | SE |
|
|
487 | |
|
|
488 | #define APPEND_CH(ch) SB \ |
|
|
489 | APPEND_GROW (1); \ |
|
|
490 | *cur++ = (ch); \ |
|
|
491 | SE |
|
|
492 | |
|
|
493 | static SV * |
1048 | static SV * |
494 | decode_str (dec_t *dec) |
1049 | decode_str (dec_t *dec) |
495 | { |
1050 | { |
496 | SV *sv = NEWSV (0,2); |
1051 | SV *sv = 0; |
497 | int utf8 = 0; |
1052 | int utf8 = 0; |
498 | char *cur = SvPVX (sv); |
1053 | char *dec_cur = dec->cur; |
499 | char *end = SvEND (sv); |
|
|
500 | |
1054 | |
501 | for (;;) |
1055 | do |
502 | { |
1056 | { |
503 | unsigned char ch = *(unsigned char *)dec->cur; |
1057 | char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; |
|
|
1058 | char *cur = buf; |
504 | |
1059 | |
505 | if (ch == '"') |
1060 | do |
506 | break; |
|
|
507 | else if (ch == '\\') |
|
|
508 | { |
1061 | { |
509 | switch (*++dec->cur) |
1062 | unsigned char ch = *(unsigned char *)dec_cur++; |
|
|
1063 | |
|
|
1064 | if (expect_false (ch == '"')) |
510 | { |
1065 | { |
511 | case '\\': |
1066 | --dec_cur; |
512 | case '/': |
1067 | break; |
513 | case '"': APPEND_CH (*dec->cur++); break; |
1068 | } |
514 | |
1069 | else if (expect_false (ch == '\\')) |
515 | case 'b': APPEND_CH ('\010'); ++dec->cur; break; |
1070 | { |
516 | case 't': APPEND_CH ('\011'); ++dec->cur; break; |
1071 | switch (*dec_cur) |
517 | case 'n': APPEND_CH ('\012'); ++dec->cur; break; |
|
|
518 | case 'f': APPEND_CH ('\014'); ++dec->cur; break; |
|
|
519 | case 'r': APPEND_CH ('\015'); ++dec->cur; break; |
|
|
520 | |
|
|
521 | case 'u': |
|
|
522 | { |
1072 | { |
523 | UV lo, hi; |
1073 | case '\\': |
524 | ++dec->cur; |
1074 | case '/': |
|
|
1075 | case '"': *cur++ = *dec_cur++; break; |
525 | |
1076 | |
526 | hi = decode_4hex (dec); |
1077 | case 'b': ++dec_cur; *cur++ = '\010'; break; |
527 | if (hi == (UV)-1) |
1078 | case 't': ++dec_cur; *cur++ = '\011'; break; |
528 | goto fail; |
1079 | case 'n': ++dec_cur; *cur++ = '\012'; break; |
|
|
1080 | case 'f': ++dec_cur; *cur++ = '\014'; break; |
|
|
1081 | case 'r': ++dec_cur; *cur++ = '\015'; break; |
529 | |
1082 | |
530 | // possibly a surrogate pair |
1083 | case 'u': |
531 | if (hi >= 0xd800 && hi < 0xdc00) |
|
|
532 | { |
1084 | { |
533 | if (dec->cur [0] != '\\' || dec->cur [1] != 'u') |
1085 | UV lo, hi; |
534 | ERR ("illegal surrogate character"); |
1086 | ++dec_cur; |
535 | |
1087 | |
536 | dec->cur += 2; |
1088 | dec->cur = dec_cur; |
537 | |
|
|
538 | lo = decode_4hex (dec); |
1089 | hi = decode_4hex (dec); |
|
|
1090 | dec_cur = dec->cur; |
539 | if (lo == (UV)-1) |
1091 | if (hi == (UV)-1) |
540 | goto fail; |
1092 | goto fail; |
541 | |
1093 | |
|
|
1094 | // possibly a surrogate pair |
|
|
1095 | if (hi >= 0xd800) |
|
|
1096 | if (hi < 0xdc00) |
|
|
1097 | { |
|
|
1098 | if (dec_cur [0] != '\\' || dec_cur [1] != 'u') |
|
|
1099 | ERR ("missing low surrogate character in surrogate pair"); |
|
|
1100 | |
|
|
1101 | dec_cur += 2; |
|
|
1102 | |
|
|
1103 | dec->cur = dec_cur; |
|
|
1104 | lo = decode_4hex (dec); |
|
|
1105 | dec_cur = dec->cur; |
|
|
1106 | if (lo == (UV)-1) |
|
|
1107 | goto fail; |
|
|
1108 | |
542 | if (lo < 0xdc00 || lo >= 0xe000) |
1109 | if (lo < 0xdc00 || lo >= 0xe000) |
543 | ERR ("surrogate pair expected"); |
1110 | ERR ("surrogate pair expected"); |
544 | |
1111 | |
545 | hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; |
1112 | hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; |
|
|
1113 | } |
|
|
1114 | else if (hi < 0xe000) |
|
|
1115 | ERR ("missing high surrogate character in surrogate pair"); |
|
|
1116 | |
|
|
1117 | if (hi >= 0x80) |
|
|
1118 | { |
|
|
1119 | utf8 = 1; |
|
|
1120 | |
|
|
1121 | cur = encode_utf8 (cur, hi); |
|
|
1122 | } |
|
|
1123 | else |
|
|
1124 | *cur++ = hi; |
546 | } |
1125 | } |
547 | else if (lo >= 0xdc00 && lo < 0xe000) |
|
|
548 | ERR ("illegal surrogate character"); |
|
|
549 | |
|
|
550 | if (hi >= 0x80) |
|
|
551 | { |
1126 | break; |
552 | utf8 = 1; |
|
|
553 | |
1127 | |
554 | APPEND_GROW (4); // at most 4 bytes for 21 bits |
|
|
555 | cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); |
|
|
556 | } |
|
|
557 | else |
1128 | default: |
558 | APPEND_CH (hi); |
1129 | --dec_cur; |
|
|
1130 | ERR ("illegal backslash escape sequence in string"); |
559 | } |
1131 | } |
|
|
1132 | } |
|
|
1133 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
|
|
1134 | *cur++ = ch; |
|
|
1135 | else if (ch >= 0x80) |
|
|
1136 | { |
|
|
1137 | STRLEN clen; |
|
|
1138 | |
|
|
1139 | --dec_cur; |
|
|
1140 | |
|
|
1141 | decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
|
|
1142 | if (clen == (STRLEN)-1) |
|
|
1143 | ERR ("malformed UTF-8 character in JSON string"); |
|
|
1144 | |
|
|
1145 | do |
|
|
1146 | *cur++ = *dec_cur++; |
|
|
1147 | while (--clen); |
|
|
1148 | |
|
|
1149 | utf8 = 1; |
|
|
1150 | } |
|
|
1151 | else if (ch == '\t' && dec->json.flags & F_RELAXED) |
|
|
1152 | *cur++ = ch; |
|
|
1153 | else |
|
|
1154 | { |
|
|
1155 | --dec_cur; |
|
|
1156 | |
|
|
1157 | if (!ch) |
|
|
1158 | ERR ("unexpected end of string while parsing JSON string"); |
560 | break; |
1159 | else |
|
|
1160 | ERR ("invalid character encountered while parsing JSON string"); |
561 | } |
1161 | } |
562 | } |
1162 | } |
563 | else if (ch >= 0x20 && ch <= 0x7f) |
1163 | while (cur < buf + SHORT_STRING_LEN); |
564 | APPEND_CH (*dec->cur++); |
1164 | |
565 | else if (ch >= 0x80) |
1165 | { |
|
|
1166 | STRLEN len = cur - buf; |
|
|
1167 | |
|
|
1168 | if (sv) |
566 | { |
1169 | { |
567 | STRLEN clen; |
1170 | STRLEN cur = SvCUR (sv); |
568 | UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); |
|
|
