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 | // three extra for rounding, sign, and end of string |
|
|
23 | #define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3) |
|
|
24 | |
9 | #define F_ASCII 0x00000001 |
25 | #define F_ASCII 0x00000001UL |
|
|
26 | #define F_LATIN1 0x00000002UL |
10 | #define F_UTF8 0x00000002 |
27 | #define F_UTF8 0x00000004UL |
11 | #define F_INDENT 0x00000004 |
28 | #define F_INDENT 0x00000008UL |
12 | #define F_CANONICAL 0x00000008 |
29 | #define F_CANONICAL 0x00000010UL |
13 | #define F_SPACE_BEFORE 0x00000010 |
30 | #define F_SPACE_BEFORE 0x00000020UL |
14 | #define F_SPACE_AFTER 0x00000020 |
31 | #define F_SPACE_AFTER 0x00000040UL |
15 | #define F_ALLOW_NONREF 0x00000080 |
32 | #define F_ALLOW_NONREF 0x00000100UL |
16 | #define F_SHRINK 0x00000100 |
33 | #define F_SHRINK 0x00000200UL |
|
|
34 | #define F_ALLOW_BLESSED 0x00000400UL |
|
|
35 | #define F_CONV_BLESSED 0x00000800UL |
|
|
36 | #define F_RELAXED 0x00001000UL |
|
|
37 | #define F_ALLOW_UNKNOWN 0x00002000UL |
|
|
38 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
17 | |
39 | |
18 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
40 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
19 | #define F_DEFAULT 0 |
|
|
20 | |
41 | |
21 | #define INIT_SIZE 32 // initial scalar size to be allocated |
42 | #define INIT_SIZE 32 // initial scalar size to be allocated |
22 | #define INDENT_STEP 3 // spaces per indentation level |
43 | #define INDENT_STEP 3 // spaces per indentation level |
23 | |
44 | |
24 | #define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character |
|
|
25 | #define SHORT_STRING_LEN 256 // special-case strings of up to this size |
45 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
26 | |
46 | |
27 | #define SB do { |
47 | #define SB do { |
28 | #define SE } while (0) |
48 | #define SE } while (0) |
29 | |
49 | |
|
|
50 | #if __GNUC__ >= 3 |
|
|
51 | # define expect(expr,value) __builtin_expect ((expr), (value)) |
|
|
52 | # define INLINE static inline |
|
|
53 | #else |
|
|
54 | # define expect(expr,value) (expr) |
|
|
55 | # define INLINE static |
|
|
56 | #endif |
|
|
57 | |
|
|
58 | #define expect_false(expr) expect ((expr) != 0, 0) |
|
|
59 | #define expect_true(expr) expect ((expr) != 0, 1) |
|
|
60 | |
|
|
61 | #define IN_RANGE_INC(type,val,beg,end) \ |
|
|
62 | ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \ |
|
|
63 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
|
|
64 | |
|
|
65 | #define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)" |
|
|
66 | |
|
|
67 | #ifdef USE_ITHREADS |
|
|
68 | # define JSON_SLOW 1 |
|
|
69 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
|
|
70 | #else |
|
|
71 | # define JSON_SLOW 0 |
|
|
72 | # define JSON_STASH json_stash |
|
|
73 | #endif |
|
|
74 | |
30 | static HV *json_stash; // JSON::XS:: |
75 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
|
|
76 | static SV *json_true, *json_false; |
|
|
77 | |
|
|
78 | enum { |
|
|
79 | INCR_M_WS = 0, // initial whitespace skipping, must be 0 |
|
|
80 | INCR_M_STR, // inside string |
|
|
81 | INCR_M_BS, // inside backslash |
|
|
82 | INCR_M_C0, // inside comment in initial whitespace sequence |
|
|
83 | INCR_M_C1, // inside comment in other places |
|
|
84 | INCR_M_JSON // outside anything, count nesting |
|
|
85 | }; |
|
|
86 | |
|
|
87 | #define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) |
|
|
88 | |
|
|
89 | typedef struct { |
|
|
90 | U32 flags; |
|
|
91 | U32 max_depth; |
|
|
92 | STRLEN max_size; |
|
|
93 | |
|
|
94 | SV *cb_object; |
|
|
95 | HV *cb_sk_object; |
|
|
96 | |
|
|
97 | // for the incremental parser |
|
|
98 | SV *incr_text; // the source text so far |
|
|
99 | STRLEN incr_pos; // the current offset into the text |
|
|
100 | int incr_nest; // {[]}-nesting level |
|
|
101 | unsigned char incr_mode; |
|
|
102 | } JSON; |
|
|
103 | |
|
|
104 | INLINE void |
|
|
105 | json_init (JSON *json) |
|
|
106 | { |
|
|
107 | Zero (json, 1, JSON); |
|
|
108 | json->max_depth = 512; |
|
|
109 | } |
31 | |
110 | |
32 | ///////////////////////////////////////////////////////////////////////////// |
111 | ///////////////////////////////////////////////////////////////////////////// |
33 | // utility functions |
112 | // utility functions |
34 | |
113 | |
35 | static UV * |
114 | INLINE SV * |
36 | SvJSON (SV *sv) |
115 | get_bool (const char *name) |
37 | { |
116 | { |
38 | if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) |
117 | SV *sv = get_sv (name, 1); |
39 | croak ("object is not of type JSON::XS"); |
|
|
40 | |
118 | |
41 | return &SvUVX (SvRV (sv)); |
119 | SvREADONLY_on (sv); |
42 | } |
120 | SvREADONLY_on (SvRV (sv)); |
43 | |
121 | |
44 | static void |
122 | return sv; |
|
|
123 | } |
|
|
124 | |
|
|
125 | INLINE void |
45 | shrink (SV *sv) |
126 | shrink (SV *sv) |
46 | { |
127 | { |
47 | sv_utf8_downgrade (sv, 1); |
128 | sv_utf8_downgrade (sv, 1); |
|
|
129 | |
48 | if (SvLEN (sv) > SvCUR (sv) + 1) |
130 | if (SvLEN (sv) > SvCUR (sv) + 1) |
49 | { |
131 | { |
50 | #ifdef SvPV_shrink_to_cur |
132 | #ifdef SvPV_shrink_to_cur |
51 | SvPV_shrink_to_cur (sv); |
133 | SvPV_shrink_to_cur (sv); |
52 | #elif defined (SvPV_renew) |
134 | #elif defined (SvPV_renew) |
… | |
… | |
58 | // decode an utf-8 character and return it, or (UV)-1 in |
140 | // decode an utf-8 character and return it, or (UV)-1 in |
59 | // case of an error. |
141 | // case of an error. |
60 | // we special-case "safe" characters from U+80 .. U+7FF, |
142 | // we special-case "safe" characters from U+80 .. U+7FF, |
61 | // but use the very good perl function to parse anything else. |
143 | // but use the very good perl function to parse anything else. |
62 | // note that we never call this function for a ascii codepoints |
144 | // note that we never call this function for a ascii codepoints |
63 | static UV |
145 | INLINE UV |
64 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
146 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
65 | { |
147 | { |
66 | if (s[0] > 0xdf || s[0] < 0xc2) |
148 | if (expect_true (len >= 2 |
67 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
149 | && IN_RANGE_INC (char, s[0], 0xc2, 0xdf) |
68 | else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
150 | && IN_RANGE_INC (char, s[1], 0x80, 0xbf))) |
69 | { |
151 | { |
70 | *clen = 2; |
152 | *clen = 2; |
71 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
153 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
72 | } |
154 | } |
73 | else |
155 | else |
74 | return (UV)-1; |
156 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
75 | } |
157 | } |
76 | |
158 | |
|
|
159 | // likewise for encoding, also never called for ascii codepoints |
|
|
160 | // this function takes advantage of this fact, although current gccs |
|
|
161 | // seem to optimise the check for >= 0x80 away anyways |
|
|
162 | INLINE unsigned char * |
|
|
163 | encode_utf8 (unsigned char *s, UV ch) |
|
|
164 | { |
|
|
165 | if (expect_false (ch < 0x000080)) |
|
|
166 | *s++ = ch; |
|
|
167 | else if (expect_true (ch < 0x000800)) |
|
|
168 | *s++ = 0xc0 | ( ch >> 6), |
|
|
169 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
170 | else if ( ch < 0x010000) |
|
|
171 | *s++ = 0xe0 | ( ch >> 12), |
|
|
172 | *s++ = 0x80 | ((ch >> 6) & 0x3f), |
|
|
173 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
174 | else if ( ch < 0x110000) |
|
|
175 | *s++ = 0xf0 | ( ch >> 18), |
|
|
176 | *s++ = 0x80 | ((ch >> 12) & 0x3f), |
|
|
177 | *s++ = 0x80 | ((ch >> 6) & 0x3f), |
|
|
178 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
179 | |
|
|
180 | return s; |
|
|
181 | } |
|
|
182 | |
|
|
183 | // convert offset pointer to character index, sv must be string |
|
|
184 | static STRLEN |
|
|
185 | ptr_to_index (SV *sv, char *offset) |
|
|
186 | { |
|
|
187 | return SvUTF8 (sv) |
|
|
188 | ? utf8_distance (offset, SvPVX (sv)) |
|
|
189 | : offset - SvPVX (sv); |
|
|
190 | } |
|
|
191 | |
|
|
192 | ///////////////////////////////////////////////////////////////////////////// |
|
|
193 | // fp hell |
|
|
194 | |
|
|
195 | // scan a group of digits, and a trailing exponent |
|
|
196 | static void |
|
|
197 | json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp) |
|
|
198 | { |
|
|
199 | UV uaccum = 0; |
|
|
200 | int eaccum = 0; |
|
|
201 | |
|
|
202 | for (;;) |
|
|
203 | { |
|
|
204 | U8 dig = (U8)*s - '0'; |
|
|
205 | |
|
|
206 | if (expect_false (dig >= 10)) |
|
|
207 | { |
|
|
208 | if (dig == (U8)((U8)'.' - (U8)'0')) |
|
|
209 | { |
|
|
210 | ++s; |
|
|
211 | json_atof_scan1 (s, accum, expo, 1); |
|
|
212 | } |
|
|
213 | else if ((dig | ' ') == 'e' - '0') |
|
|
214 | { |
|
|
215 | int exp2 = 0; |
|
|
216 | int neg = 0; |
|
|
217 | |
|
|
218 | ++s; |
|
|
219 | |
|
|
220 | if (*s == '-') |
|
|
221 | { |
|
|
222 | ++s; |
|
|
223 | neg = 1; |
|
|
224 | } |
|
|
225 | else if (*s == '+') |
|
|
226 | ++s; |
|
|
227 | |
|
|
228 | while ((dig = (U8)*s - '0') < 10) |
|
|
229 | exp2 = exp2 * 10 + *s++ - '0'; |
|
|
230 | |
|
|
231 | *expo += neg ? -exp2 : exp2; |
|
|
232 | } |
|
|
233 | |
|
|
234 | break; |
|
|
235 | } |
|
|
236 | |
|
|
237 | ++s; |
|
|
238 | |
|
|
239 | uaccum = uaccum * 10 + dig; |
|
|
240 | ++eaccum; |
|
|
241 | |
|
|
242 | // if we have too many digits, then recurse for more |
|
|
243 | // we actually do this for rather few digits |
|
|
244 | if (uaccum >= (UV_MAX - 9) / 10) |
|
|
245 | { |
|
|
246 | if (postdp) *expo -= eaccum; |
|
|
247 | json_atof_scan1 (s, accum, expo, postdp); |
|
|
248 | if (postdp) *expo += eaccum; |
|
|
249 | |
|
|
250 | break; |
