… | |
… | |
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" |
8 | |
9 | |
|
|
10 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
|
|
11 | # define snprintf _snprintf // C compilers have this in stdio.h |
|
|
12 | #endif |
|
|
13 | |
9 | #define F_ASCII 0x00000001 |
14 | #define F_ASCII 0x00000001UL |
10 | #define F_UTF8 0x00000002 |
15 | #define F_UTF8 0x00000002UL |
11 | #define F_INDENT 0x00000004 |
16 | #define F_INDENT 0x00000004UL |
12 | #define F_CANONICAL 0x00000008 |
17 | #define F_CANONICAL 0x00000008UL |
13 | #define F_SPACE_BEFORE 0x00000010 |
18 | #define F_SPACE_BEFORE 0x00000010UL |
14 | #define F_SPACE_AFTER 0x00000020 |
19 | #define F_SPACE_AFTER 0x00000020UL |
|
|
20 | #define F_ALLOW_NONREF 0x00000080UL |
15 | #define F_JSON_RPC 0x00000040 |
21 | #define F_SHRINK 0x00000100UL |
|
|
22 | #define F_MAXDEPTH 0xf8000000UL |
|
|
23 | #define S_MAXDEPTH 27 |
|
|
24 | |
|
|
25 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
|
|
26 | |
|
|
27 | // F_SELFCONVERT? <=> to_json/toJson |
|
|
28 | // F_BLESSED? <=> { $__class__$ => } |
16 | |
29 | |
17 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
30 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
18 | #define F_DEFAULT 0 |
31 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
19 | |
32 | |
20 | #define INIT_SIZE 32 // initial scalar size to be allocated |
33 | #define INIT_SIZE 32 // initial scalar size to be allocated |
|
|
34 | #define INDENT_STEP 3 // spaces per indentation level |
|
|
35 | |
|
|
36 | #define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character |
|
|
37 | #define SHORT_STRING_LEN 512 // special-case strings of up to this size |
21 | |
38 | |
22 | #define SB do { |
39 | #define SB do { |
23 | #define SE } while (0) |
40 | #define SE } while (0) |
24 | |
41 | |
25 | static HV *json_stash; |
42 | static HV *json_stash; // JSON::XS:: |
|
|
43 | |
|
|
44 | ///////////////////////////////////////////////////////////////////////////// |
|
|
45 | // utility functions |
|
|
46 | |
|
|
47 | static UV * |
|
|
48 | SvJSON (SV *sv) |
|
|
49 | { |
|
|
50 | if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) |
|
|
51 | croak ("object is not of type JSON::XS"); |
|
|
52 | |
|
|
53 | return &SvUVX (SvRV (sv)); |
|
|
54 | } |
|
|
55 | |
|
|
56 | static void |
|
|
57 | shrink (SV *sv) |
|
|
58 | { |
|
|
59 | sv_utf8_downgrade (sv, 1); |
|
|
60 | if (SvLEN (sv) > SvCUR (sv) + 1) |
|
|
61 | { |
|
|
62 | #ifdef SvPV_shrink_to_cur |
|
|
63 | SvPV_shrink_to_cur (sv); |
|
|
64 | #elif defined (SvPV_renew) |
|
|
65 | SvPV_renew (sv, SvCUR (sv) + 1); |
|
|
66 | #endif |
|
|
67 | } |
|
|
68 | } |
|
|
69 | |
|
|
70 | // decode an utf-8 character and return it, or (UV)-1 in |
|
|
71 | // case of an error. |
|
|
72 | // we special-case "safe" characters from U+80 .. U+7FF, |
|
|
73 | // but use the very good perl function to parse anything else. |
|
|
74 | // note that we never call this function for a ascii codepoints |
|
|
75 | static UV |
|
|
76 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
|
|
77 | { |
|
|
78 | if (s[0] > 0xdf || s[0] < 0xc2) |
|
|
79 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
|
|
80 | else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
|
|
81 | { |
|
|
82 | *clen = 2; |
|
|
83 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
|
|
84 | } |
|
|
85 | else |
|
|
86 | { |
|
|
87 | *clen = (STRLEN)-1; |
|
|
88 | return (UV)-1; |
|
|
89 | } |
|
|
90 | } |
|
|
91 | |
|
|