569 | if (clen < 0) |
|
|
570 | ERR ("malformed UTF-8 character in string, cannot convert to JSON"); |
|
|
571 | |
1171 | |
572 | APPEND_GROW (clen); |
1172 | if (SvLEN (sv) <= cur + len) |
573 | memcpy (cur, dec->cur, clen); |
1173 | SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
574 | cur += clen; |
1174 | |
575 | dec->cur += clen; |
1175 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
|
|
1176 | SvCUR_set (sv, SvCUR (sv) + len); |
576 | } |
1177 | } |
577 | else |
1178 | else |
578 | ERR ("invalid character encountered"); |
1179 | sv = newSVpvn (buf, len); |
579 | } |
1180 | } |
|
|
1181 | } |
|
|
1182 | while (*dec_cur != '"'); |
580 | |
1183 | |
581 | ++dec->cur; |
1184 | ++dec_cur; |
582 | |
1185 | |
583 | SvCUR_set (sv, cur - SvPVX (sv)); |
1186 | if (sv) |
584 | |
1187 | { |
585 | SvPOK_only (sv); |
1188 | SvPOK_only (sv); |
586 | *SvEND (sv) = 0; |
1189 | *SvEND (sv) = 0; |
587 | |
1190 | |
588 | if (utf8) |
1191 | if (utf8) |
589 | SvUTF8_on (sv); |
1192 | SvUTF8_on (sv); |
|
|
1193 | } |
|
|
1194 | else |
|
|
1195 | sv = newSVpvn ("", 0); |
590 | |
1196 | |
|
|
1197 | dec->cur = dec_cur; |
591 | return sv; |
1198 | return sv; |
592 | |
1199 | |
593 | fail: |
1200 | fail: |
594 | SvREFCNT_dec (sv); |
1201 | dec->cur = dec_cur; |
595 | return 0; |
1202 | return 0; |
596 | } |
1203 | } |
597 | |
1204 | |
598 | static SV * |
1205 | static SV * |
599 | decode_num (dec_t *dec) |
1206 | decode_num (dec_t *dec) |
… | |
… | |
609 | { |
1216 | { |
610 | ++dec->cur; |
1217 | ++dec->cur; |
611 | if (*dec->cur >= '0' && *dec->cur <= '9') |
1218 | if (*dec->cur >= '0' && *dec->cur <= '9') |
612 | ERR ("malformed number (leading zero must not be followed by another digit)"); |
1219 | ERR ("malformed number (leading zero must not be followed by another digit)"); |
613 | } |
1220 | } |
614 | |
1221 | else if (*dec->cur < '0' || *dec->cur > '9') |
615 | // int |
1222 | ERR ("malformed number (no digits after initial minus)"); |
|
|
1223 | else |
|
|
1224 | do |
|
|
1225 | { |
|
|
1226 | ++dec->cur; |
|
|
1227 | } |
616 | while (*dec->cur >= '0' && *dec->cur <= '9') |
1228 | while (*dec->cur >= '0' && *dec->cur <= '9'); |
617 | ++dec->cur; |
|
|
618 | |
1229 | |
619 | // [frac] |
1230 | // [frac] |
620 | if (*dec->cur == '.') |
1231 | if (*dec->cur == '.') |
621 | { |
1232 | { |
622 | is_nv = 1; |
1233 | ++dec->cur; |
|
|
1234 | |
|
|
1235 | if (*dec->cur < '0' || *dec->cur > '9') |
|
|
1236 | ERR ("malformed number (no digits after decimal point)"); |
623 | |
1237 | |
624 | do |
1238 | do |
625 | { |
1239 | { |
626 | ++dec->cur; |
1240 | ++dec->cur; |
627 | } |
1241 | } |
628 | while (*dec->cur >= '0' && *dec->cur <= '9'); |
1242 | while (*dec->cur >= '0' && *dec->cur <= '9'); |
|
|
1243 | |
|
|
1244 | is_nv = 1; |
629 | } |
1245 | } |
630 | |
1246 | |
631 | // [exp] |
1247 | // [exp] |
632 | if (*dec->cur == 'e' || *dec->cur == 'E') |
1248 | if (*dec->cur == 'e' || *dec->cur == 'E') |
633 | { |
1249 | { |
634 | is_nv = 1; |
|
|
635 | |
|
|
636 | ++dec->cur; |
1250 | ++dec->cur; |
|
|
1251 | |
637 | if (*dec->cur == '-' || *dec->cur == '+') |
1252 | if (*dec->cur == '-' || *dec->cur == '+') |
638 | ++dec->cur; |
1253 | ++dec->cur; |
639 | |
1254 | |
|
|
1255 | if (*dec->cur < '0' || *dec->cur > '9') |
|
|
1256 | ERR ("malformed number (no digits after exp sign)"); |
|
|
1257 | |
|
|
1258 | do |
|
|
1259 | { |
|
|
1260 | ++dec->cur; |
|
|
1261 | } |
640 | while (*dec->cur >= '0' && *dec->cur <= '9') |
1262 | while (*dec->cur >= '0' && *dec->cur <= '9'); |
641 | ++dec->cur; |
1263 | |
|
|
1264 | is_nv = 1; |
642 | } |
1265 | } |
643 | |
1266 | |
644 | if (!is_nv) |
1267 | if (!is_nv) |
645 | { |
1268 | { |
646 | UV uv; |
1269 | int len = dec->cur - start; |
647 | int numtype = grok_number (start, dec->cur - start, &uv); |
1270 | |
648 | if (numtype & IS_NUMBER_IN_UV) |
1271 | // special case the rather common 1..5-digit-int case |
649 | if (numtype & IS_NUMBER_NEG) |
1272 | if (*start == '-') |
|
|
1273 | switch (len) |
650 | { |
1274 | { |
651 | if (uv < (UV)IV_MIN) |
1275 | case 2: return newSViv (-(IV)( start [1] - '0' * 1)); |
652 | return newSViv (-(IV)uv); |
1276 | case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11)); |
|
|
1277 | case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
|
|
1278 | case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
|
|
1279 | case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
653 | } |
1280 | } |
|
|
1281 | else |
|
|
1282 | switch (len) |
|
|
1283 | { |
|
|
1284 | case 1: return newSViv ( start [0] - '0' * 1); |
|
|
1285 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
|
|
1286 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
|
|
1287 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
|
|
1288 | case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
|
|
1289 | } |
|
|
1290 | |
|
|
1291 | { |
|
|
1292 | UV uv; |
|
|
1293 | int numtype = grok_number (start, len, &uv); |
|
|
1294 | if (numtype & IS_NUMBER_IN_UV) |
|
|
1295 | if (numtype & IS_NUMBER_NEG) |
|
|
1296 | { |
|
|
1297 | if (uv < (UV)IV_MIN) |
|
|
1298 | return newSViv (-(IV)uv); |
|
|
1299 | } |
654 | else |
1300 | else |
655 | return newSVuv (uv); |
1301 | return newSVuv (uv); |
656 | } |
1302 | } |
657 | |
1303 | |
|
|
1304 | len -= *start == '-' ? 1 : 0; |
|
|
1305 | |
|
|
1306 | // does not fit into IV or UV, try NV |
|
|
1307 | if (len <= NV_DIG) |
|
|
1308 | // fits into NV without loss of precision |
|
|
1309 | return newSVnv (json_atof (start)); |
|
|
1310 | |
|
|
1311 | // everything else fails, convert it to a string |
|
|
1312 | return newSVpvn (start, dec->cur - start); |