|
|
251 | } |
|
|
252 | } |
|
|
253 | |
|
|
254 | // this relies greatly on the quality of the pow () |
|
|
255 | // implementation of the platform, but a good |
|
|
256 | // implementation is hard to beat. |
|
|
257 | if (postdp) *expo -= eaccum; |
|
|
258 | *accum += uaccum * Perl_pow (10., *expo); |
|
|
259 | *expo += eaccum; |
|
|
260 | } |
|
|
261 | |
|
|
262 | static NV |
|
|
263 | json_atof (const char *s) |
|
|
264 | { |
|
|
265 | NV accum = 0.; |
|
|
266 | int expo = 0; |
|
|
267 | int neg = 0; |
|
|
268 | |
|
|
269 | if (*s == '-') |
|
|
270 | { |
|
|
271 | ++s; |
|
|
272 | neg = 1; |
|
|
273 | } |
|
|
274 | |
|
|
275 | json_atof_scan1 (s, &accum, &expo, 0); |
|
|
276 | |
|
|
277 | return neg ? -accum : accum; |
|
|
278 | } |
77 | ///////////////////////////////////////////////////////////////////////////// |
279 | ///////////////////////////////////////////////////////////////////////////// |
78 | // encoder |
280 | // encoder |
79 | |
281 | |
80 | // structure used for encoding JSON |
282 | // structure used for encoding JSON |
81 | typedef struct |
283 | typedef struct |
82 | { |
284 | { |
83 | char *cur; // SvPVX (sv) + current output position |
285 | char *cur; // SvPVX (sv) + current output position |
84 | char *end; // SvEND (sv) |
286 | char *end; // SvEND (sv) |
85 | SV *sv; // result scalar |
287 | SV *sv; // result scalar |
86 | UV flags; // F_* |
288 | JSON json; |
87 | int indent; // indentation level |
289 | U32 indent; // indentation level |
88 | int max_depth; // max. recursion level |
290 | UV limit; // escape character values >= this value when encoding |
89 | } enc_t; |
291 | } enc_t; |
90 | |
292 | |
91 | static void |
293 | INLINE void |
92 | need (enc_t *enc, STRLEN len) |
294 | need (enc_t *enc, STRLEN len) |
93 | { |
295 | { |
94 | if (enc->cur + len >= enc->end) |
296 | if (expect_false (enc->cur + len >= enc->end)) |
95 | { |
297 | { |
96 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
298 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
97 | SvGROW (enc->sv, cur + len + 1); |
299 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
98 | enc->cur = SvPVX (enc->sv) + cur; |
300 | enc->cur = SvPVX (enc->sv) + cur; |
99 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); |
301 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
100 | } |
302 | } |
101 | } |
303 | } |
102 | |
304 | |
103 | static void |
305 | INLINE void |
104 | encode_ch (enc_t *enc, char ch) |
306 | encode_ch (enc_t *enc, char ch) |
105 | { |
307 | { |
106 | need (enc, 1); |
308 | need (enc, 1); |
107 | *enc->cur++ = ch; |
309 | *enc->cur++ = ch; |
108 | } |
310 | } |
… | |
… | |
116 | |
318 | |
117 | while (str < end) |
319 | while (str < end) |
118 | { |
320 | { |
119 | unsigned char ch = *(unsigned char *)str; |
321 | unsigned char ch = *(unsigned char *)str; |
120 | |
322 | |
121 | if (ch >= 0x20 && ch < 0x80) // most common case |
323 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
122 | { |
324 | { |
123 | if (ch == '"') // but with slow exceptions |
325 | if (expect_false (ch == '"')) // but with slow exceptions |
124 | { |
326 | { |
125 | need (enc, len += 1); |
327 | need (enc, len += 1); |
126 | *enc->cur++ = '\\'; |
328 | *enc->cur++ = '\\'; |
127 | *enc->cur++ = '"'; |
329 | *enc->cur++ = '"'; |
128 | } |
330 | } |
129 | else if (ch == '\\') |
331 | else if (expect_false (ch == '\\')) |
130 | { |
332 | { |
131 | need (enc, len += 1); |
333 | need (enc, len += 1); |
132 | *enc->cur++ = '\\'; |
334 | *enc->cur++ = '\\'; |
133 | *enc->cur++ = '\\'; |
335 | *enc->cur++ = '\\'; |
134 | } |
336 | } |
… | |
… | |
152 | STRLEN clen; |
354 | STRLEN clen; |
153 | UV uch; |
355 | UV uch; |
154 | |
356 | |
155 | if (is_utf8) |
357 | if (is_utf8) |
156 | { |
358 | { |
157 | //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); |
|
|
158 | uch = decode_utf8 (str, end - str, &clen); |
359 | uch = decode_utf8 (str, end - str, &clen); |
159 | if (clen == (STRLEN)-1) |
360 | if (clen == (STRLEN)-1) |
160 | croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
361 | croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
161 | } |
362 | } |
162 | else |
363 | else |
163 | { |
364 | { |
164 | uch = ch; |
365 | uch = ch; |
165 | clen = 1; |
366 | clen = 1; |
166 | } |
367 | } |
167 | |
368 | |
168 | if (uch > 0x10FFFFUL) |
369 | if (uch < 0x80/*0x20*/ || uch >= enc->limit) |
169 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
|
|
170 | |
|
|
171 | if (uch < 0x80 || enc->flags & F_ASCII) |
|
|
172 | { |
370 | { |
173 | if (uch > 0xFFFFUL) |
371 | if (uch >= 0x10000UL) |
174 | { |
372 | { |
|
|
373 | if (uch >= 0x110000UL) |
|
|
374 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
|
|
375 | |
175 | need (enc, len += 11); |
376 | need (enc, len += 11); |
176 | sprintf (enc->cur, "\\u%04x\\u%04x", |
377 | sprintf (enc->cur, "\\u%04x\\u%04x", |
177 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
378 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
178 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
379 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
179 | enc->cur += 12; |
380 | enc->cur += 12; |
180 | } |
381 | } |
181 | else |
382 | else |
182 | { |
383 | { |
183 | static char hexdigit [16] = "0123456789abcdef"; |
|
|
184 | need (enc, len += 5); |
384 | need (enc, len += 5); |
185 | *enc->cur++ = '\\'; |
385 | *enc->cur++ = '\\'; |
186 | *enc->cur++ = 'u'; |
386 | *enc->cur++ = 'u'; |
187 | *enc->cur++ = hexdigit [ uch >> 12 ]; |
387 | *enc->cur++ = PL_hexdigit [ uch >> 12 ]; |
188 | *enc->cur++ = hexdigit [(uch >> 8) & 15]; |
388 | *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; |
189 | *enc->cur++ = hexdigit [(uch >> 4) & 15]; |
389 | *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; |
190 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
390 | *enc->cur++ = PL_hexdigit [(uch >> 0) & 15]; |
191 | } |
391 | } |
192 | |
392 | |
|
|
393 | str += clen; |
|
|
394 | } |
|
|
395 | else if (enc->json.flags & F_LATIN1) |
|
|
396 | { |
|
|
397 | *enc->cur++ = uch; |
193 | str += clen; |
398 | str += clen; |
194 | } |
399 | } |
195 | else if (is_utf8) |
400 | else if (is_utf8) |
196 | { |
401 | { |
197 | need (enc, len += clen); |
402 | need (enc, len += clen); |
… | |
… | |
201 | } |
406 | } |
202 | while (--clen); |
407 | while (--clen); |
203 | } |
408 | } |
204 | else |
409 | else |
205 | { |
410 | { |
206 | need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed |
411 | need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
207 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
412 | enc->cur = encode_utf8 (enc->cur, uch); |
208 | ++str; |
413 | ++str; |
209 | } |
414 | } |
210 | } |
415 | } |
211 | } |
416 | } |
212 | } |
417 | } |
213 | |
418 | |
214 | --len; |
419 | --len; |
215 | } |
420 | } |
216 | } |
421 | } |
217 | |
422 | |
218 | static void |
423 | INLINE void |
219 | encode_indent (enc_t *enc) |
424 | encode_indent (enc_t *enc) |
220 | { |
425 | { |
221 | if (enc->flags & F_INDENT) |
426 | if (enc->json.flags & F_INDENT) |
222 | { |
427 | { |
223 | int spaces = enc->indent * INDENT_STEP; |
428 | int spaces = enc->indent * INDENT_STEP; |
224 | |
429 | |
225 | need (enc, spaces); |
430 | need (enc, spaces); |
226 | memset (enc->cur, ' ', spaces); |
431 | memset (enc->cur, ' ', spaces); |
227 | enc->cur += spaces; |
432 | enc->cur += spaces; |
228 | } |
433 | } |
229 | } |
434 | } |
230 | |
435 | |
231 | static void |
436 | INLINE void |
232 | encode_space (enc_t *enc) |
437 | encode_space (enc_t *enc) |
233 | { |
438 | { |
234 | need (enc, 1); |
439 | need (enc, 1); |
235 | encode_ch (enc, ' '); |
440 | encode_ch (enc, ' '); |
236 | } |
441 | } |
237 | |
442 | |
238 | static void |
443 | INLINE void |
239 | encode_nl (enc_t *enc) |
444 | encode_nl (enc_t *enc) |
240 | { |
445 | { |
241 | if (enc->flags & F_INDENT) |
446 | if (enc->json.flags & F_INDENT) |
242 | { |
447 | { |
243 | need (enc, 1); |
448 | need (enc, 1); |
244 | encode_ch (enc, '\n'); |
449 | encode_ch (enc, '\n'); |
245 | } |
450 | } |
246 | } |
451 | } |
247 | |
452 | |
248 | static void |
453 | INLINE void |
249 | encode_comma (enc_t *enc) |
454 | encode_comma (enc_t *enc) |
250 | { |
455 | { |
251 | encode_ch (enc, ','); |
456 | encode_ch (enc, ','); |
252 | |
457 | |
253 | if (enc->flags & F_INDENT) |
458 | if (enc->json.flags & F_INDENT) |
254 | encode_nl (enc); |
459 | encode_nl (enc); |
255 | else if (enc->flags & F_SPACE_AFTER) |
460 | else if (enc->json.flags & F_SPACE_AFTER) |
256 | encode_space (enc); |
461 | encode_space (enc); |
257 | } |
462 | } |
258 | |
463 | |
259 | static void encode_sv (enc_t *enc, SV *sv); |
464 | static void encode_sv (enc_t *enc, SV *sv); |
260 | |
465 | |
261 | static void |
466 | static void |
262 | encode_av (enc_t *enc, AV *av) |
467 | encode_av (enc_t *enc, AV *av) |
263 | { |
468 | { |
264 | int i, len = av_len (av); |
469 | int i, len = av_len (av); |
265 | |
470 | |
266 | encode_ch (enc, '['); encode_nl (enc); |
471 | if (enc->indent >= enc->json.max_depth) |
267 | ++enc->indent; |
472 | croak (ERR_NESTING_EXCEEDED); |
268 | |
473 | |
|
|
474 | encode_ch (enc, '['); |
|
|
475 | |
|
|
476 | if (len >= 0) |
|
|
477 | { |
|
|
478 | encode_nl (enc); ++enc->indent; |
|
|
479 | |
269 | for (i = 0; i <= len; ++i) |
480 | for (i = 0; i <= len; ++i) |
270 | { |
481 | { |
|
|
482 | SV **svp = av_fetch (av, i, 0); |
|
|
483 | |
271 | encode_indent (enc); |
484 | encode_indent (enc); |
272 | encode_sv (enc, *av_fetch (av, i, 0)); |
|
|
273 | |
485 | |
|
|
486 | if (svp) |
|
|
487 | encode_sv (enc, *svp); |