92 | ///////////////////////////////////////////////////////////////////////////// |
|
|
93 | // encoder |
26 | |
94 | |
27 | // structure used for encoding JSON |
95 | // structure used for encoding JSON |
28 | typedef struct |
96 | typedef struct |
29 | { |
97 | { |
30 | char *cur; |
98 | char *cur; // SvPVX (sv) + current output position |
31 | STRLEN len; // SvLEN (sv) |
|
|
32 | char *end; // SvEND (sv) |
99 | char *end; // SvEND (sv) |
33 | SV *sv; |
100 | SV *sv; // result scalar |
34 | UV flags; |
101 | U32 flags; // F_* |
35 | int max_recurse; |
102 | U32 indent; // indentation level |
36 | int indent; |
103 | U32 maxdepth; // max. indentation/recursion level |
37 | } enc_t; |
104 | } enc_t; |
38 | |
|
|
39 | // structure used for decoding JSON |
|
|
40 | typedef struct |
|
|
41 | { |
|
|
42 | char *cur; |
|
|
43 | char *end; |
|
|
44 | char *err; |
|
|
45 | UV flags; |
|
|
46 | } dec_t; |
|
|
47 | |
|
|
48 | static UV * |
|
|
49 | SvJSON (SV *sv) |
|
|
50 | { |
|
|
51 | if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) |
|
|
52 | croak ("object is not of type JSON::XS"); |
|
|
53 | |
|
|
54 | return &SvUVX (SvRV (sv)); |
|
|
55 | } |
|
|
56 | |
|
|
57 | ///////////////////////////////////////////////////////////////////////////// |
|
|
58 | |
105 | |
59 | static void |
106 | static void |
60 | need (enc_t *enc, STRLEN len) |
107 | need (enc_t *enc, STRLEN len) |
61 | { |
108 | { |
62 | if (enc->cur + len >= enc->end) |
109 | if (enc->cur + len >= enc->end) |
63 | { |
110 | { |
64 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
111 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
65 | SvGROW (enc->sv, cur + len + 1); |
112 | SvGROW (enc->sv, cur + len + 1); |
66 | enc->cur = SvPVX (enc->sv) + cur; |
113 | enc->cur = SvPVX (enc->sv) + cur; |
67 | enc->end = SvEND (enc->sv); |
114 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
68 | } |
115 | } |
69 | } |
116 | } |
70 | |
117 | |
71 | static void |
118 | static void |
72 | encode_ch (enc_t *enc, char ch) |
119 | encode_ch (enc_t *enc, char ch) |
… | |
… | |
78 | static void |
125 | static void |
79 | encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) |
126 | encode_str (enc_t *enc, char *str, STRLEN len, int is_utf8) |
80 | { |
127 | { |
81 | char *end = str + len; |
128 | char *end = str + len; |
82 | |
129 | |
|
|
130 | need (enc, len); |
|
|
131 | |
83 | while (str < end) |
132 | while (str < end) |
84 | { |
133 | { |
85 | unsigned char ch = *(unsigned char *)str; |
134 | unsigned char ch = *(unsigned char *)str; |
|
|
135 | |
86 | if (ch >= 0x20 && ch < 0x80) // most common case |
136 | if (ch >= 0x20 && ch < 0x80) // most common case |
87 | { |
137 | { |
|
|
138 | if (ch == '"') // but with slow exceptions |
|
|
139 | { |
|
|
140 | need (enc, len += 1); |
|
|
141 | *enc->cur++ = '\\'; |
|
|
142 | *enc->cur++ = '"'; |
|
|
143 | } |
|
|
144 | else if (ch == '\\') |
|
|
145 | { |
|
|
146 | need (enc, len += 1); |
|
|
147 | *enc->cur++ = '\\'; |
|
|
148 | *enc->cur++ = '\\'; |
|
|
149 | } |
|
|
150 | else |
88 | *enc->cur++ = ch; |
151 | *enc->cur++ = ch; |
|
|
152 | |
89 | str++; |
153 | ++str; |
90 | } |
154 | } |
91 | else |
155 | else |
92 | { |
156 | { |
93 | STRLEN clen; |
157 | switch (ch) |
94 | UV uch; |
|
|
95 | |
|
|
96 | if (is_utf8) |
|
|
97 | { |
158 | { |
98 | uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); |
159 | case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; |
99 | if (clen < 0) |
160 | case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; |