|
|
1313 | } |
|
|
1314 | |
|
|
1315 | // loss of precision here |
658 | return newSVnv (Atof (start)); |
1316 | return newSVnv (json_atof (start)); |
659 | |
1317 | |
660 | fail: |
1318 | fail: |
661 | return 0; |
1319 | return 0; |
662 | } |
1320 | } |
663 | |
1321 | |
664 | static SV * |
1322 | static SV * |
665 | decode_av (dec_t *dec) |
1323 | decode_av (dec_t *dec) |
666 | { |
1324 | { |
667 | AV *av = newAV (); |
1325 | AV *av = newAV (); |
668 | |
1326 | |
|
|
1327 | DEC_INC_DEPTH; |
|
|
1328 | decode_ws (dec); |
|
|
1329 | |
|
|
1330 | if (*dec->cur == ']') |
|
|
1331 | ++dec->cur; |
|
|
1332 | else |
669 | for (;;) |
1333 | for (;;) |
670 | { |
1334 | { |
671 | SV *value; |
1335 | SV *value; |
672 | |
1336 | |
673 | value = decode_sv (dec); |
1337 | value = decode_sv (dec); |
674 | if (!value) |
1338 | if (!value) |
675 | goto fail; |
1339 | goto fail; |
676 | |
1340 | |
677 | av_push (av, value); |
1341 | av_push (av, value); |
678 | |
1342 | |
679 | WS; |
1343 | decode_ws (dec); |
680 | |
1344 | |
681 | if (*dec->cur == ']') |
1345 | if (*dec->cur == ']') |
682 | { |
1346 | { |
683 | ++dec->cur; |
1347 | ++dec->cur; |
684 | break; |
1348 | break; |
685 | } |
1349 | } |
686 | |
1350 | |
687 | if (*dec->cur != ',') |
1351 | if (*dec->cur != ',') |
688 | ERR (", or ] expected while parsing array"); |
1352 | ERR (", or ] expected while parsing array"); |
689 | |
1353 | |
690 | ++dec->cur; |
1354 | ++dec->cur; |
|
|
1355 | |
|
|
1356 | decode_ws (dec); |
|
|
1357 | |
|
|
1358 | if (*dec->cur == ']' && dec->json.flags & F_RELAXED) |
|
|
1359 | { |
|
|
1360 | ++dec->cur; |
|
|
1361 | break; |
|
|
1362 | } |
691 | } |
1363 | } |
692 | |
1364 | |
|
|
1365 | DEC_DEC_DEPTH; |
693 | return newRV_noinc ((SV *)av); |
1366 | return newRV_noinc ((SV *)av); |
694 | |
1367 | |
695 | fail: |
1368 | fail: |
696 | SvREFCNT_dec (av); |
1369 | SvREFCNT_dec (av); |
|
|
1370 | DEC_DEC_DEPTH; |
697 | return 0; |
1371 | return 0; |
698 | } |
1372 | } |
699 | |
1373 | |
700 | static SV * |
1374 | static SV * |
701 | decode_hv (dec_t *dec) |
1375 | decode_hv (dec_t *dec) |
702 | { |
1376 | { |
|
|
1377 | SV *sv; |
703 | HV *hv = newHV (); |
1378 | HV *hv = newHV (); |
704 | |
1379 | |
|
|
1380 | DEC_INC_DEPTH; |
|
|
1381 | decode_ws (dec); |
|
|
1382 | |
|
|
1383 | if (*dec->cur == '}') |
|
|
1384 | ++dec->cur; |
|
|
1385 | else |
705 | for (;;) |
1386 | for (;;) |
706 | { |
1387 | { |
707 | SV *key, *value; |
|
|
708 | |
|
|
709 | WS; EXPECT_CH ('"'); |
1388 | EXPECT_CH ('"'); |
710 | |
1389 | |
711 | key = decode_str (dec); |
1390 | // heuristic: assume that |
712 | if (!key) |
1391 | // a) decode_str + hv_store_ent are abysmally slow. |
713 | goto fail; |
1392 | // b) most hash keys are short, simple ascii text. |
714 | |
1393 | // => try to "fast-match" such strings to avoid |
715 | WS; EXPECT_CH (':'); |
1394 | // the overhead of decode_str + hv_store_ent. |
716 | |
|
|
717 | value = decode_sv (dec); |
|
|
718 | if (!value) |
|
|
719 | { |
1395 | { |
|
|
1396 | SV *value; |
|
|
1397 | char *p = dec->cur; |
|
|
1398 | char *e = p + 24; // only try up to 24 bytes |
|
|
1399 | |
|
|
1400 | for (;;) |
|
|
1401 | { |
|
|
1402 | // the >= 0x80 is false on most architectures |
|
|
1403 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
|
|
1404 | { |
|
|
1405 | // slow path, back up and use decode_str |
|
|
1406 | SV *key = decode_str (dec); |
|
|
1407 | if (!key) |
|
|
1408 | goto fail; |
|
|
1409 | |
|
|
1410 | decode_ws (dec); EXPECT_CH (':'); |
|
|
1411 | |
|
|
1412 | decode_ws (dec); |
|
|
1413 | value = decode_sv (dec); |
|
|
1414 | if (!value) |
|
|
1415 | { |
|
|
1416 | SvREFCNT_dec (key); |
|
|
1417 | goto fail; |
|
|
1418 | } |
|
|
1419 | |
|
|
1420 | hv_store_ent (hv, key, value, 0); |
720 | SvREFCNT_dec (key); |
1421 | SvREFCNT_dec (key); |
|
|
1422 | |
|
|
1423 | break; |
|
|
1424 | } |
|
|
1425 | else if (*p == '"') |
|
|
1426 | { |
|
|
1427 | // fast path, got a simple key |
|
|
1428 | char *key = dec->cur; |
|
|
1429 | int len = p - key; |
|
|
1430 | dec->cur = p + 1; |
|
|
1431 | |
|
|
1432 | decode_ws (dec); EXPECT_CH (':'); |
|
|
1433 | |
|
|
1434 | decode_ws (dec); |
|
|
1435 | value = decode_sv (dec); |
|
|
1436 | if (!value) |
721 | goto fail; |
1437 | goto fail; |
|
|
1438 | |
|
|
1439 | hv_store (hv, key, len, value, 0); |
|
|
1440 | |
|
|
1441 | break; |
|
|
1442 | } |
|
|
1443 | |
|
|
1444 | ++p; |
|
|
1445 | } |
722 | } |
1446 | } |
723 | |
1447 | |
724 | //TODO: optimise |
1448 | decode_ws (dec); |
725 | hv_store_ent (hv, key, value, 0); |
|
|
726 | |
1449 | |
727 | WS; |
|
|
728 | |
|
|
729 | if (*dec->cur == '}') |
1450 | if (*dec->cur == '}') |
|
|
1451 | { |
|
|
1452 | ++dec->cur; |
|
|
1453 | break; |
|
|
1454 | } |
|
|
1455 | |
|
|
1456 | if (*dec->cur != ',') |
|
|
1457 | ERR (", or } expected while parsing object/hash"); |
|
|
1458 | |
|
|
1459 | ++dec->cur; |
|
|
1460 | |
|
|
1461 | decode_ws (dec); |
|
|
1462 | |
|
|
1463 | if (*dec->cur == '}' && dec->json.flags & F_RELAXED) |
|
|
1464 | { |
|
|
1465 | ++dec->cur; |
|
|
1466 | break; |
|
|
1467 | } |
|
|
1468 | } |
|
|
1469 | |
|
|
1470 | DEC_DEC_DEPTH; |
|
|
1471 | sv = newRV_noinc ((SV *)hv); |
|
|
1472 | |
|
|
1473 | // check filter callbacks |
|
|
1474 | if (dec->json.flags & F_HOOK) |
|
|
1475 | { |
|
|
1476 | if (dec->json.cb_sk_object && HvKEYS (hv) == 1) |
730 | { |
1477 | { |
731 | ++dec->cur; |
1478 | HE *cb, *he; |
732 | break; |
1479 | |
|
|
1480 | hv_iterinit (hv); |
|
|
1481 | he = hv_iternext (hv); |
|
|
1482 | hv_iterinit (hv); |
|
|
1483 | |
|
|
1484 | // the next line creates a mortal sv each time it's called. |