|
|
488 | else |
|
|
489 | encode_str (enc, "null", 4, 0); |
|
|
490 | |
274 | if (i < len) |
491 | if (i < len) |
275 | encode_comma (enc); |
492 | encode_comma (enc); |
276 | } |
493 | } |
277 | |
494 | |
|
|
495 | encode_nl (enc); --enc->indent; encode_indent (enc); |
|
|
496 | } |
|
|
497 | |
278 | encode_nl (enc); |
498 | encode_ch (enc, ']'); |
279 | |
|
|
280 | --enc->indent; |
|
|
281 | encode_indent (enc); encode_ch (enc, ']'); |
|
|
282 | } |
499 | } |
283 | |
500 | |
284 | static void |
501 | static void |
285 | encode_he (enc_t *enc, HE *he) |
502 | encode_hk (enc_t *enc, HE *he) |
286 | { |
503 | { |
287 | encode_ch (enc, '"'); |
504 | encode_ch (enc, '"'); |
288 | |
505 | |
289 | if (HeKLEN (he) == HEf_SVKEY) |
506 | if (HeKLEN (he) == HEf_SVKEY) |
290 | { |
507 | { |
… | |
… | |
300 | else |
517 | else |
301 | encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
518 | encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
302 | |
519 | |
303 | encode_ch (enc, '"'); |
520 | encode_ch (enc, '"'); |
304 | |
521 | |
305 | if (enc->flags & F_SPACE_BEFORE) encode_space (enc); |
522 | if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); |
306 | encode_ch (enc, ':'); |
523 | encode_ch (enc, ':'); |
307 | if (enc->flags & F_SPACE_AFTER ) encode_space (enc); |
524 | if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); |
308 | encode_sv (enc, HeVAL (he)); |
|
|
309 | } |
525 | } |
310 | |
526 | |
311 | // compare hash entries, used when all keys are bytestrings |
527 | // compare hash entries, used when all keys are bytestrings |
312 | static int |
528 | static int |
313 | he_cmp_fast (const void *a_, const void *b_) |
529 | he_cmp_fast (const void *a_, const void *b_) |
… | |
… | |
318 | HE *b = *(HE **)b_; |
534 | HE *b = *(HE **)b_; |
319 | |
535 | |
320 | STRLEN la = HeKLEN (a); |
536 | STRLEN la = HeKLEN (a); |
321 | STRLEN lb = HeKLEN (b); |
537 | STRLEN lb = HeKLEN (b); |
322 | |
538 | |
323 | if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) |
539 | if (!(cmp = memcmp (HeKEY (b), HeKEY (a), lb < la ? lb : la))) |
324 | cmp = la - lb; |
540 | cmp = lb - la; |
325 | |
541 | |
326 | return cmp; |
542 | return cmp; |
327 | } |
543 | } |
328 | |
544 | |
329 | // compare hash entries, used when some keys are sv's or utf-x |
545 | // compare hash entries, used when some keys are sv's or utf-x |
330 | static int |
546 | static int |
331 | he_cmp_slow (const void *a, const void *b) |
547 | he_cmp_slow (const void *a, const void *b) |
332 | { |
548 | { |
333 | return sv_cmp (HeSVKEY_force (*(HE **)a), HeSVKEY_force (*(HE **)b)); |
549 | return sv_cmp (HeSVKEY_force (*(HE **)b), HeSVKEY_force (*(HE **)a)); |
334 | } |
550 | } |
335 | |
551 | |
336 | static void |
552 | static void |
337 | encode_hv (enc_t *enc, HV *hv) |
553 | encode_hv (enc_t *enc, HV *hv) |
338 | { |
554 | { |
339 | int count, i; |
555 | HE *he; |
340 | |
556 | |
341 | encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; |
557 | if (enc->indent >= enc->json.max_depth) |
|
|
558 | croak (ERR_NESTING_EXCEEDED); |
342 | |
559 | |
343 | if ((count = hv_iterinit (hv))) |
560 | encode_ch (enc, '{'); |
344 | { |
561 | |
345 | // for canonical output we have to sort by keys first |
562 | // for canonical output we have to sort by keys first |
346 | // actually, this is mostly due to the stupid so-called |
563 | // actually, this is mostly due to the stupid so-called |
347 | // security workaround added somewhere in 5.8.x. |
564 | // security workaround added somewhere in 5.8.x |
348 | // that randomises hash orderings |
565 | // that randomises hash orderings |
349 | if (enc->flags & F_CANONICAL) |
566 | if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv)) |
|
|
567 | { |
|
|
568 | int count = hv_iterinit (hv); |
|
|
569 | |
|
|
570 | if (SvMAGICAL (hv)) |
350 | { |
571 | { |
|
|
572 | // need to count by iterating. could improve by dynamically building the vector below |
|
|
573 | // but I don't care for the speed of this special case. |
|
|
574 | // note also that we will run into undefined behaviour when the two iterations |
|
|
575 | // do not result in the same count, something I might care for in some later release. |
|
|
576 | |
|
|
577 | count = 0; |
|
|
578 | while (hv_iternext (hv)) |
|
|
579 | ++count; |
|
|
580 | |
|
|
581 | hv_iterinit (hv); |
|
|
582 | } |
|
|
583 | |
|
|
584 | if (count) |
|
|
585 | { |
|
|
586 | int i, fast = 1; |
|
|
587 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
|
|
588 | HE **hes = _alloca (count * sizeof (HE)); |
|
|
589 | #else |
351 | HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode |
590 | HE *hes [count]; // if your compiler dies here, you need to enable C99 mode |
352 | int fast = 1; |
591 | #endif |
353 | |
592 | |
354 | i = 0; |
593 | i = 0; |
355 | while ((he = hv_iternext (hv))) |
594 | while ((he = hv_iternext (hv))) |
356 | { |
595 | { |
357 | hes [i++] = he; |
596 | hes [i++] = he; |
… | |
… | |
379 | |
618 | |
380 | FREETMPS; |
619 | FREETMPS; |
381 | LEAVE; |
620 | LEAVE; |
382 | } |
621 | } |
383 | |
622 | |
384 | for (i = 0; i < count; ++i) |
623 | encode_nl (enc); ++enc->indent; |
|
|
624 | |
|
|
625 | while (count--) |
385 | { |
626 | { |
386 | encode_indent (enc); |
627 | encode_indent (enc); |
|
|
628 | he = hes [count]; |
387 | encode_he (enc, hes [i]); |
629 | encode_hk (enc, he); |
|
|
630 | encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
388 | |
631 | |
389 | if (i < count - 1) |
632 | if (count) |
390 | encode_comma (enc); |
633 | encode_comma (enc); |
391 | } |
634 | } |
392 | |
635 | |
|
|
636 | encode_nl (enc); --enc->indent; encode_indent (enc); |
|
|
637 | } |
|
|
638 | } |
|
|
639 | else |
|
|
640 | { |
|
|
641 | if (hv_iterinit (hv) || SvMAGICAL (hv)) |
|
|
642 | if ((he = hv_iternext (hv))) |
|
|
643 | { |
|
|
644 | encode_nl (enc); ++enc->indent; |
|
|
645 | |
|
|
646 | for (;;) |
|
|
647 | { |
393 | encode_nl (enc); |
648 | encode_indent (enc); |
|
|
649 | encode_hk (enc, he); |
|
|
650 | encode_sv (enc, expect_false (SvMAGICAL (hv)) ? hv_iterval (hv, he) : HeVAL (he)); |
|
|
651 | |
|
|
652 | if (!(he = hv_iternext (hv))) |
|
|
653 | break; |
|
|
654 | |
|
|
655 | encode_comma (enc); |
|
|
656 | } |
|
|
657 | |
|
|
658 | encode_nl (enc); --enc->indent; encode_indent (enc); |
|
|
659 | } |
|
|
660 | } |
|
|
661 | |
|
|
662 | encode_ch (enc, '}'); |
|
|
663 | } |
|
|
664 | |
|
|
665 | // encode objects, arrays and special \0=false and \1=true values. |
|
|
666 | static void |
|
|
667 | encode_rv (enc_t *enc, SV *sv) |
|
|
668 | { |
|
|
669 | svtype svt; |
|
|
670 | |
|
|
671 | SvGETMAGIC (sv); |
|
|
672 | svt = SvTYPE (sv); |
|
|
673 | |
|
|
674 | if (expect_false (SvOBJECT (sv))) |
|
|
675 | { |
|
|
676 | HV *stash = !JSON_SLOW || json_boolean_stash |
|
|
677 | ? json_boolean_stash |
|
|
678 | : gv_stashpv ("JSON::XS::Boolean", 1); |
|
|
679 | |
|
|
680 | if (SvSTASH (sv) == stash) |
|
|
681 | { |
|
|
682 | if (SvIV (sv)) |
|
|
683 | encode_str (enc, "true", 4, 0); |
|
|
684 | else |
|
|
685 | encode_str (enc, "false", 5, 0); |
394 | } |
686 | } |
395 | else |
687 | else |
396 | { |
688 | { |
397 | SV *sv; |
689 | #if 0 |
398 | HE *he = hv_iternext (hv); |
690 | if (0 && sv_derived_from (rv, "JSON::Literal")) |
399 | |
|
|
400 | for (;;) |
|
|
401 | { |
691 | { |
402 | encode_indent (enc); |
692 | // not yet |
403 | encode_he (enc, he); |
|
|
404 | |
|
|
405 | if (!(he = hv_iternext (hv))) |
|
|
406 | break; |
|
|
407 | |
|
|
408 | encode_comma (enc); |
|
|
409 | } |
693 | } |
|
|
694 | #endif |
|
|
695 | if (enc->json.flags & F_CONV_BLESSED) |
|
|
696 | { |
|
|
697 | // we re-bless the reference to get overload and other niceties right |
|
|
698 | GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0); |
410 | |
699 | |
|
|
700 | if (to_json) |
|
|
701 | { |
|
|
702 | dSP; |
|
|
703 | |
|
|
704 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
705 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); |
|
|
706 | |
|
|
707 | // calling with G_SCALAR ensures that we always get a 1 return value |
|
|
708 | PUTBACK; |
|
|
709 | call_sv ((SV *)GvCV (to_json), G_SCALAR); |
|
|
710 | SPAGAIN; |
|
|
711 | |
|
|
712 | // catch this surprisingly common error |
|
|
713 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
714 | croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
715 | |
|
|
716 | sv = POPs; |
|
|
717 | PUTBACK; |
|
|
718 | |
411 | encode_nl (enc); |
719 | encode_sv (enc, sv); |
|
|
720 | |
|
|
721 | FREETMPS; LEAVE; |
|
|
722 | } |
|
|
723 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
724 | encode_str (enc, "null", 4, 0); |
|
|
725 | else |
|
|
726 | croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it", |
|
|
727 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
728 | } |
|
|
729 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
730 | encode_str (enc, "null", 4, 0); |
|
|
731 | else |
|
|
732 | croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", |
|
|
733 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
412 | } |
734 | } |
413 | } |
735 | } |
|
|
736 | else if (svt == SVt_PVHV) |
|
|
737 | encode_hv (enc, (HV *)sv); |
|
|
738 | else if (svt == SVt_PVAV) |
|
|
739 | encode_av (enc, (AV *)sv); |
|
|
740 | else if (svt < SVt_PVAV) |
|
|
741 | { |
|
|
742 | STRLEN len = 0; |
|
|
743 | char *pv = svt ? SvPV (sv, len) : 0; |
414 | |
744 | |
415 | --enc->indent; encode_indent (enc); encode_ch (enc, '}'); |
745 | if (len == 1 && *pv == '1') |
|
|
746 | encode_str (enc, "true", 4, 0); |
|
|
747 | else if (len == 1 && *pv == '0') |
|
|