|
|
1485 | // might want to optimise this for common cases. |
|
|
1486 | cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); |
|
|
1487 | |
|
|
1488 | if (cb) |
|
|
1489 | { |
|
|
1490 | dSP; |
|
|
1491 | int count; |
|
|
1492 | |
|
|
1493 | ENTER; SAVETMPS; |
|
|
1494 | SAVESTACK_POS (); |
|
|
1495 | PUSHMARK (SP); |
|
|
1496 | XPUSHs (HeVAL (he)); |
|
|
1497 | sv_2mortal (sv); |
|
|
1498 | |
|
|
1499 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
|
|
1500 | |
|
|
1501 | if (count == 1) |
|
|
1502 | { |
|
|
1503 | sv = newSVsv (POPs); |
|
|
1504 | FREETMPS; LEAVE; |
|
|
1505 | return sv; |
|
|
1506 | } |
|
|
1507 | |
|
|
1508 | SvREFCNT_inc (sv); |
|
|
1509 | FREETMPS; LEAVE; |
|
|
1510 | } |
733 | } |
1511 | } |
734 | |
1512 | |
735 | if (*dec->cur != ',') |
1513 | if (dec->json.cb_object) |
736 | ERR (", or } expected while parsing object/hash"); |
1514 | { |
|
|
1515 | dSP; |
|
|
1516 | int count; |
737 | |
1517 | |
738 | ++dec->cur; |
1518 | ENTER; SAVETMPS; |
739 | } |
1519 | SAVESTACK_POS (); |
|
|
1520 | PUSHMARK (SP); |
|
|
1521 | XPUSHs (sv_2mortal (sv)); |
740 | |
1522 | |
741 | return newRV_noinc ((SV *)hv); |
1523 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
|
|
1524 | |
|
|
1525 | if (count == 1) |
|
|
1526 | { |
|
|
1527 | sv = newSVsv (POPs); |
|
|
1528 | FREETMPS; LEAVE; |
|
|
1529 | return sv; |
|
|
1530 | } |
|
|
1531 | |
|
|
1532 | SvREFCNT_inc (sv); |
|
|
1533 | FREETMPS; LEAVE; |
|
|
1534 | } |
|
|
1535 | } |
|
|
1536 | |
|
|
1537 | return sv; |
742 | |
1538 | |
743 | fail: |
1539 | fail: |
744 | SvREFCNT_dec (hv); |
1540 | SvREFCNT_dec (hv); |
|
|
1541 | DEC_DEC_DEPTH; |
|
|
1542 | return 0; |
|
|
1543 | } |
|
|
1544 | |
|
|
1545 | static SV * |
|
|
1546 | decode_tag (dec_t *dec) |
|
|
1547 | { |
|
|
1548 | SV *tag = 0; |
|
|
1549 | SV *val = 0; |
|
|
1550 | |
|
|
1551 | if (!(dec->json.flags & F_ALLOW_TAGS)) |
|
|
1552 | ERR ("malformed JSON string, neither array, object, number, string or atom"); |
|
|
1553 | |
|
|
1554 | ++dec->cur; |
|
|
1555 | |
|
|
1556 | decode_ws (dec); |
|
|
1557 | |
|
|
1558 | tag = decode_sv (dec); |
|
|
1559 | if (!tag) |
|
|
1560 | goto fail; |
|
|
1561 | |
|
|
1562 | if (!SvPOK (tag)) |
|
|
1563 | ERR ("malformed JSON string, (tag) must be a string"); |
|
|
1564 | |
|
|
1565 | decode_ws (dec); |
|
|
1566 | |
|
|
1567 | if (*dec->cur != ')') |
|
|
1568 | ERR (") expected after tag"); |
|
|
1569 | |
|
|
1570 | ++dec->cur; |
|
|
1571 | |
|
|
1572 | decode_ws (dec); |
|
|
1573 | |
|
|
1574 | val = decode_sv (dec); |
|
|
1575 | if (!val) |
|
|
1576 | goto fail; |
|
|
1577 | |
|
|
1578 | if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV) |
|
|
1579 | ERR ("malformed JSON string, tag value must be an array"); |
|
|
1580 | |
|
|
1581 | { |
|
|
1582 | AV *av = (AV *)SvRV (val); |
|
|
1583 | int i, len = av_len (av) + 1; |
|
|
1584 | HV *stash = gv_stashsv (tag, 0); |
|
|
1585 | SV *sv; |
|
|
1586 | |
|
|
1587 | if (!stash) |
|
|
1588 | ERR ("cannot decode perl-object (package does not exist)"); |
|
|
1589 | |
|
|
1590 | GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); |
|
|
1591 | |
|
|
1592 | if (!method) |
|
|
1593 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
|
|
1594 | |
|
|
1595 | dSP; |
|
|
1596 | |
|
|
1597 | ENTER; SAVETMPS; |
|
|
1598 | PUSHMARK (SP); |
|
|
1599 | EXTEND (SP, len + 2); |
|
|
1600 | // we re-bless the reference to get overload and other niceties right |
|
|
1601 | PUSHs (tag); |
|
|
1602 | PUSHs (sv_json); |
|
|
1603 | |
|
|
1604 | for (i = 0; i < len; ++i) |
|
|
1605 | PUSHs (*av_fetch (av, i, 1)); |
|
|
1606 | |
|
|
1607 | PUTBACK; |
|
|
1608 | call_sv ((SV *)GvCV (method), G_SCALAR); |
|
|
1609 | SPAGAIN; |
|
|
1610 | |
|
|
1611 | SvREFCNT_dec (tag); |
|
|
1612 | SvREFCNT_dec (val); |
|
|
1613 | sv = SvREFCNT_inc (POPs); |
|
|
1614 | |
|
|
1615 | PUTBACK; |
|
|
1616 | |
|
|
1617 | FREETMPS; LEAVE; |
|
|
1618 | |
|
|
1619 | return sv; |
|
|
1620 | } |
|
|
1621 | |
|
|
1622 | fail: |
|
|
1623 | SvREFCNT_dec (tag); |
|
|
1624 | SvREFCNT_dec (val); |
745 | return 0; |
1625 | return 0; |
746 | } |
1626 | } |
747 | |
1627 | |
748 | static SV * |
1628 | static SV * |
749 | decode_sv (dec_t *dec) |
1629 | decode_sv (dec_t *dec) |
750 | { |
1630 | { |
751 | WS; |
1631 | // the beauty of JSON: you need exactly one character lookahead |
|
|
1632 | // to parse everything. |
752 | switch (*dec->cur) |
1633 | switch (*dec->cur) |
753 | { |
1634 | { |
754 | case '"': ++dec->cur; return decode_str (dec); |
1635 | case '"': ++dec->cur; return decode_str (dec); |
755 | case '[': ++dec->cur; return decode_av (dec); |
1636 | case '[': ++dec->cur; return decode_av (dec); |
756 | case '{': ++dec->cur; return decode_hv (dec); |
1637 | case '{': ++dec->cur; return decode_hv (dec); |
|
|
1638 | case '(': return decode_tag (dec); |
757 | |
1639 | |
758 | case '-': |
1640 | case '-': |
759 | case '0': case '1': case '2': case '3': case '4': |
1641 | case '0': case '1': case '2': case '3': case '4': |
760 | case '5': case '6': case '7': case '8': case '9': |
1642 | case '5': case '6': case '7': case '8': case '9': |
761 | return decode_num (dec); |
1643 | return decode_num (dec); |
762 | |
1644 | |
763 | case 't': |
1645 | case 't': |
764 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1646 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
765 | { |
1647 | { |
766 | dec->cur += 4; |
1648 | dec->cur += 4; |
|
|
1649 | #if JSON_SLOW |
|
|
1650 | bool_true = get_bool ("Types::Serialiser::true"); |
|
|
1651 | #endif |
767 | return newSViv (1); |
1652 | return newSVsv (bool_true); |
768 | } |
1653 | } |
769 | else |
1654 | else |
770 | ERR ("'true' expected"); |
1655 | ERR ("'true' expected"); |
771 | |
1656 | |
772 | break; |
1657 | break; |
773 | |
1658 | |
774 | case 'f': |
1659 | case 'f': |
775 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1660 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
776 | { |
1661 | { |
777 | dec->cur += 5; |
1662 | dec->cur += 5; |
|
|
1663 | #if JSON_SLOW |
|
|
1664 | bool_false = get_bool ("Types::Serialiser::false"); |
|
|
1665 | #endif |
778 | return newSViv (0); |
1666 | return newSVsv (bool_false); |
779 | } |
1667 | } |
780 | else |
1668 | else |
781 | ERR ("'false' expected"); |
1669 | ERR ("'false' expected"); |
782 | |
1670 | |
783 | break; |
1671 | break; |
784 | |
1672 | |
785 | case 'n': |
1673 | case 'n': |
786 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
1674 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
787 | { |
1675 | { |
788 | dec->cur += 4; |
1676 | dec->cur += 4; |
789 | return newSViv (1); |
1677 | return newSVsv (&PL_sv_undef); |
790 | } |
1678 | } |
791 | else |
1679 | else |
792 | ERR ("'null' expected"); |
1680 | ERR ("'null' expected"); |
793 | |
1681 | |
794 | break; |
1682 | break; |
795 | |
1683 | |
796 | default: |
1684 | default: |
797 | ERR ("malformed json string"); |
1685 | ERR ("malformed JSON string, neither tag, array, object, number, string or atom"); |
798 | break; |
1686 | break; |
799 | } |
1687 | } |
800 | |
1688 | |
801 | fail: |
1689 | fail: |
802 | return 0; |
1690 | return 0; |
803 | } |
1691 | } |
804 | |
1692 | |
805 | static SV * |
1693 | static SV * |
806 | decode_json (SV *string, UV flags) |
1694 | decode_json (SV *string, JSON *json, char **offset_return) |
807 | { |
1695 | { |
|
|
1696 | dec_t dec; |
808 | SV *sv; |
1697 | SV *sv; |
809 | |
1698 | |
810 | if (!(flags & F_UTF8)) |
1699 | /* work around bugs in 5.10 where manipulating magic values |
|
|
1700 | * makes perl ignore the magic in subsequent accesses. |
|
|
1701 | * also make a copy of non-PV values, to get them into a clean |
|
|
1702 | * state (SvPV should do that, but it's buggy, see below). |
|
|
1703 | * |
|
|
1704 | * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16), |
|
|
1705 | * as reported by Reini Urban. |
|
|
1706 | */ |
|
|
1707 | /*SvGETMAGIC (string);*/ |
|
|
1708 | if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string)) |
|
|
1709 | string = sv_2mortal (newSVsv (string)); |
|
|
1710 | |
|
|
1711 | SvUPGRADE (string, SVt_PV); |
|
|
1712 | |
|
|
1713 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
|
|
1714 | * assertion with -DDEBUGGING, although SvCUR is documented to |
|
|
1715 | * return the xpv_cur field which certainly exists after upgrading. |
|
|
1716 | * according to nicholas clark, calling SvPOK fixes this. |
|
|
1717 | * But it doesn't fix it, so try another workaround, call SvPV_nolen |
|
|
1718 | * and hope for the best. |
|
|
1719 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
|
|
1720 | * assertion business is seriously broken, try yet another workaround |
|
|
1721 | * for the broken -DDEBUGGING. |
|
|
1722 | */ |
|
|
1723 | { |
|
|
1724 | #ifdef DEBUGGING |
|
|
1725 | STRLEN offset = SvOK (string) ? sv_len (string) : 0; |
|
|
1726 | #else |
|
|
1727 | STRLEN offset = SvCUR (string); |
|
|
1728 | #endif |
|
|
1729 | |
|
|
1730 | if (offset > json->max_size && json->max_size) |
|
|
1731 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
|
|
1732 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
|
|
1733 | } |
|
|
1734 | |
|
|
1735 | if (DECODE_WANTS_OCTETS (json)) |
|
|
1736 | sv_utf8_downgrade (string, 0); |
|
|
1737 | else |
811 | sv_utf8_upgrade (string); |
1738 | sv_utf8_upgrade (string); |
812 | |
1739 | |
813 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1740 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
814 | |
1741 | |
815 | dec_t dec; |
1742 | dec.json = *json; |
816 | dec.flags = flags; |
|
|
817 | dec.cur = SvPVX (string); |
1743 | dec.cur = SvPVX (string); |
818 | dec.end = SvEND (string); |
1744 | dec.end = SvEND (string); |
819 | dec.err = 0; |
1745 | dec.err = 0; |
|
|
1746 | dec.depth = 0; |
820 | |
1747 | |
821 | *dec.end = 1; // invalid anywhere |
1748 | if (dec.json.cb_object || dec.json.cb_sk_object) |
|
|
1749 | dec.json.flags |= F_HOOK; |
|
|
1750 | |
|
|
1751 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
|
|
1752 | |
|
|
1753 | decode_ws (&dec); |
822 | sv = decode_sv (&dec); |
1754 | sv = decode_sv (&dec); |
823 | *dec.end = 0; |
1755 | |
|
|
1756 | if (offset_return) |
|
|
1757 | *offset_return = dec.cur; |
|
|
1758 | |
|
|
1759 | if (!(offset_return || !sv)) |
|
|
1760 | { |
|
|
1761 | // check for trailing garbage |
|
|
1762 | decode_ws (&dec); |
|
|
1763 | |
|
|
1764 | if (*dec.cur) |
|
|
1765 | { |
|
|
1766 | dec.err = "garbage after JSON object"; |
|
|
1767 | SvREFCNT_dec (sv); |
|
|
1768 | sv = 0; |
|
|
1769 | } |
|
|
1770 | } |
824 | |
1771 | |
825 | if (!sv) |
1772 | if (!sv) |
826 | { |
1773 | { |
827 | IV offset = utf8_distance (dec.cur, SvPVX (string)); |
|
|
828 | SV *uni = sv_newmortal (); |
1774 | SV *uni = sv_newmortal (); |
829 | |
1775 | |
|
|
1776 | // horrible hack to silence warning inside pv_uni_display |
|
|
1777 | COP cop = *PL_curcop; |
|
|
1778 | cop.cop_warnings = pWARN_NONE; |
|
|
1779 | ENTER; |
|
|
1780 | SAVEVPTR (PL_curcop); |
|
|
1781 | PL_curcop = &cop; |
830 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1782 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
|
|
1783 | LEAVE; |
|
|
1784 | |
831 | croak ("%s, at character %d (%s)", |