748 | encode_str (enc, "false", 5, 0); |
|
|
749 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
750 | encode_str (enc, "null", 4, 0); |
|
|
751 | else |
|
|
752 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
|
|
753 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
754 | } |
|
|
755 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
756 | encode_str (enc, "null", 4, 0); |
|
|
757 | else |
|
|
758 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
|
|
759 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
416 | } |
760 | } |
417 | |
761 | |
418 | static void |
762 | static void |
419 | encode_sv (enc_t *enc, SV *sv) |
763 | encode_sv (enc_t *enc, SV *sv) |
420 | { |
764 | { |
… | |
… | |
428 | encode_str (enc, str, len, SvUTF8 (sv)); |
772 | encode_str (enc, str, len, SvUTF8 (sv)); |
429 | encode_ch (enc, '"'); |
773 | encode_ch (enc, '"'); |
430 | } |
774 | } |
431 | else if (SvNOKp (sv)) |
775 | else if (SvNOKp (sv)) |
432 | { |
776 | { |
|
|
777 | // trust that perl will do the right thing w.r.t. JSON syntax. |
433 | need (enc, NV_DIG + 32); |
778 | need (enc, NV_DIG + 32); |
434 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
779 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
435 | enc->cur += strlen (enc->cur); |
780 | enc->cur += strlen (enc->cur); |
436 | } |
781 | } |
437 | else if (SvIOKp (sv)) |
782 | else if (SvIOKp (sv)) |
438 | { |
783 | { |
|
|
784 | // we assume we can always read an IV as a UV and vice versa |
|
|
785 | // we assume two's complement |
|
|
786 | // we assume no aliasing issues in the union |
|
|
787 | if (SvIsUV (sv) ? SvUVX (sv) <= 59000 |
|
|
788 | : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000) |
|
|
789 | { |
|
|
790 | // optimise the "small number case" |
|
|
791 | // code will likely be branchless and use only a single multiplication |
|
|
792 | // works for numbers up to 59074 |
|
|
793 | I32 i = SvIVX (sv); |
|
|
794 | U32 u; |
|
|
795 | char digit, nz = 0; |
|
|
796 | |
439 | need (enc, 64); |
797 | need (enc, 6); |
|
|
798 | |
|
|
799 | *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0; |
|
|
800 | u = i < 0 ? -i : i; |
|
|
801 | |
|
|
802 | // convert to 4.28 fixed-point representation |
|
|
803 | u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits |
|
|
804 | |
|
|
805 | // now output digit by digit, each time masking out the integer part |
|
|
806 | // and multiplying by 5 while moving the decimal point one to the right, |
|
|
807 | // resulting in a net multiplication by 10. |
|
|
808 | // we always write the digit to memory but conditionally increment |
|
|
809 | // the pointer, to enable the use of conditional move instructions. |
|
|
810 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5; |
|
|
811 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5; |
|
|
812 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5; |
|
|
813 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5; |
|
|
814 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
|
|
815 | } |
|
|
816 | else |
|
|
817 | { |
|
|
818 | // large integer, use the (rather slow) snprintf way. |
|
|
819 | need (enc, IVUV_MAXCHARS); |
440 | enc->cur += |
820 | enc->cur += |
441 | SvIsUV(sv) |
821 | SvIsUV(sv) |
442 | ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) |
822 | ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) |
443 | : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); |
823 | : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); |
|
|
824 | } |
444 | } |
825 | } |
445 | else if (SvROK (sv)) |
826 | else if (SvROK (sv)) |
446 | { |
827 | encode_rv (enc, SvRV (sv)); |
447 | SV *rv = SvRV (sv); |
828 | else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) |
448 | |
|
|
449 | if (enc->indent >= enc->max_depth) |
|
|
450 | croak ("data structure too deep (hit recursion limit)"); |
|
|
451 | |
|
|
452 | switch (SvTYPE (rv)) |
|
|
453 | { |
|
|
454 | case SVt_PVAV: encode_av (enc, (AV *)rv); break; |
|
|
455 | case SVt_PVHV: encode_hv (enc, (HV *)rv); break; |
|
|
456 | |
|
|
457 | default: |
|
|
458 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
|
|
459 | SvPV_nolen (sv)); |
|
|
460 | } |
|
|
461 | } |
|
|
462 | else if (!SvOK (sv)) |
|
|
463 | encode_str (enc, "null", 4, 0); |
829 | encode_str (enc, "null", 4, 0); |
464 | else |
830 | else |
465 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
831 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
466 | SvPV_nolen (sv), SvFLAGS (sv)); |
832 | SvPV_nolen (sv), SvFLAGS (sv)); |
467 | } |
833 | } |
468 | |
834 | |
469 | static SV * |
835 | static SV * |
470 | encode_json (SV *scalar, UV flags) |
836 | encode_json (SV *scalar, JSON *json) |
471 | { |
837 | { |
|
|
838 | enc_t enc; |
|
|
839 | |
472 | if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
840 | if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
473 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
841 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
474 | |
842 | |
475 | enc_t enc; |
843 | enc.json = *json; |
476 | enc.flags = flags; |
|
|
477 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
844 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
478 | enc.cur = SvPVX (enc.sv); |
845 | enc.cur = SvPVX (enc.sv); |
479 | enc.end = SvEND (enc.sv); |
846 | enc.end = SvEND (enc.sv); |
480 | enc.indent = 0; |
847 | enc.indent = 0; |
481 | enc.max_depth = 0x7fffffffUL; |
848 | enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
|
|
849 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
|
|
850 | : 0x110000UL; |
482 | |
851 | |
483 | SvPOK_only (enc.sv); |
852 | SvPOK_only (enc.sv); |
484 | encode_sv (&enc, scalar); |
853 | encode_sv (&enc, scalar); |
|
|
854 | encode_nl (&enc); |
485 | |
855 | |
|
|
856 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
|
|
857 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
|
|
858 | |
486 | if (!(flags & (F_ASCII | F_UTF8))) |
859 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
487 | SvUTF8_on (enc.sv); |
860 | SvUTF8_on (enc.sv); |
488 | |
861 | |
489 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
|
|
490 | |
|
|
491 | if (enc.flags & F_SHRINK) |
862 | if (enc.json.flags & F_SHRINK) |
492 | shrink (enc.sv); |
863 | shrink (enc.sv); |
493 | |
864 | |
494 | return enc.sv; |
865 | return enc.sv; |
495 | } |
866 | } |
496 | |
867 | |
… | |
… | |
501 | typedef struct |
872 | typedef struct |
502 | { |
873 | { |
503 | char *cur; // current parser pointer |
874 | char *cur; // current parser pointer |
504 | char *end; // end of input string |
875 | char *end; // end of input string |
505 | const char *err; // parse error, if != 0 |
876 | const char *err; // parse error, if != 0 |
506 | UV flags; // F_* |
877 | JSON json; |
|
|
878 | U32 depth; // recursion depth |
|
|
879 | U32 maxdepth; // recursion depth limit |
507 | } dec_t; |
880 | } dec_t; |
508 | |
881 | |
509 | static void |
882 | INLINE void |
|
|
883 | decode_comment (dec_t *dec) |
|
|
884 | { |
|
|
885 | // only '#'-style comments allowed a.t.m. |
|
|
886 | |
|
|
887 | while (*dec->cur && *dec->cur != 0x0a && *dec->cur != 0x0d) |
|
|
888 | ++dec->cur; |
|
|
889 | } |
|
|
890 | |
|
|
891 | INLINE void |
510 | decode_ws (dec_t *dec) |
892 | decode_ws (dec_t *dec) |
511 | { |
893 | { |
512 | for (;;) |
894 | for (;;) |
513 | { |
895 | { |
514 | char ch = *dec->cur; |
896 | char ch = *dec->cur; |
515 | |
897 | |
516 | if (ch > 0x20 |
898 | if (ch > 0x20) |
|
|
899 | { |
|
|
900 | if (expect_false (ch == '#')) |
|
|
901 | { |
|
|
902 | if (dec->json.flags & F_RELAXED) |
|
|
903 | decode_comment (dec); |
|
|
904 | else |
|
|
905 | break; |
|
|
906 | } |
|
|
907 | else |
|
|
908 | break; |
|
|
909 | } |
517 | || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) |
910 | else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09) |
518 | break; |
911 | break; // parse error, but let higher level handle it, gives better error messages |
519 | |
912 | |
520 | ++dec->cur; |
913 | ++dec->cur; |
521 | } |
914 | } |
522 | } |
915 | } |
523 | |
916 | |
524 | #define ERR(reason) SB dec->err = reason; goto fail; SE |
917 | #define ERR(reason) SB dec->err = reason; goto fail; SE |
|
|
918 | |
525 | #define EXPECT_CH(ch) SB \ |
919 | #define EXPECT_CH(ch) SB \ |
526 | if (*dec->cur != ch) \ |
920 | if (*dec->cur != ch) \ |
527 | ERR (# ch " expected"); \ |
921 | ERR (# ch " expected"); \ |
528 | ++dec->cur; \ |
922 | ++dec->cur; \ |
529 | SE |
923 | SE |
530 | |
924 | |
|
|
925 | #define DEC_INC_DEPTH if (++dec->depth > dec->json.max_depth) ERR (ERR_NESTING_EXCEEDED) |
|
|
926 | #define DEC_DEC_DEPTH --dec->depth |
|
|
927 | |
531 | static SV *decode_sv (dec_t *dec); |
928 | static SV *decode_sv (dec_t *dec); |
532 | |
929 | |
533 | static signed char decode_hexdigit[256]; |
930 | static signed char decode_hexdigit[256]; |
534 | |
931 | |
535 | static UV |
932 | static UV |
536 | decode_4hex (dec_t *dec) |
933 | decode_4hex (dec_t *dec) |
537 | { |
934 | { |
538 | signed char d1, d2, d3, d4; |
935 | signed char d1, d2, d3, d4; |
539 | unsigned char *cur = (unsigned char *)dec->cur; |
936 | unsigned char *cur = (unsigned char *)dec->cur; |
540 | |
937 | |
541 | d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); |
938 | d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected"); |
542 | d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); |
939 | d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected"); |
543 | d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); |
940 | d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected"); |
544 | d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); |
941 | d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected"); |
545 | |
942 | |
546 | dec->cur += 4; |
943 | dec->cur += 4; |
547 | |
944 | |
548 | return ((UV)d1) << 12 |
945 | return ((UV)d1) << 12 |
549 | | ((UV)d2) << 8 |
946 | | ((UV)d2) << 8 |
… | |
… | |
557 | static SV * |
954 | static SV * |
558 | decode_str (dec_t *dec) |
955 | decode_str (dec_t *dec) |
559 | { |
956 | { |
560 | SV *sv = 0; |
957 | SV *sv = 0; |
561 | int utf8 = 0; |
958 | int utf8 = 0; |
|
|
959 | char *dec_cur = dec->cur; |
562 | |
960 | |
563 | do |
961 | do |
564 | { |
962 | { |
565 | char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; |
963 | char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; |
566 | char *cur = buf; |
964 | char *cur = buf; |
567 | |
965 | |
568 | do |
966 | do |
569 | { |
967 | { |
570 | unsigned char ch = *(unsigned char *)dec->cur++; |
968 | unsigned char ch = *(unsigned char *)dec_cur++; |
571 | |
969 | |
572 | if (ch == '"') |
970 | if (expect_false (ch == '"')) |
573 | { |
971 | { |
574 | --dec->cur; |
972 | --dec_cur; |
575 | break; |
973 | break; |
576 | } |
974 | } |
577 | else if (ch == '\\') |
975 | else if (expect_false (ch == '\\')) |
578 | { |
976 | { |
579 | switch (*dec->cur) |
977 | switch (*dec_cur) |
580 | { |
978 | { |
581 | case '\\': |
979 | case '\\': |
582 | case '/': |
980 | case '/': |
583 | case '"': *cur++ = *dec->cur++; break; |
981 | case '"': *cur++ = *dec_cur++; break; |
584 | |
982 | |
585 | case 'b': ++dec->cur; *cur++ = '\010'; break; |
983 | case 'b': ++dec_cur; *cur++ = '\010'; break; |
586 | case 't': ++dec->cur; *cur++ = '\011'; break; |
984 | case 't': ++dec_cur; *cur++ = '\011'; break; |
587 | case 'n': ++dec->cur; *cur++ = '\012'; break; |
985 | case 'n': ++dec_cur; *cur++ = '\012'; break; |
588 | case 'f': ++dec->cur; *cur++ = '\014'; break; |
986 | case 'f': ++dec_cur; *cur++ = '\014'; break; |
589 | case 'r': ++dec->cur; *cur++ = '\015'; break; |
987 | case 'r': ++dec_cur; *cur++ = '\015'; break; |
590 | |
988 | |
591 | case 'u': |
989 | case 'u': |
592 | { |
990 | { |
593 | UV lo, hi; |
991 | UV lo, hi; |
594 | ++dec->cur; |
992 | ++dec_cur; |
595 | |
993 | |
|
|
994 | dec->cur = dec_cur; |
596 | hi = decode_4hex (dec); |
995 | hi = decode_4hex (dec); |
|
|
996 | dec_cur = dec->cur; |
597 | if (hi == (UV)-1) |
997 | if (hi == (UV)-1) |
598 | goto fail; |
998 | goto fail; |
599 | |
999 | |
600 | // possibly a surrogate pair |
1000 | // possibly a surrogate pair |
601 | if (hi >= 0xd800) |
1001 | if (hi >= 0xd800) |
602 | if (hi < 0xdc00) |
1002 | if (hi < 0xdc00) |
603 | { |
1003 | { |
604 | if (dec->cur [0] != '\\' || dec->cur [1] != 'u') |
1004 | if (dec_cur [0] != '\\' || dec_cur [1] != 'u') |
605 | ERR ("missing low surrogate character in surrogate pair"); |
1005 | ERR ("missing low surrogate character in surrogate pair"); |
606 | |
1006 | |
607 | dec->cur += 2; |
1007 | dec_cur += 2; |
608 | |
1008 | |
|
|
1009 | dec->cur = dec_cur; |
609 | lo = decode_4hex (dec); |
1010 | lo = decode_4hex (dec); |
|
|
1011 | dec_cur = dec->cur; |
610 | if (lo == (UV)-1) |
1012 | if (lo == (UV)-1) |
611 | goto fail; |
1013 | goto fail; |
612 | |
1014 | |
613 | if (lo < 0xdc00 || lo >= 0xe000) |
1015 | if (lo < 0xdc00 || lo >= 0xe000) |
614 | ERR ("surrogate pair expected"); |
1016 | ERR ("surrogate pair expected"); |
… | |
… | |
620 | |
1022 | |
621 | if (hi >= 0x80) |
1023 | if (hi >= 0x80) |
622 | { |
1024 | { |
623 | utf8 = 1; |
1025 | utf8 = 1; |
624 | |
1026 | |
625 | cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); |
1027 | cur = encode_utf8 (cur, hi); |
626 | } |
1028 | } |
627 | else |
1029 | else |
628 | *cur++ = hi; |
1030 | *cur++ = hi; |
629 | } |
1031 | } |
630 | break; |
1032 | break; |
631 | |
1033 | |
632 | default: |
1034 | default: |
633 | --dec->cur; |
1035 | --dec_cur; |
634 | ERR ("illegal backslash escape sequence in string"); |
1036 | ERR ("illegal backslash escape sequence in string"); |
635 | } |
1037 | } |
636 | } |
1038 | } |
637 | else if (ch >= 0x20 && ch <= 0x7f) |
1039 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
638 | *cur++ = ch; |
1040 | *cur++ = ch; |
639 | else if (ch >= 0x80) |
1041 | else if (ch >= 0x80) |
640 | { |
1042 | { |
641 | --dec->cur; |
|
|
642 | |
|
|
643 | STRLEN clen; |
1043 | STRLEN clen; |
|
|
1044 | |
|
|
1045 | --dec_cur; |
|
|
1046 | |
644 | UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); |
1047 | decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
645 | if (clen == (STRLEN)-1) |
1048 | if (clen == (STRLEN)-1) |
646 | ERR ("malformed UTF-8 character in JSON string"); |
1049 | ERR ("malformed UTF-8 character in JSON string"); |
647 | |
1050 | |
648 | do |
1051 | do |
649 | { |
|
|
650 | *cur++ = *dec->cur++; |
1052 | *cur++ = *dec_cur++; |
651 | } |
|
|
652 | while (--clen); |
1053 | while (--clen); |
653 | |
1054 | |
654 | utf8 = 1; |
1055 | utf8 = 1; |
655 | } |
1056 | } |
656 | else if (!ch) |
|
|
657 | ERR ("unexpected end of string while parsing json string"); |
|
|
658 | else |
1057 | else |
|
|
1058 | { |
|
|
1059 | --dec_cur; |
|
|
1060 | |
|
|
1061 | if (!ch) |
|
|
1062 | ERR ("unexpected end of string while parsing JSON string"); |
|
|
1063 | else |
659 | ERR ("invalid character encountered"); |
1064 | ERR ("invalid character encountered while parsing JSON string"); |
660 | |
1065 | } |
661 | } |
1066 | } |
662 | while (cur < buf + SHORT_STRING_LEN); |
1067 | while (cur < buf + SHORT_STRING_LEN); |
663 | |
1068 | |
|
|
1069 | { |
664 | STRLEN len = cur - buf; |
1070 | STRLEN len = cur - buf; |
665 | |
1071 | |
666 | if (sv) |
1072 | if (sv) |
667 | { |
1073 | { |
668 | SvGROW (sv, SvCUR (sv) + len + 1); |
1074 | STRLEN cur = SvCUR (sv); |
|
|
1075 | |
|
|
1076 | if (SvLEN (sv) <= cur + len) |
|
|
1077 | SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
|
|
1078 | |
669 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
1079 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
670 | SvCUR_set (sv, SvCUR (sv) + len); |
1080 | SvCUR_set (sv, SvCUR (sv) + len); |
671 | } |
1081 | } |
672 | else |
1082 | else |
673 | sv = newSVpvn (buf, len); |
1083 | sv = newSVpvn (buf, len); |
674 | } |
1084 | } |
|
|
1085 | } |
675 | while (*dec->cur != '"'); |
1086 | while (*dec_cur != '"'); |
676 | |
1087 | |
677 | ++dec->cur; |
1088 | ++dec_cur; |
678 | |
1089 | |
679 | if (sv) |
1090 | if (sv) |
680 | { |
1091 | { |
681 | SvPOK_only (sv); |
1092 | SvPOK_only (sv); |
682 | *SvEND (sv) = 0; |
1093 | *SvEND (sv) = 0; |
… | |
… | |
685 | SvUTF8_on (sv); |
1096 | SvUTF8_on (sv); |
686 | } |
1097 | } |
687 | else |
1098 | else |
688 | sv = newSVpvn ("", 0); |
1099 | sv = newSVpvn ("", 0); |
689 | |
1100 | |
|
|
1101 | dec->cur = dec_cur; |
690 | return sv; |
1102 | return sv; |
691 | |
1103 | |
692 | fail: |
1104 | fail: |
|
|
1105 | dec->cur = dec_cur; |
693 | return 0; |
1106 | return 0; |
694 | } |
1107 | } |
695 | |
1108 | |
696 | static SV * |
1109 | static SV * |
697 | decode_num (dec_t *dec) |
1110 | decode_num (dec_t *dec) |
… | |
… | |
755 | is_nv = 1; |
1168 | is_nv = 1; |
756 | } |
1169 | } |
757 | |
1170 | |
758 | if (!is_nv) |
1171 | if (!is_nv) |
759 | { |
1172 | { |
760 | UV uv; |
1173 | int len = dec->cur - start; |
761 | int numtype = grok_number (start, dec->cur - start, &uv); |
1174 | |
762 | if (numtype & IS_NUMBER_IN_UV) |
1175 | // special case the rather common 1..5-digit-int case |
763 | if (numtype & IS_NUMBER_NEG) |
1176 | if (*start == '-') |
|
|
1177 | switch (len) |
764 | { |
1178 | { |
765 | if (uv < (UV)IV_MIN) |
1179 | case 2: return newSViv (-(IV)( start [1] - '0' * 1)); |
766 | return newSViv (-(IV)uv); |
1180 | case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11)); |
|
|
1181 | case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
|
|
1182 | case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
|
|
1183 | case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); |
767 | } |
1184 | } |
|
|
1185 | else |
|
|
1186 | switch (len) |
|
|
1187 | { |
|
|
1188 | case 1: return newSViv ( start [0] - '0' * 1); |
|
|
1189 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
|
|
1190 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
|
|
1191 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
|
|
1192 | case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); |
|
|
1193 | } |
|
|
1194 | |
|
|
1195 | { |
|
|
1196 | UV uv; |
|
|
1197 | int numtype = grok_number (start, len, &uv); |
|
|
1198 | if (numtype & IS_NUMBER_IN_UV) |
|
|
1199 | if (numtype & IS_NUMBER_NEG) |
|
|
1200 | { |
|
|
1201 | if (uv < (UV)IV_MIN) |
|
|
1202 | return newSViv (-(IV)uv); |
|
|
1203 | } |
768 | else |
1204 | else |
769 | return newSVuv (uv); |
1205 | return newSVuv (uv); |
770 | } |
1206 | } |
771 | |
1207 | |
|
|
1208 | len -= *start == '-' ? 1 : 0; |
|
|
1209 | |
|
|
1210 | // does not fit into IV or UV, try NV |
|
|
1211 | if (len <= NV_DIG) |
|
|
1212 | // fits into NV without loss of precision |
|
|
1213 | return newSVnv (json_atof (start)); |
|
|
1214 | |
|
|
1215 | // everything else fails, convert it to a string |
|
|
1216 | return newSVpvn (start, dec->cur - start); |
|
|
1217 | } |
|
|
1218 | |
|
|
1219 | // loss of precision here |
772 | return newSVnv (Atof (start)); |
1220 | return newSVnv (json_atof (start)); |
773 | |
1221 | |
774 | fail: |
1222 | fail: |
775 | return 0; |
1223 | return 0; |
776 | } |
1224 | } |
777 | |
1225 | |
778 | static SV * |
1226 | static SV * |
779 | decode_av (dec_t *dec) |
1227 | decode_av (dec_t *dec) |
780 | { |