1785 | croak ("%s, at character offset %d (before \"%s\")", |
832 | dec.err, |
1786 | dec.err, |
833 | (int)offset, |
1787 | (int)ptr_to_index (string, dec.cur), |
834 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1788 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
835 | } |
1789 | } |
836 | |
1790 | |
837 | sv = sv_2mortal (sv); |
1791 | sv = sv_2mortal (sv); |
838 | |
1792 | |
839 | if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1793 | if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv)) |
840 | croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)"); |
1794 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
841 | |
1795 | |
842 | return sv; |
1796 | return sv; |
843 | } |
1797 | } |
844 | |
1798 | |
|
|
1799 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1800 | // incremental parser |
|
|
1801 | |
|
|
1802 | static void |
|
|
1803 | incr_parse (JSON *self) |
|
|
1804 | { |
|
|
1805 | const char *p = SvPVX (self->incr_text) + self->incr_pos; |
|
|
1806 | |
|
|
1807 | // the state machine here is a bit convoluted and could be simplified a lot |
|
|
1808 | // but this would make it slower, so... |
|
|
1809 | |
|
|
1810 | for (;;) |
|
|
1811 | { |
|
|
1812 | //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 |
|
|
1813 | switch (self->incr_mode) |
|
|
1814 | { |
|
|
1815 | // only used for initial whitespace skipping |
|
|
1816 | case INCR_M_WS: |
|
|
1817 | for (;;) |
|
|
1818 | { |
|
|
1819 | if (*p > 0x20) |
|
|
1820 | { |
|
|
1821 | if (*p == '#') |
|
|
1822 | { |
|
|
1823 | self->incr_mode = INCR_M_C0; |
|
|
1824 | goto incr_m_c; |
|
|
1825 | } |
|
|
1826 | else |
|
|
1827 | { |
|
|
1828 | self->incr_mode = INCR_M_JSON; |
|
|
1829 | goto incr_m_json; |
|
|
1830 | } |
|
|
1831 | } |
|
|
1832 | else if (!*p) |
|
|
1833 | goto interrupt; |
|
|
1834 | |
|
|
1835 | ++p; |
|
|
1836 | } |
|
|
1837 | |
|
|
1838 | // skip a single char inside a string (for \\-processing) |
|
|
1839 | case INCR_M_BS: |
|
|
1840 | if (!*p) |
|
|
1841 | goto interrupt; |
|
|
1842 | |
|
|
1843 | ++p; |
|
|
1844 | self->incr_mode = INCR_M_STR; |
|
|
1845 | goto incr_m_str; |
|
|
1846 | |
|
|
1847 | // inside #-style comments |
|
|
1848 | case INCR_M_C0: |
|
|
1849 | case INCR_M_C1: |
|
|
1850 | incr_m_c: |
|
|
1851 | for (;;) |
|
|
1852 | { |
|
|
1853 | if (*p == '\n') |
|
|
1854 | { |
|
|
1855 | self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON; |
|
|
1856 | break; |
|
|
1857 | } |
|
|
1858 | else if (!*p) |
|
|
1859 | goto interrupt; |
|
|
1860 | |
|
|
1861 | ++p; |
|
|
1862 | } |
|
|
1863 | |
|
|
1864 | break; |
|
|
1865 | |
|
|
1866 | // inside a string |
|
|
1867 | case INCR_M_STR: |
|
|
1868 | incr_m_str: |
|
|
1869 | for (;;) |
|
|
1870 | { |
|
|
1871 | if (*p == '"') |
|
|
1872 | { |
|
|
1873 | ++p; |
|
|
1874 | self->incr_mode = INCR_M_JSON; |
|
|
1875 | |
|
|
1876 | if (!self->incr_nest) |
|
|
1877 | goto interrupt; |
|
|
1878 | |
|
|
1879 | goto incr_m_json; |
|
|
1880 | } |
|
|
1881 | else if (*p == '\\') |
|
|
1882 | { |
|
|
1883 | ++p; // "virtually" consumes character after \ |
|
|
1884 | |
|
|
1885 | if (!*p) // if at end of string we have to switch modes |
|
|
1886 | { |
|
|
1887 | self->incr_mode = INCR_M_BS; |
|
|
1888 | goto interrupt; |
|
|
1889 | } |
|
|
1890 | } |
|
|
1891 | else if (!*p) |
|
|
1892 | goto interrupt; |
|
|
1893 | |
|
|
1894 | ++p; |
|
|
1895 | } |
|
|
1896 | |
|
|
1897 | // after initial ws, outside string |
|
|
1898 | case INCR_M_JSON: |
|
|
1899 | incr_m_json: |
|
|
1900 | for (;;) |
|
|
1901 | { |
|
|
1902 | switch (*p++) |
|
|
1903 | { |
|
|
1904 | case 0: |
|
|
1905 | --p; |
|
|
1906 | goto interrupt; |
|
|
1907 | |
|
|
1908 | case 0x09: |
|
|
1909 | case 0x0a: |
|
|
1910 | case 0x0d: |
|
|
1911 | case 0x20: |
|
|
1912 | if (!self->incr_nest) |
|
|
1913 | { |
|
|
1914 | --p; // do not eat the whitespace, let the next round do it |
|
|
1915 | goto interrupt; |
|
|
1916 | } |
|
|
1917 | break; |
|
|
1918 | |
|
|
1919 | case '"': |
|
|
1920 | self->incr_mode = INCR_M_STR; |
|
|
1921 | goto incr_m_str; |
|
|
1922 | |
|
|
1923 | case '[': |
|
|
1924 | case '{': |
|
|
1925 | case '(': |
|
|
1926 | if (++self->incr_nest > self->max_depth) |
|
|
1927 | croak (ERR_NESTING_EXCEEDED); |
|
|
1928 | break; |
|
|
1929 | |
|
|
1930 | case ']': |
|
|
1931 | case '}': |
|
|
1932 | if (--self->incr_nest <= 0) |
|
|
1933 | goto interrupt; |
|
|
1934 | break; |
|
|
1935 | |
|
|
1936 | case ')': |
|
|
1937 | --self->incr_nest; |
|
|
1938 | break; |
|
|
1939 | |
|
|
1940 | case '#': |
|
|
1941 | self->incr_mode = INCR_M_C1; |
|
|
1942 | goto incr_m_c; |
|
|
1943 | } |
|
|
1944 | } |
|
|
1945 | } |
|
|
1946 | |
|
|
1947 | modechange: |
|
|
1948 | ; |
|
|
1949 | } |
|
|
1950 | |
|
|
1951 | interrupt: |
|
|
1952 | self->incr_pos = p - SvPVX (self->incr_text); |
|
|
1953 | //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D |
|
|
1954 | //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D |
|
|
1955 | } |
|
|
1956 | |
|
|
1957 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1958 | // XS interface functions |
|
|
1959 | |
845 | MODULE = JSON::XS PACKAGE = JSON::XS |
1960 | MODULE = JSON::XS PACKAGE = JSON::XS |
846 | |
1961 | |
847 | BOOT: |
1962 | BOOT: |
848 | { |
1963 | { |
849 | int i; |
1964 | int i; |
850 | |
1965 | |
851 | memset (decode_hexdigit, 0xff, 256); |
|
|
852 | for (i = 10; i--; ) |
1966 | for (i = 0; i < 256; ++i) |
853 | decode_hexdigit ['0' + i] = i; |
1967 | decode_hexdigit [i] = |
|
|
1968 | i >= '0' && i <= '9' ? i - '0' |
|
|
1969 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
|
|
1970 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