1228 | { |
781 | AV *av = newAV (); |
1229 | AV *av = newAV (); |
782 | |
1230 | |
|
|
1231 | DEC_INC_DEPTH; |
783 | decode_ws (dec); |
1232 | decode_ws (dec); |
|
|
1233 | |
784 | if (*dec->cur == ']') |
1234 | if (*dec->cur == ']') |
785 | ++dec->cur; |
1235 | ++dec->cur; |
786 | else |
1236 | else |
787 | for (;;) |
1237 | for (;;) |
788 | { |
1238 | { |
… | |
… | |
804 | |
1254 | |
805 | if (*dec->cur != ',') |
1255 | if (*dec->cur != ',') |
806 | ERR (", or ] expected while parsing array"); |
1256 | ERR (", or ] expected while parsing array"); |
807 | |
1257 | |
808 | ++dec->cur; |
1258 | ++dec->cur; |
|
|
1259 | |
|
|
1260 | decode_ws (dec); |
|
|
1261 | |
|
|
1262 | if (*dec->cur == ']' && dec->json.flags & F_RELAXED) |
|
|
1263 | { |
|
|
1264 | ++dec->cur; |
|
|
1265 | break; |
|
|
1266 | } |
809 | } |
1267 | } |
810 | |
1268 | |
|
|
1269 | DEC_DEC_DEPTH; |
811 | return newRV_noinc ((SV *)av); |
1270 | return newRV_noinc ((SV *)av); |
812 | |
1271 | |
813 | fail: |
1272 | fail: |
814 | SvREFCNT_dec (av); |
1273 | SvREFCNT_dec (av); |
|
|
1274 | DEC_DEC_DEPTH; |
815 | return 0; |
1275 | return 0; |
816 | } |
1276 | } |
817 | |
1277 | |
818 | static SV * |
1278 | static SV * |
819 | decode_hv (dec_t *dec) |
1279 | decode_hv (dec_t *dec) |
820 | { |
1280 | { |
|
|
1281 | SV *sv; |
821 | HV *hv = newHV (); |
1282 | HV *hv = newHV (); |
822 | |
1283 | |
|
|
1284 | DEC_INC_DEPTH; |
823 | decode_ws (dec); |
1285 | decode_ws (dec); |
|
|
1286 | |
824 | if (*dec->cur == '}') |
1287 | if (*dec->cur == '}') |
825 | ++dec->cur; |
1288 | ++dec->cur; |
826 | else |
1289 | else |
827 | for (;;) |
1290 | for (;;) |
828 | { |
1291 | { |
|
|
1292 | EXPECT_CH ('"'); |
|
|
1293 | |
|
|
1294 | // heuristic: assume that |
|
|
1295 | // a) decode_str + hv_store_ent are abysmally slow. |
|
|
1296 | // b) most hash keys are short, simple ascii text. |
|
|
1297 | // => try to "fast-match" such strings to avoid |
|
|
1298 | // the overhead of decode_str + hv_store_ent. |
|
|
1299 | { |
829 | SV *key, *value; |
1300 | SV *value; |
|
|
1301 | char *p = dec->cur; |
|
|
1302 | char *e = p + 24; // only try up to 24 bytes |
830 | |
1303 | |
831 | decode_ws (dec); EXPECT_CH ('"'); |
1304 | for (;;) |
832 | |
|
|
833 | key = decode_str (dec); |
|
|
834 | if (!key) |
|
|
835 | goto fail; |
|
|
836 | |
|
|
837 | decode_ws (dec); EXPECT_CH (':'); |
|
|
838 | |
|
|
839 | value = decode_sv (dec); |
|
|
840 | if (!value) |
|
|
841 | { |
1305 | { |
|
|
1306 | // the >= 0x80 is false on most architectures |
|
|
1307 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
|
|
1308 | { |
|
|
1309 | // slow path, back up and use decode_str |
|
|
1310 | SV *key = decode_str (dec); |
|
|
1311 | if (!key) |
|
|
1312 | goto fail; |
|
|
1313 | |
|
|
1314 | decode_ws (dec); EXPECT_CH (':'); |
|
|
1315 | |
|
|
1316 | decode_ws (dec); |
|
|
1317 | value = decode_sv (dec); |
|
|
1318 | if (!value) |
|
|
1319 | { |
|
|
1320 | SvREFCNT_dec (key); |
|
|
1321 | goto fail; |
|
|
1322 | } |
|
|
1323 | |
|
|
1324 | hv_store_ent (hv, key, value, 0); |
842 | SvREFCNT_dec (key); |
1325 | SvREFCNT_dec (key); |
|
|
1326 | |
|
|
1327 | break; |
|
|
1328 | } |
|
|
1329 | else if (*p == '"') |
|
|
1330 | { |
|
|
1331 | // fast path, got a simple key |
|
|
1332 | char *key = dec->cur; |
|
|
1333 | int len = p - key; |
|
|
1334 | dec->cur = p + 1; |
|
|
1335 | |
|
|
1336 | decode_ws (dec); EXPECT_CH (':'); |
|
|
1337 | |
|
|
1338 | decode_ws (dec); |
|
|
1339 | value = decode_sv (dec); |
|
|
1340 | if (!value) |
843 | goto fail; |
1341 | goto fail; |
|
|
1342 | |
|
|
1343 | hv_store (hv, key, len, value, 0); |
|
|
1344 | |
|
|
1345 | break; |
|
|
1346 | } |
|
|
1347 | |
|
|
1348 | ++p; |
844 | } |
1349 | } |
845 | |
1350 | } |
846 | //TODO: optimise |
|
|
847 | hv_store_ent (hv, key, value, 0); |
|
|
848 | |
1351 | |
849 | decode_ws (dec); |
1352 | decode_ws (dec); |
850 | |
1353 | |
851 | if (*dec->cur == '}') |
1354 | if (*dec->cur == '}') |
852 | { |
1355 | { |
… | |
… | |
856 | |
1359 | |
857 | if (*dec->cur != ',') |
1360 | if (*dec->cur != ',') |
858 | ERR (", or } expected while parsing object/hash"); |
1361 | ERR (", or } expected while parsing object/hash"); |
859 | |
1362 | |
860 | ++dec->cur; |
1363 | ++dec->cur; |
|
|
1364 | |
|
|
1365 | decode_ws (dec); |
|
|
1366 | |
|
|
1367 | if (*dec->cur == '}' && dec->json.flags & F_RELAXED) |
|
|
1368 | { |
|
|
1369 | ++dec->cur; |
|
|
1370 | break; |
|
|
1371 | } |
861 | } |
1372 | } |
862 | |
1373 | |
|
|
1374 | DEC_DEC_DEPTH; |
863 | return newRV_noinc ((SV *)hv); |
1375 | sv = newRV_noinc ((SV *)hv); |
|
|
1376 | |
|
|
1377 | // check filter callbacks |
|
|
1378 | if (dec->json.flags & F_HOOK) |
|
|
1379 | { |
|
|
1380 | if (dec->json.cb_sk_object && HvKEYS (hv) == 1) |
|
|
1381 | { |
|
|
1382 | HE *cb, *he; |
|
|
1383 | |
|
|
1384 | hv_iterinit (hv); |
|
|
1385 | he = hv_iternext (hv); |
|
|
1386 | hv_iterinit (hv); |
|
|
1387 | |
|
|
1388 | // the next line creates a mortal sv each time its called. |
|
|
1389 | // might want to optimise this for common cases. |
|
|
1390 | cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); |
|
|
1391 | |
|
|
1392 | if (cb) |
|
|
1393 | { |
|
|
1394 | dSP; |
|
|
1395 | int count; |
|
|
1396 | |
|
|
1397 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
1398 | XPUSHs (HeVAL (he)); |
|
|
1399 | |
|
|
1400 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
|
|
1401 | |
|
|
1402 | if (count == 1) |
|
|
1403 | { |
|
|
1404 | sv = newSVsv (POPs); |
|
|
1405 | FREETMPS; LEAVE; |
|
|
1406 | return sv; |
|
|
1407 | } |
|
|
1408 | |
|
|
1409 | FREETMPS; LEAVE; |
|
|
1410 | } |
|
|
1411 | } |
|
|
1412 | |
|
|
1413 | if (dec->json.cb_object) |
|
|
1414 | { |
|
|
1415 | dSP; |
|
|
1416 | int count; |
|
|
1417 | |
|
|
1418 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
1419 | XPUSHs (sv_2mortal (sv)); |
|
|
1420 | |
|
|
1421 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
|
|
1422 | |
|
|
1423 | if (count == 1) |
|
|
1424 | { |
|
|
1425 | sv = newSVsv (POPs); |
|
|
1426 | FREETMPS; LEAVE; |
|
|
1427 | return sv; |
|
|
1428 | } |
|
|
1429 | |
|
|
1430 | SvREFCNT_inc (sv); |
|
|
1431 | FREETMPS; LEAVE; |
|
|
1432 | } |
|
|
1433 | } |
|
|
1434 | |
|
|
1435 | return sv; |
864 | |
1436 | |
865 | fail: |
1437 | fail: |
866 | SvREFCNT_dec (hv); |
1438 | SvREFCNT_dec (hv); |
|
|
1439 | DEC_DEC_DEPTH; |
867 | return 0; |
1440 | return 0; |
868 | } |
1441 | } |
869 | |
1442 | |
870 | static SV * |
1443 | static SV * |
871 | decode_sv (dec_t *dec) |
1444 | decode_sv (dec_t *dec) |
872 | { |
1445 | { |
873 | decode_ws (dec); |
1446 | // the beauty of JSON: you need exactly one character lookahead |
|
|
1447 | // to parse everything. |
874 | switch (*dec->cur) |
1448 | switch (*dec->cur) |
875 | { |
1449 | { |
876 | case '"': ++dec->cur; return decode_str (dec); |
1450 | case '"': ++dec->cur; return decode_str (dec); |
877 | case '[': ++dec->cur; return decode_av (dec); |
1451 | case '[': ++dec->cur; return decode_av (dec); |
878 | case '{': ++dec->cur; return decode_hv (dec); |
1452 | case '{': ++dec->cur; return decode_hv (dec); |
879 | |
1453 | |
880 | case '-': |
1454 | case '-': |
881 | case '0': case '1': case '2': case '3': case '4': |
1455 | case '0': case '1': case '2': case '3': case '4': |
882 | case '5': case '6': case '7': case '8': case '9': |
1456 | case '5': case '6': case '7': case '8': case '9': |
883 | return decode_num (dec); |
1457 | return decode_num (dec); |
884 | |
1458 | |
885 | case 't': |
1459 | case 't': |
886 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1460 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
887 | { |
1461 | { |
888 | dec->cur += 4; |
1462 | dec->cur += 4; |
|
|
1463 | #if JSON_SLOW |
|
|
1464 | json_true = get_bool ("JSON::XS::true"); |
|
|
1465 | #endif |
889 | return newSViv (1); |
1466 | return newSVsv (json_true); |
890 | } |
1467 | } |
891 | else |
1468 | else |
892 | ERR ("'true' expected"); |
1469 | ERR ("'true' expected"); |
893 | |
1470 | |
894 | break; |
1471 | break; |
895 | |
1472 | |
896 | case 'f': |
1473 | case 'f': |
897 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1474 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
898 | { |
1475 | { |
899 | dec->cur += 5; |
1476 | dec->cur += 5; |
|
|
1477 | #if JSON_SLOW |
|
|
1478 | json_false = get_bool ("JSON::XS::false"); |
|
|
1479 | #endif |
900 | return newSViv (0); |
1480 | return newSVsv (json_false); |
901 | } |
1481 | } |
902 | else |
1482 | else |
903 | ERR ("'false' expected"); |
1483 | ERR ("'false' expected"); |
904 | |
1484 | |
905 | break; |
1485 | break; |
… | |
… | |
914 | ERR ("'null' expected"); |
1494 | ERR ("'null' expected"); |
915 | |
1495 | |
916 | break; |
1496 | break; |
917 | |
1497 | |
918 | default: |
1498 | default: |
919 | ERR ("malformed json string, neither array, object, number, string or atom"); |
1499 | ERR ("malformed JSON string, neither array, object, number, string or atom"); |
920 | break; |
1500 | break; |
921 | } |
1501 | } |
922 | |
1502 | |
923 | fail: |
1503 | fail: |
924 | return 0; |
1504 | return 0; |
925 | } |
1505 | } |
926 | |
1506 | |
927 | static SV * |
1507 | static SV * |
928 | decode_json (SV *string, UV flags) |
1508 | decode_json (SV *string, JSON *json, char **offset_return) |
929 | { |
1509 | { |
|
|
1510 | dec_t dec; |
930 | SV *sv; |
1511 | SV *sv; |
931 | |
1512 | |
|
|
1513 | /* work around bugs in 5.10 where manipulating magic values |
|
|
1514 | * will perl ignore the magic in subsequent accesses |
|
|
1515 | */ |
|
|
1516 | /*SvGETMAGIC (string);*/ |
|
|