|
|
1971 | : -1; |
854 | |
1972 | |
855 | for (i = 7; i--; ) |
|
|
856 | { |
|
|
857 | decode_hexdigit ['a' + i] = 10 + i; |
|
|
858 | decode_hexdigit ['A' + i] = 10 + i; |
|
|
859 | } |
|
|
860 | |
|
|
861 | json_stash = gv_stashpv ("JSON::XS", 1); |
1973 | json_stash = gv_stashpv ("JSON::XS" , 1); |
|
|
1974 | bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); |
|
|
1975 | bool_true = get_bool ("Types::Serialiser::true"); |
|
|
1976 | bool_false = get_bool ("Types::Serialiser::false"); |
|
|
1977 | |
|
|
1978 | sv_json = newSVpv ("JSON", 0); |
|
|
1979 | SvREADONLY_on (sv_json); |
|
|
1980 | |
|
|
1981 | CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ |
862 | } |
1982 | } |
863 | |
1983 | |
864 | PROTOTYPES: DISABLE |
1984 | PROTOTYPES: DISABLE |
865 | |
1985 | |
866 | SV *new (char *dummy) |
1986 | void CLONE (...) |
867 | CODE: |
1987 | CODE: |
868 | RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); |
1988 | json_stash = 0; |
|
|
1989 | bool_stash = 0; |
|
|
1990 | |
|
|
1991 | void new (char *klass) |
|
|
1992 | PPCODE: |
|
|
1993 | { |
|
|
1994 | SV *pv = NEWSV (0, sizeof (JSON)); |
|
|
1995 | SvPOK_only (pv); |
|
|
1996 | json_init ((JSON *)SvPVX (pv)); |
|
|
1997 | XPUSHs (sv_2mortal (sv_bless ( |
|
|
1998 | newRV_noinc (pv), |
|
|
1999 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
|
|
2000 | ))); |
|
|
2001 | } |
|
|
2002 | |
|
|
2003 | void ascii (JSON *self, int enable = 1) |
|
|
2004 | ALIAS: |
|
|
2005 | ascii = F_ASCII |
|
|
2006 | latin1 = F_LATIN1 |
|
|
2007 | utf8 = F_UTF8 |
|
|
2008 | indent = F_INDENT |
|
|
2009 | canonical = F_CANONICAL |
|
|
2010 | space_before = F_SPACE_BEFORE |
|
|
2011 | space_after = F_SPACE_AFTER |
|
|
2012 | pretty = F_PRETTY |
|
|
2013 | allow_nonref = F_ALLOW_NONREF |
|
|
2014 | shrink = F_SHRINK |
|
|
2015 | allow_blessed = F_ALLOW_BLESSED |
|
|
2016 | convert_blessed = F_CONV_BLESSED |
|
|
2017 | relaxed = F_RELAXED |
|
|
2018 | allow_unknown = F_ALLOW_UNKNOWN |
|
|
2019 | allow_tags = F_ALLOW_TAGS |
|
|
2020 | PPCODE: |
|
|
2021 | { |
|
|
2022 | if (enable) |
|
|
2023 | self->flags |= ix; |
|
|
2024 | else |
|
|
2025 | self->flags &= ~ix; |
|
|
2026 | |
|
|
2027 | XPUSHs (ST (0)); |
|
|
2028 | } |
|
|
2029 | |
|
|
2030 | void get_ascii (JSON *self) |
|
|
2031 | ALIAS: |
|
|
2032 | get_ascii = F_ASCII |
|
|
2033 | get_latin1 = F_LATIN1 |
|
|
2034 | get_utf8 = F_UTF8 |
|
|
2035 | get_indent = F_INDENT |
|
|
2036 | get_canonical = F_CANONICAL |
|
|
2037 | get_space_before = F_SPACE_BEFORE |
|
|
2038 | get_space_after = F_SPACE_AFTER |
|
|
2039 | get_allow_nonref = F_ALLOW_NONREF |
|
|
2040 | get_shrink = F_SHRINK |
|
|
2041 | get_allow_blessed = F_ALLOW_BLESSED |
|
|
2042 | get_convert_blessed = F_CONV_BLESSED |
|
|
2043 | get_relaxed = F_RELAXED |
|
|
2044 | get_allow_unknown = F_ALLOW_UNKNOWN |
|
|
2045 | get_allow_tags = F_ALLOW_TAGS |
|
|
2046 | PPCODE: |
|
|
2047 | XPUSHs (boolSV (self->flags & ix)); |
|
|
2048 | |
|
|
2049 | void max_depth (JSON *self, U32 max_depth = 0x80000000UL) |
|
|
2050 | PPCODE: |
|
|
2051 | self->max_depth = max_depth; |
|
|
2052 | XPUSHs (ST (0)); |
|
|
2053 | |
|
|
2054 | U32 get_max_depth (JSON *self) |
|
|
2055 | CODE: |
|
|
2056 | RETVAL = self->max_depth; |
869 | OUTPUT: |
2057 | OUTPUT: |
870 | RETVAL |
2058 | RETVAL |
871 | |
2059 | |
872 | SV *ascii (SV *self, int enable) |
2060 | void max_size (JSON *self, U32 max_size = 0) |
873 | ALIAS: |
2061 | PPCODE: |
874 | ascii = F_ASCII |
2062 | self->max_size = max_size; |
875 | utf8 = F_UTF8 |
2063 | XPUSHs (ST (0)); |
876 | indent = F_INDENT |
2064 | |
877 | canonical = F_CANONICAL |
2065 | int get_max_size (JSON *self) |
878 | space_before = F_SPACE_BEFORE |
|
|
879 | space_after = F_SPACE_AFTER |
|
|
880 | json_rpc = F_JSON_RPC |
|
|
881 | pretty = F_PRETTY |
|
|
882 | allow_nonref = F_ALLOW_NONREF |
|
|
883 | CODE: |
2066 | CODE: |
884 | { |
2067 | RETVAL = self->max_size; |
885 | UV *uv = SvJSON (self); |
|
|
886 | if (enable) |
|
|
887 | *uv |= ix; |
|
|
888 | else |
|
|
889 | *uv &= ~ix; |
|
|
890 | |
|
|
891 | RETVAL = newSVsv (self); |
|
|
892 | } |
|
|
893 | OUTPUT: |
2068 | OUTPUT: |
894 | RETVAL |
2069 | RETVAL |
895 | |
2070 | |
896 | void encode (SV *self, SV *scalar) |
2071 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
897 | PPCODE: |
2072 | PPCODE: |
898 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
2073 | { |
|
|
2074 | SvREFCNT_dec (self->cb_object); |
|
|
2075 | self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; |
899 | |
2076 | |
900 | void decode (SV *self, SV *jsonstr) |
2077 | XPUSHs (ST (0)); |
|
|
2078 | } |
|
|
2079 | |
|
|
2080 | void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) |
901 | PPCODE: |
2081 | PPCODE: |
902 | XPUSHs (decode_json (jsonstr, *SvJSON (self))); |
2082 | { |
|
|
2083 | if (!self->cb_sk_object) |
|
|
2084 | self->cb_sk_object = newHV (); |
|
|
2085 | |
|
|
2086 | if (SvOK (cb)) |
|
|
2087 | hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
|
|
2088 | else |
|
|
2089 | { |
|
|
2090 | hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0); |
|
|
2091 | |
|
|
2092 | if (!HvKEYS (self->cb_sk_object)) |
|
|
2093 | { |
|
|
2094 | SvREFCNT_dec (self->cb_sk_object); |
|
|
2095 | self->cb_sk_object = 0; |
|
|
2096 | } |
|
|
2097 | } |
|
|
2098 | |
|
|
2099 | XPUSHs (ST (0)); |
|
|
2100 | } |
|
|
2101 | |
|
|
2102 | void encode (JSON *self, SV *scalar) |
|
|
2103 | PPCODE: |
|
|
2104 | PUTBACK; scalar = encode_json (scalar, self); SPAGAIN; |
|
|
2105 | XPUSHs (scalar); |
|
|
2106 | |
|
|
2107 | void decode (JSON *self, SV *jsonstr) |
|
|
2108 | PPCODE: |
|
|