1517 | if (SvMAGICAL (string)) |
|
|
1518 | string = sv_2mortal (newSVsv (string)); |
|
|
1519 | |
|
|
1520 | SvUPGRADE (string, SVt_PV); |
|
|
1521 | |
|
|
1522 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
|
|
1523 | * assertion with -DDEBUGGING, although SvCUR is documented to |
|
|
1524 | * return the xpv_cur field which certainly exists after upgrading. |
|
|
1525 | * according to nicholas clark, calling SvPOK fixes this. |
|
|
1526 | * But it doesn't fix it, so try another workaround, call SvPV_nolen |
|
|
1527 | * and hope for the best. |
|
|
1528 | * Damnit, SvPV_nolen still trips over yet another assertion. This |
|
|
1529 | * assertion business is seriously broken, try yet another workaround |
|
|
1530 | * for the broken -DDEBUGGING. |
|
|
1531 | */ |
|
|
1532 | { |
|
|
1533 | #ifdef DEBUGGING |
|
|
1534 | STRLEN offset = SvOK (string) ? sv_len (string) : 0; |
|
|
1535 | #else |
|
|
1536 | STRLEN offset = SvCUR (string); |
|
|
1537 | #endif |
|
|
1538 | |
|
|
1539 | if (offset > json->max_size && json->max_size) |
|
|
1540 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
|
|
1541 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
|
|
1542 | } |
|
|
1543 | |
932 | if (flags & F_UTF8) |
1544 | if (json->flags & F_UTF8) |
933 | sv_utf8_downgrade (string, 0); |
1545 | sv_utf8_downgrade (string, 0); |
934 | else |
1546 | else |
935 | sv_utf8_upgrade (string); |
1547 | sv_utf8_upgrade (string); |
936 | |
1548 | |
937 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1549 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
938 | |
1550 | |
939 | dec_t dec; |
1551 | dec.json = *json; |
940 | dec.flags = flags; |
|
|
941 | dec.cur = SvPVX (string); |
1552 | dec.cur = SvPVX (string); |
942 | dec.end = SvEND (string); |
1553 | dec.end = SvEND (string); |
943 | dec.err = 0; |
1554 | dec.err = 0; |
|
|
1555 | dec.depth = 0; |
944 | |
1556 | |
|
|
1557 | if (dec.json.cb_object || dec.json.cb_sk_object) |
|
|
1558 | dec.json.flags |= F_HOOK; |
|
|
1559 | |
|
|
1560 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
|
|
1561 | |
|
|
1562 | decode_ws (&dec); |
945 | sv = decode_sv (&dec); |
1563 | sv = decode_sv (&dec); |
946 | |
1564 | |
|
|
1565 | if (offset_return) |
|
|
1566 | *offset_return = dec.cur; |
|
|
1567 | |
|
|
1568 | if (!(offset_return || !sv)) |
|
|
1569 | { |
|
|
1570 | // check for trailing garbage |
|
|
1571 | decode_ws (&dec); |
|
|
1572 | |
|
|
1573 | if (*dec.cur) |
|
|
1574 | { |
|
|
1575 | dec.err = "garbage after JSON object"; |
|
|
1576 | SvREFCNT_dec (sv); |
|
|
1577 | sv = 0; |
|
|
1578 | } |
|
|
1579 | } |
|
|
1580 | |
947 | if (!sv) |
1581 | if (!sv) |
948 | { |
1582 | { |
949 | IV offset = dec.flags & F_UTF8 |
|
|
950 | ? dec.cur - SvPVX (string) |
|
|
951 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
952 | SV *uni = sv_newmortal (); |
1583 | SV *uni = sv_newmortal (); |
953 | |
1584 | |
954 | // horrible hack to silence warning inside pv_uni_display |
1585 | // horrible hack to silence warning inside pv_uni_display |
955 | COP cop = *PL_curcop; |
1586 | COP cop = *PL_curcop; |
956 | cop.cop_warnings = pWARN_NONE; |
1587 | cop.cop_warnings = pWARN_NONE; |
… | |
… | |
958 | SAVEVPTR (PL_curcop); |
1589 | SAVEVPTR (PL_curcop); |
959 | PL_curcop = &cop; |
1590 | PL_curcop = &cop; |
960 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1591 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
961 | LEAVE; |
1592 | LEAVE; |
962 | |
1593 | |
963 | croak ("%s, at character offset %d (%s)", |
1594 | croak ("%s, at character offset %d (before \"%s\")", |
964 | dec.err, |
1595 | dec.err, |
965 | (int)offset, |
1596 | ptr_to_index (string, dec.cur), |
966 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1597 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
967 | } |
1598 | } |
968 | |
1599 | |
969 | sv = sv_2mortal (sv); |
1600 | sv = sv_2mortal (sv); |
970 | |
1601 | |
971 | if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1602 | if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
972 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1603 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
973 | |
1604 | |
974 | return sv; |
1605 | return sv; |
|
|
1606 | } |
|
|
1607 | |
|
|
1608 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1609 | // incremental parser |
|
|
1610 | |
|
|
1611 | static void |
|
|
1612 | incr_parse (JSON *self) |
|
|
1613 | { |
|
|
1614 | const char *p = SvPVX (self->incr_text) + self->incr_pos; |
|
|
1615 | |
|
|
1616 | // the state machine here is a bit convoluted and could be simplified a lot |
|
|
1617 | // but this would make it slower, so... |
|
|
1618 | |
|
|
1619 | for (;;) |
|
|
1620 | { |
|
|
1621 | //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 |
|
|
1622 | switch (self->incr_mode) |
|
|
1623 | { |
|
|
1624 | // only used for initial whitespace skipping |
|
|
1625 | case INCR_M_WS: |
|
|
1626 | for (;;) |
|
|
1627 | { |
|
|
1628 | if (*p > 0x20) |
|
|
1629 | { |
|
|
1630 | if (*p == '#') |
|
|
1631 | { |
|
|
1632 | self->incr_mode = INCR_M_C0; |
|
|
1633 | goto incr_m_c; |
|
|
1634 | } |
|
|
1635 | else |
|
|
1636 | { |
|
|
1637 | self->incr_mode = INCR_M_JSON; |
|
|
1638 | goto incr_m_json; |
|
|
1639 | } |
|
|
1640 | } |
|
|
1641 | else if (!*p) |
|
|
1642 | goto interrupt; |
|
|
1643 | |
|
|
1644 | ++p; |
|
|
1645 | } |
|
|
1646 | |
|
|
1647 | // skip a single char inside a string (for \\-processing) |
|
|
1648 | case INCR_M_BS: |
|
|
1649 | if (!*p) |
|
|
1650 | goto interrupt; |
|
|
1651 | |
|
|
1652 | ++p; |
|
|
1653 | self->incr_mode = INCR_M_STR; |
|
|
1654 | goto incr_m_str; |
|
|
1655 | |
|
|
1656 | // inside #-style comments |
|
|
1657 | case INCR_M_C0: |
|
|
1658 | case INCR_M_C1: |
|
|
1659 | incr_m_c: |
|
|
1660 | for (;;) |
|
|
1661 | { |
|
|
1662 | if (*p == '\n') |
|
|
1663 | { |
|
|
1664 | self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON; |
|
|
1665 | break; |
|
|
1666 | } |
|
|
1667 | else if (!*p) |
|
|
1668 | goto interrupt; |
|
|
1669 | |
|
|
1670 | ++p; |
|
|
1671 | } |
|
|
1672 | |
|
|
1673 | break; |
|
|
1674 | |
|
|
1675 | // inside a string |
|
|
1676 | case INCR_M_STR: |
|
|
1677 | incr_m_str: |
|
|
1678 | for (;;) |
|
|
1679 | { |
|
|
1680 | if (*p == '"') |
|
|
1681 | { |
|
|
1682 | ++p; |
|
|
1683 | self->incr_mode = INCR_M_JSON; |
|
|
1684 | |
|
|
1685 | if (!self->incr_nest) |
|
|
1686 | goto interrupt; |
|
|
1687 | |
|
|
1688 | goto incr_m_json; |
|
|
1689 | } |
|
|
1690 | else if (*p == '\\') |
|
|
1691 | { |
|
|
1692 | ++p; // "virtually" consumes character after \ |
|
|
1693 | |
|
|
1694 | if (!*p) // if at end of string we have to switch modes |
|
|
1695 | { |
|
|
1696 | self->incr_mode = INCR_M_BS; |
|
|
1697 | goto interrupt; |
|
|
1698 | } |
|
|
1699 | } |
|
|
1700 | else if (!*p) |
|
|
1701 | goto interrupt; |
|
|
1702 | |
|
|
1703 | ++p; |
|
|
1704 | } |
|
|
1705 | |
|
|
1706 | // after initial ws, outside string |
|
|
1707 | case INCR_M_JSON: |
|
|
1708 | incr_m_json: |
|
|
1709 | for (;;) |
|
|
1710 | { |
|
|
1711 | switch (*p++) |
|
|
1712 | { |
|
|
1713 | case 0: |
|
|
1714 | --p; |
|
|
1715 | goto interrupt; |
|
|
1716 | |
|
|
1717 | case 0x09: |
|
|
1718 | case 0x0a: |
|
|
1719 | case 0x0d: |
|
|
1720 | case 0x20: |
|
|
1721 | if (!self->incr_nest) |
|
|
1722 | { |
|
|
1723 | --p; // do not eat the whitespace, let the next round do it |
|
|
1724 | goto interrupt; |
|
|
1725 | } |
|
|
1726 | break; |
|
|
1727 | |
|
|
1728 | case '"': |
|
|
1729 | self->incr_mode = INCR_M_STR; |
|
|
1730 | goto incr_m_str; |
|
|
1731 | |
|
|
1732 | case '[': |
|
|
1733 | case '{': |
|
|
1734 | if (++self->incr_nest > self->max_depth) |
|
|
1735 | croak (ERR_NESTING_EXCEEDED); |
|
|
1736 | break; |
|
|
1737 | |
|
|
1738 | case ']': |
|
|
1739 | case '}': |
|
|
1740 | if (--self->incr_nest <= 0) |
|
|
1741 | goto interrupt; |
|
|
1742 | break; |
|
|
1743 | |
|
|
1744 | case '#': |
|
|
1745 | self->incr_mode = INCR_M_C1; |
|
|
1746 | goto incr_m_c; |
|
|
1747 | } |
|
|
1748 | } |
|
|
1749 | } |
|
|
1750 | |
|
|
1751 | modechange: |
|
|
1752 | ; |
|
|
1753 | } |
|
|
1754 | |
|
|
1755 | interrupt: |
|
|
1756 | self->incr_pos = p - SvPVX (self->incr_text); |
|
|
1757 | //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D |
|
|
1758 | //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D |
975 | } |
1759 | } |
976 | |
1760 | |
977 | ///////////////////////////////////////////////////////////////////////////// |
1761 | ///////////////////////////////////////////////////////////////////////////// |
978 | // XS interface functions |
1762 | // XS interface functions |
979 | |
1763 | |
… | |
… | |
981 | |
1765 | |
982 | BOOT: |
1766 | BOOT: |
983 | { |
1767 | { |
984 | int i; |
1768 | int i; |
985 | |
1769 | |
986 | memset (decode_hexdigit, 0xff, 256); |
|
|
987 | for (i = 10; i--; ) |
1770 | for (i = 0; i < 256; ++i) |
988 | decode_hexdigit ['0' + i] = i; |
1771 | decode_hexdigit [i] = |
|
|
1772 | i >= '0' && i <= '9' ? i - '0' |
|
|
1773 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
|
|
1774 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
|
|
1775 | : -1; |
989 | |
1776 | |
990 | for (i = 7; i--; ) |
|
|
991 | { |
|
|
992 | decode_hexdigit ['a' + i] = 10 + i; |
|
|
993 | decode_hexdigit ['A' + i] = 10 + i; |
|
|
994 | } |
|
|
995 | |
|
|
996 | json_stash = gv_stashpv ("JSON::XS", 1); |
1777 | json_stash = gv_stashpv ("JSON::XS" , 1); |
|
|
1778 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
|
|
1779 | |
|
|
1780 | json_true = get_bool ("JSON::XS::true"); |
|
|
1781 | json_false = get_bool ("JSON::XS::false"); |
|
|
1782 | |
|
|
1783 | CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ |
997 | } |
1784 | } |
998 | |
1785 | |
999 | PROTOTYPES: DISABLE |
1786 | PROTOTYPES: DISABLE |
1000 | |
1787 | |
1001 | SV *new (char *dummy) |
1788 | void CLONE (...) |
1002 | CODE: |
1789 | CODE: |
1003 | RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); |
1790 | json_stash = 0; |
|
|
1791 | json_boolean_stash = 0; |
|
|
1792 | |
|
|
1793 | void new (char *klass) |
|
|
1794 | PPCODE: |
|
|
1795 | { |
|
|
1796 | SV *pv = NEWSV (0, sizeof (JSON)); |
|
|
1797 | SvPOK_only (pv); |
|
|
1798 | json_init ((JSON *)SvPVX (pv)); |
|
|
1799 | XPUSHs (sv_2mortal (sv_bless ( |
|
|
1800 | newRV_noinc (pv), |
|
|
1801 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
|
|
1802 | ))); |
|
|
1803 | } |
|
|
1804 | |
|
|
1805 | void ascii (JSON *self, int enable = 1) |
|
|
1806 | ALIAS: |
|
|
1807 | ascii = F_ASCII |
|
|
1808 | latin1 = F_LATIN1 |
|
|
1809 | utf8 = F_UTF8 |
|
|
1810 | indent = F_INDENT |
|
|
1811 | canonical = F_CANONICAL |
|
|
1812 | space_before = F_SPACE_BEFORE |
|
|
1813 | space_after = F_SPACE_AFTER |
|
|
1814 | pretty = F_PRETTY |
|
|
1815 | allow_nonref = F_ALLOW_NONREF |
|
|
1816 | shrink = F_SHRINK |
|
|
1817 | allow_blessed = F_ALLOW_BLESSED |
|
|
1818 | convert_blessed = F_CONV_BLESSED |
|
|
1819 | relaxed = F_RELAXED |
|
|
1820 | allow_unknown = F_ALLOW_UNKNOWN |
|
|
1821 | PPCODE: |
|
|
1822 | { |
|
|
1823 | if (enable) |
|
|
1824 | self->flags |= ix; |
|
|
1825 | else |
|
|
1826 | self->flags &= ~ix; |
|
|
1827 | |
|
|
1828 | XPUSHs (ST (0)); |
|
|
1829 | } |
|
|
1830 | |
|
|
1831 | void get_ascii (JSON *self) |
|
|
1832 | ALIAS: |
|
|
1833 | get_ascii = F_ASCII |
|
|
1834 | get_latin1 = F_LATIN1 |
|
|
1835 | get_utf8 = F_UTF8 |
|
|
1836 | get_indent = F_INDENT |
|
|
1837 | get_canonical = F_CANONICAL |
|
|
1838 | get_space_before = F_SPACE_BEFORE |
|
|
1839 | get_space_after = F_SPACE_AFTER |
|
|
1840 | get_allow_nonref = F_ALLOW_NONREF |
|
|
1841 | get_shrink = F_SHRINK |
|
|
1842 | get_allow_blessed = F_ALLOW_BLESSED |
|
|
1843 | get_convert_blessed = F_CONV_BLESSED |
|
|
1844 | get_relaxed = F_RELAXED |
|
|
1845 | get_allow_unknown = F_ALLOW_UNKNOWN |
|
|
1846 | PPCODE: |
|
|
1847 | XPUSHs (boolSV (self->flags & ix)); |
|
|
1848 | |
|
|
1849 | void max_depth (JSON *self, U32 max_depth = 0x80000000UL) |
|
|
1850 | PPCODE: |
|
|
1851 | self->max_depth = max_depth; |
|
|
1852 | XPUSHs (ST (0)); |
|
|
1853 | |
|
|
1854 | U32 get_max_depth (JSON *self) |
|
|
1855 | CODE: |
|
|
1856 | RETVAL = self->max_depth; |
1004 | OUTPUT: |
1857 | OUTPUT: |
1005 | RETVAL |
1858 | RETVAL |
1006 | |
1859 | |
1007 | SV *ascii (SV *self, int enable = 1) |
1860 | void max_size (JSON *self, U32 max_size = 0) |
1008 | ALIAS: |
1861 | PPCODE: |
1009 | ascii = F_ASCII |
1862 | self->max_size = max_size; |
1010 | utf8 = F_UTF8 |
1863 | XPUSHs (ST (0)); |
1011 | indent = F_INDENT |
1864 | |
1012 | canonical = F_CANONICAL |
1865 | int get_max_size (JSON *self) |
1013 | space_before = F_SPACE_BEFORE |
|
|
1014 | space_after = F_SPACE_AFTER |
|
|
1015 | pretty = F_PRETTY |
|
|
1016 | allow_nonref = F_ALLOW_NONREF |
|
|
1017 | shrink = F_SHRINK |
|
|
1018 | CODE: |
1866 | CODE: |
1019 | { |
1867 | RETVAL = self->max_size; |
1020 | UV *uv = SvJSON (self); |
|
|
1021 | if (enable) |
|
|
1022 | *uv |= ix; |
|
|
1023 | else |
|
|
1024 | *uv &= ~ix; |
|
|
1025 | |
|
|
1026 | RETVAL = newSVsv (self); |
|
|
1027 | } |
|
|
1028 | OUTPUT: |
1868 | OUTPUT: |
1029 | RETVAL |
1869 | RETVAL |
1030 | |
1870 | |
1031 | void encode (SV *self, SV *scalar) |
1871 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
1032 | PPCODE: |
1872 | PPCODE: |
1033 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
1873 | { |
|
|
1874 | SvREFCNT_dec (self->cb_object); |
|
|
1875 | self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; |
1034 | |
1876 | |
1035 | void decode (SV *self, SV *jsonstr) |
1877 | XPUSHs (ST (0)); |
|
|
1878 | } |
|
|
1879 | |
|
|
1880 | void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) |
1036 | PPCODE: |
1881 | PPCODE: |
|
|
1882 | { |
|
|
1883 | if (!self->cb_sk_object) |
|
|
1884 | self->cb_sk_object = newHV (); |
|
|
1885 | |
|
|
1886 | if (SvOK (cb)) |
|
|
1887 | hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
|
|
1888 | else |
|
|
1889 | { |
|
|
1890 | hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0); |
|
|
1891 | |
|
|
1892 | if (!HvKEYS (self->cb_sk_object)) |
|
|
1893 | { |
|
|
1894 | SvREFCNT_dec (self->cb_sk_object); |
|
|
1895 | self->cb_sk_object = 0; |
|
|
1896 | } |
|
|
1897 | } |
|
|
1898 | |
|
|
1899 | XPUSHs (ST (0)); |
|
|
1900 | } |
|
|
1901 | |
|
|
1902 | void encode (JSON *self, SV *scalar) |
|
|
1903 | PPCODE: |
|
|
1904 | XPUSHs (encode_json (scalar, self)); |
|
|
1905 | |
|
|
1906 | void decode (JSON *self, SV *jsonstr) |
|
|
1907 | PPCODE: |
1037 | XPUSHs (decode_json (jsonstr, *SvJSON (self))); |
1908 | XPUSHs (decode_json (jsonstr, self, 0)); |
|
|
1909 | |
|
|
1910 | void decode_prefix (JSON *self, SV *jsonstr) |
|
|
1911 | PPCODE: |
|
|
1912 | { |
|
|
1913 | char *offset; |
|
|
1914 | EXTEND (SP, 2); |
|
|
1915 | PUSHs (decode_json (jsonstr, self, &offset)); |
|
|
1916 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); |
|
|
1917 | } |
|
|
1918 | |
|
|
1919 | void incr_parse (JSON *self, SV *jsonstr = 0) |
|
|
1920 | PPCODE: |
|
|
1921 | { |
|
|
1922 | if (!self->incr_text) |
|
|
1923 | self->incr_text = newSVpvn ("", 0); |
|
|
1924 | |
|
|
1925 | // append data, if any |
|
|
1926 | if (jsonstr) |
|
|
1927 | { |
|
|
1928 | if (SvUTF8 (jsonstr)) |
|
|
1929 | { |
|
|
1930 | if (!SvUTF8 (self->incr_text)) |
|
|
1931 | { |
|
|
1932 | /* utf-8-ness differs, need to upgrade */ |
|
|
1933 | sv_utf8_upgrade (self->incr_text); |
|
|
1934 | |
|
|
1935 | if (self->incr_pos) |
|
|
1936 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
|
|
1937 | - (U8 *)SvPVX (self->incr_text); |
|
|
1938 | } |
|
|
1939 | } |
|
|
1940 | else if (SvUTF8 (self->incr_text)) |
|
|
1941 | sv_utf8_upgrade (jsonstr); |
|
|
1942 | |
|
|
1943 | { |
|
|
1944 | STRLEN len; |
|
|
1945 | const char *str = SvPV (jsonstr, len); |
|
|
1946 | STRLEN cur = SvCUR (self->incr_text); |
|
|
1947 | |
|
|
1948 | if (SvLEN (self->incr_text) <= cur + len) |
|
|
1949 | SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
|
|
1950 | |
|
|
1951 | Move (str, SvEND (self->incr_text), len, char); |
|
|
1952 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
|
|
1953 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
|
|
1954 | } |
|
|
1955 | } |
|
|
1956 | |
|
|
1957 | if (GIMME_V != G_VOID) |
|
|
1958 | do |
|
|
1959 | { |
|
|
1960 | char *offset; |
|
|
1961 | |
|
|
1962 | if (!INCR_DONE (self)) |
|
|
1963 | { |
|
|
1964 | incr_parse (self); |
|
|
1965 | |
|
|
1966 | if (self->incr_pos > self->max_size && self->max_size) |
|
|
1967 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
|
|
1968 | (unsigned long)self->incr_pos, (unsigned long)self->max_size); |
|
|
1969 | |
|
|
1970 | if (!INCR_DONE (self)) |
|
|
1971 | break; |
|
|
1972 | } |
|
|
1973 | |
|
|
1974 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
|
|
1975 | |
|
|
1976 | self->incr_pos -= offset - SvPVX (self->incr_text); |
|
|
1977 | self->incr_nest = 0; |
|
|
1978 | self->incr_mode = 0; |
|
|
1979 | |
|
|
1980 | sv_chop (self->incr_text, offset); |
|
|
1981 | } |
|
|
1982 | while (GIMME_V == G_ARRAY); |
|
|
1983 | } |
|
|
1984 | |
|
|
1985 | SV *incr_text (JSON *self) |
|
|
1986 | ATTRS: lvalue |
|
|
1987 | CODE: |
|
|
1988 | { |
|
|
1989 | if (self->incr_pos) |
|
|
1990 | croak ("incr_text can not be called when the incremental parser already started parsing"); |
|
|
1991 | |
|
|
1992 | RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef; |
|
|
1993 | } |
|
|
1994 | OUTPUT: |
|
|
1995 | RETVAL |
|
|
1996 | |
|
|
1997 | void incr_skip (JSON *self) |
|
|
1998 | CODE: |
|
|
1999 | { |
|
|
2000 | if (self->incr_pos) |
|
|
2001 | { |
|
|
2002 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos); |
|
|
2003 | self->incr_pos = 0; |
|
|
2004 | self->incr_nest = 0; |
|
|
2005 | self->incr_mode = 0; |
|
|
2006 | } |
|
|
2007 | } |
|
|
2008 | |
|
|
2009 | void incr_reset (JSON *self) |
|
|
2010 | CODE: |
|
|
2011 | { |
|
|
2012 | SvREFCNT_dec (self->incr_text); |
|
|
2013 | self->incr_text = 0; |
|
|
2014 | self->incr_pos = 0; |
|
|
2015 | self->incr_nest = 0; |
|
|
2016 | self->incr_mode = 0; |
|
|
2017 | } |
|
|
2018 | |
|
|
2019 | void DESTROY (JSON *self) |
|
|
2020 | CODE: |
|
|
2021 | SvREFCNT_dec (self->cb_sk_object); |
|
|
2022 | SvREFCNT_dec (self->cb_object); |
|
|
2023 | SvREFCNT_dec (self->incr_text); |
1038 | |
2024 | |
1039 | PROTOTYPES: ENABLE |
2025 | PROTOTYPES: ENABLE |
1040 | |
2026 | |
1041 | void to_json (SV *scalar) |
2027 | void encode_json (SV *scalar) |
|
|
2028 | ALIAS: |
|
|
2029 | to_json_ = 0 |
|
|
2030 | encode_json = F_UTF8 |
1042 | PPCODE: |
2031 | PPCODE: |
|
|
2032 | { |
|
|
2033 | JSON json; |
|
|
2034 | json_init (&json); |
|
|
2035 | json.flags |= ix; |
1043 | XPUSHs (encode_json (scalar, F_UTF8)); |
2036 | XPUSHs (encode_json (scalar, &json)); |
|
|
2037 | } |
1044 | |
2038 | |
1045 | void from_json (SV *jsonstr) |
2039 | void decode_json (SV *jsonstr) |
|
|
2040 | ALIAS: |
|
|
2041 | from_json_ = 0 |
|
|
2042 | decode_json = F_UTF8 |
1046 | PPCODE: |
2043 | PPCODE: |
|
|
2044 | { |
|
|
2045 | JSON json; |
|
|
2046 | json_init (&json); |
|
|
2047 | json.flags |= ix; |
1047 | XPUSHs (decode_json (jsonstr, F_UTF8)); |
2048 | XPUSHs (decode_json (jsonstr, &json, 0)); |
|
|
2049 | } |
1048 | |
2050 | |