2109 | PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN; |
|
|
2110 | XPUSHs (jsonstr); |
|
|
2111 | |
|
|
2112 | void decode_prefix (JSON *self, SV *jsonstr) |
|
|
2113 | PPCODE: |
|
|
2114 | { |
|
|
2115 | SV *sv; |
|
|
2116 | char *offset; |
|
|
2117 | PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; |
|
|
2118 | EXTEND (SP, 2); |
|
|
2119 | PUSHs (sv); |
|
|
2120 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); |
|
|
2121 | } |
|
|
2122 | |
|
|
2123 | void incr_parse (JSON *self, SV *jsonstr = 0) |
|
|
2124 | PPCODE: |
|
|
2125 | { |
|
|
2126 | if (!self->incr_text) |
|
|
2127 | self->incr_text = newSVpvn ("", 0); |
|
|
2128 | |
|
|
2129 | /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */ |
|
|
2130 | if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text)) |
|
|
2131 | if (DECODE_WANTS_OCTETS (self)) |
|
|
2132 | { |
|
|
2133 | if (self->incr_pos) |
|
|
2134 | self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text), |
|
|
2135 | (U8 *)SvPVX (self->incr_text) + self->incr_pos); |
|
|
2136 | |
|
|
2137 | sv_utf8_downgrade (self->incr_text, 0); |
|
|
2138 | } |
|
|
2139 | else |
|
|
2140 | { |
|
|
2141 | sv_utf8_upgrade (self->incr_text); |
|
|
2142 | |
|
|
2143 | if (self->incr_pos) |
|
|
2144 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
|
|
2145 | - (U8 *)SvPVX (self->incr_text); |
|
|
2146 | } |
|
|
2147 | |
|
|
2148 | // append data, if any |
|
|
2149 | if (jsonstr) |
|
|
2150 | { |
|
|
2151 | /* make sure both strings have same encoding */ |
|
|
2152 | if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text)) |
|
|
2153 | if (SvUTF8 (jsonstr)) |
|
|
2154 | sv_utf8_downgrade (jsonstr, 0); |
|
|
2155 | else |
|
|
2156 | sv_utf8_upgrade (jsonstr); |
|
|
2157 | |
|
|
2158 | /* and then just blindly append */ |
|
|
2159 | { |
|
|
2160 | STRLEN len; |
|
|
2161 | const char *str = SvPV (jsonstr, len); |
|
|
2162 | STRLEN cur = SvCUR (self->incr_text); |
|
|
2163 | |
|
|
2164 | if (SvLEN (self->incr_text) <= cur + len) |
|
|
2165 | SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
|
|
2166 | |
|
|
2167 | Move (str, SvEND (self->incr_text), len, char); |
|
|
2168 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
|
|
2169 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
|
|
2170 | } |
|
|
2171 | } |
|
|
2172 | |
|
|
2173 | if (GIMME_V != G_VOID) |
|
|
2174 | do |
|
|
2175 | { |
|
|
2176 | SV *sv; |
|
|
2177 | char *offset; |
|
|
2178 | |
|
|
2179 | if (!INCR_DONE (self)) |
|
|
2180 | { |
|
|
2181 | incr_parse (self); |
|
|
2182 | |
|
|
2183 | if (self->incr_pos > self->max_size && self->max_size) |
|
|
2184 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
|
|
2185 | (unsigned long)self->incr_pos, (unsigned long)self->max_size); |
|
|
2186 | |
|
|
2187 | if (!INCR_DONE (self)) |
|
|
2188 | { |
|
|
2189 | // as an optimisation, do not accumulate white space in the incr buffer |
|
|
2190 | if (self->incr_mode == INCR_M_WS && self->incr_pos) |
|
|
2191 | { |
|
|
2192 | self->incr_pos = 0; |
|
|
2193 | SvCUR_set (self->incr_text, 0); |
|
|
2194 | } |
|
|
2195 | |
|
|
2196 | break; |
|
|
2197 | } |
|
|
2198 | } |
|
|
2199 | |
|
|
2200 | PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; |
|
|
2201 | XPUSHs (sv); |
|
|
2202 | |
|
|
2203 | self->incr_pos -= offset - SvPVX (self->incr_text); |
|
|
2204 | self->incr_nest = 0; |
|
|
2205 | self->incr_mode = 0; |
|
|
2206 | |
|
|
2207 | sv_chop (self->incr_text, offset); |
|
|
2208 | } |
|
|
2209 | while (GIMME_V == G_ARRAY); |
|
|
2210 | } |
|
|
2211 | |
|
|
2212 | SV *incr_text (JSON *self) |
|
|
2213 | ATTRS: lvalue |
|
|
2214 | CODE: |
|
|
2215 | { |
|
|
2216 | if (self->incr_pos) |
|
|
2217 | croak ("incr_text can not be called when the incremental parser already started parsing"); |
|
|
2218 | |
|
|
2219 | RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef; |
|
|
2220 | } |
|
|
2221 | OUTPUT: |
|
|
2222 | RETVAL |
|
|
2223 | |
|
|
2224 | void incr_skip (JSON *self) |
|
|
2225 | CODE: |
|
|
2226 | { |
|
|
2227 | if (self->incr_pos) |
|
|
2228 | { |
|
|
2229 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos); |
|
|
2230 | self->incr_pos = 0; |
|
|
2231 | self->incr_nest = 0; |
|
|
2232 | self->incr_mode = 0; |
|
|
2233 | } |
|
|
2234 | } |
|
|
2235 | |
|
|
2236 | void incr_reset (JSON *self) |
|
|
2237 | CODE: |
|
|
2238 | { |
|
|
2239 | SvREFCNT_dec (self->incr_text); |
|
|
2240 | self->incr_text = 0; |
|
|
2241 | self->incr_pos = 0; |
|
|
2242 | self->incr_nest = 0; |
|
|
2243 | self->incr_mode = 0; |
|
|
2244 | } |
|
|
2245 | |
|
|
2246 | void DESTROY (JSON *self) |
|
|
2247 | CODE: |
|
|
2248 | SvREFCNT_dec (self->cb_sk_object); |
|
|
2249 | SvREFCNT_dec (self->cb_object); |
|
|
2250 | SvREFCNT_dec (self->incr_text); |
903 | |
2251 | |
904 | PROTOTYPES: ENABLE |
2252 | PROTOTYPES: ENABLE |
905 | |
2253 | |
906 | void to_json (SV *scalar) |
2254 | void encode_json (SV *scalar) |
907 | PPCODE: |
2255 | PPCODE: |
908 | XPUSHs (encode_json (scalar, F_UTF8)); |
2256 | { |
|
|
2257 | JSON json; |
|
|
2258 | json_init (&json); |
|
|
2259 | json.flags |= F_UTF8; |
|
|
2260 | PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN; |
|
|
2261 | XPUSHs (scalar); |
|
|
2262 | } |
909 | |
2263 | |
910 | void from_json (SV *jsonstr) |
2264 | void decode_json (SV *jsonstr) |
911 | PPCODE: |
2265 | PPCODE: |
912 | XPUSHs (decode_json (jsonstr, F_UTF8)); |
2266 | { |
|
|
2267 | JSON json; |
|
|
2268 | json_init (&json); |
|
|
2269 | json.flags |= F_UTF8; |
|
|
2270 | PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN; |
|
|
2271 | XPUSHs (jsonstr); |
|
|
2272 | } |
913 | |
2273 | |