100 | croak ("malformed UTF-8 character in string, cannot convert to JSON"); |
161 | case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; |
101 | } |
162 | case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; |
102 | else |
163 | case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; |
103 | { |
|
|
104 | uch = ch; |
|
|
105 | clen = 1; |
|
|
106 | } |
|
|
107 | |
164 | |
108 | need (enc, len += 6); |
165 | default: |
109 | |
|
|
110 | if (uch < 0xa0 || enc->flags & F_ASCII) |
|
|
111 | { |
|
|
112 | if (uch > 0xFFFFUL) |
|
|
113 | { |
166 | { |
|
|
167 | STRLEN clen; |
|
|
168 | UV uch; |
|
|
169 | |
|
|
170 | if (is_utf8) |
|
|
171 | { |
|
|
172 | //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); |
|
|
173 | uch = decode_utf8 (str, end - str, &clen); |
|
|
174 | if (clen == (STRLEN)-1) |
|
|
175 | croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
|
|
176 | } |
|
|
177 | else |
|
|
178 | { |
|
|
179 | uch = ch; |
114 | len += 6; |
180 | clen = 1; |
|
|
181 | } |
|
|
182 | |
|
|
183 | if (uch > 0x10FFFFUL) |
|
|
184 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
|
|
185 | |
|
|
186 | if (uch < 0x80 || enc->flags & F_ASCII) |
|
|
187 | { |
|
|
188 | if (uch > 0xFFFFUL) |
|
|
189 | { |
115 | need (enc, len += 6); |
190 | need (enc, len += 11); |
116 | sprintf (enc->cur, "\\u%04x\\u%04x", |
191 | sprintf (enc->cur, "\\u%04x\\u%04x", |
117 | (uch - 0x10000) / 0x400 + 0xD800, |
192 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
118 | (uch - 0x10000) % 0x400 + 0xDC00); |
193 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
119 | enc->cur += 12; |
194 | enc->cur += 12; |
|
|
195 | } |
|
|
196 | else |
|
|
197 | { |
|
|
198 | static char hexdigit [16] = "0123456789abcdef"; |
|
|
199 | need (enc, len += 5); |
|
|
200 | *enc->cur++ = '\\'; |
|
|
201 | *enc->cur++ = 'u'; |
|
|
202 | *enc->cur++ = hexdigit [ uch >> 12 ]; |
|
|
203 | *enc->cur++ = hexdigit [(uch >> 8) & 15]; |
|
|
204 | *enc->cur++ = hexdigit [(uch >> 4) & 15]; |
|
|
205 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
|
|
206 | } |
|
|
207 | |
|
|
208 | str += clen; |
120 | } |
209 | } |
|
|
210 | else if (is_utf8) |
|
|
211 | { |
|
|
212 | need (enc, len += clen); |
|
|
213 | do |
|
|
214 | { |
|
|
215 | *enc->cur++ = *str++; |
|
|
216 | } |
|
|
217 | while (--clen); |
|
|
218 | } |
121 | else |
219 | else |
122 | { |
220 | { |
123 | sprintf (enc->cur, "\\u%04x", uch); |
221 | need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed |
124 | enc->cur += 6; |
222 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
|
|
223 | ++str; |
|
|
224 | } |
125 | } |
225 | } |
126 | } |
226 | } |
127 | else if (is_utf8) |
|
|
128 | { |
|
|
129 | memcpy (enc->cur, str, clen); |
|
|
130 | enc->cur += clen; |
|
|
131 | } |
|
|
132 | else |
|
|
133 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
|
|
134 | |
|
|
135 | str += clen; |
|
|
136 | } |
227 | } |
137 | |
228 | |
138 | --len; |
229 | --len; |
139 | } |
230 | } |
140 | } |
231 | } |
141 | |
232 | |
142 | #define INDENT SB \ |
233 | static void |
|
|
234 | encode_indent (enc_t *enc) |
|
|
235 | { |
143 | if (enc->flags & F_INDENT) \ |
236 | if (enc->flags & F_INDENT) |
144 | { \ |
237 | { |
145 | int i_; \ |
238 | int spaces = enc->indent * INDENT_STEP; |
146 | need (enc, enc->indent); \ |
|
|
147 | for (i_ = enc->indent * 3; i_--; )\ |
|
|
148 | encode_ch (enc, ' '); \ |
|
|
149 | } \ |
|
|
150 | SE |
|
|
151 | |
239 | |
152 | #define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE |
240 | need (enc, spaces); |
153 | #define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE |
241 | memset (enc->cur, ' ', spaces); |
154 | #define COMMA SB \ |
242 | enc->cur += spaces; |
|
|
243 | } |
|
|
244 | } |
|
|
245 | |
|
|
246 | static void |
|
|
247 | encode_space (enc_t *enc) |
|
|
248 | { |
|
|
249 | need (enc, 1); |
155 | encode_ch (enc, ','); \ |
250 | encode_ch (enc, ' '); |
|
|
251 | } |
|
|
252 | |
|
|
253 | static void |
|
|
254 | encode_nl (enc_t *enc) |
|
|
255 | { |
156 | if (enc->flags & F_INDENT) \ |
256 | if (enc->flags & F_INDENT) |
157 | NL; \ |
257 | { |
|
|
258 | need (enc, 1); |
|
|
259 | encode_ch (enc, '\n'); |
|
|
260 | } |
|
|
261 | } |
|
|
262 | |
|
|
263 | static void |
|
|
264 | encode_comma (enc_t *enc) |
|
|
265 | { |
|
|
266 | encode_ch (enc, ','); |
|
|
267 | |
|
|
268 | if (enc->flags & F_INDENT) |
|
|
269 | encode_nl (enc); |
158 | else if (enc->flags & F_SPACE_AFTER) \ |
270 | else if (enc->flags & F_SPACE_AFTER) |
159 | SPACE; \ |
271 | encode_space (enc); |
160 | SE |
272 | } |
161 | |
273 | |
162 | static void encode_sv (enc_t *enc, SV *sv); |
274 | static void encode_sv (enc_t *enc, SV *sv); |
163 | |
275 | |
164 | static void |
276 | static void |
165 | encode_av (enc_t *enc, AV *av) |
277 | encode_av (enc_t *enc, AV *av) |
166 | { |
278 | { |
167 | int i, len = av_len (av); |
279 | int i, len = av_len (av); |
168 | |
280 | |
169 | encode_ch (enc, '['); NL; |
281 | if (enc->indent >= enc->maxdepth) |
|
|
282 | croak ("data structure too deep (hit recursion limit)"); |
|
|
283 | |
|
|
284 | encode_ch (enc, '['); encode_nl (enc); |
170 | ++enc->indent; |
285 | ++enc->indent; |
171 | |
286 | |
172 | for (i = 0; i <= len; ++i) |
287 | for (i = 0; i <= len; ++i) |
173 | { |
288 | { |
174 | INDENT; |
289 | encode_indent (enc); |
175 | encode_sv (enc, *av_fetch (av, i, 0)); |
290 | encode_sv (enc, *av_fetch (av, i, 0)); |
176 | |
291 | |
177 | if (i < len) |
292 | if (i < len) |
178 | COMMA; |
293 | encode_comma (enc); |
179 | } |
294 | } |
180 | |
295 | |
181 | NL; |
296 | encode_nl (enc); |
182 | |
297 | |
183 | --enc->indent; |
298 | --enc->indent; |
184 | INDENT; encode_ch (enc, ']'); |
299 | encode_indent (enc); encode_ch (enc, ']'); |
185 | } |
300 | } |
186 | |
301 | |
187 | static void |
302 | static void |
188 | encode_he (enc_t *enc, HE *he) |
303 | encode_he (enc_t *enc, HE *he) |
189 | { |
304 | { |
… | |
… | |
191 | |
306 | |
192 | if (HeKLEN (he) == HEf_SVKEY) |
307 | if (HeKLEN (he) == HEf_SVKEY) |
193 | { |
308 | { |
194 | SV *sv = HeSVKEY (he); |
309 | SV *sv = HeSVKEY (he); |
195 | STRLEN len; |
310 | STRLEN len; |
|
|
311 | char *str; |
|
|
312 | |
|
|
313 | SvGETMAGIC (sv); |
196 | char *str = SvPV (sv, len); |
314 | str = SvPV (sv, len); |
197 | |
315 | |
198 | encode_str (enc, str, len, SvUTF8 (sv)); |
316 | encode_str (enc, str, len, SvUTF8 (sv)); |
199 | } |
317 | } |
200 | else |
318 | else |
201 | encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
319 | encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
202 | |
320 | |
203 | encode_ch (enc, '"'); |
321 | encode_ch (enc, '"'); |
204 | |
322 | |
205 | if (enc->flags & F_SPACE_BEFORE) SPACE; |
323 | if (enc->flags & F_SPACE_BEFORE) encode_space (enc); |
206 | encode_ch (enc, ':'); |
324 | encode_ch (enc, ':'); |
207 | if (enc->flags & F_SPACE_AFTER ) SPACE; |
325 | if (enc->flags & F_SPACE_AFTER ) encode_space (enc); |
208 | encode_sv (enc, HeVAL (he)); |
326 | encode_sv (enc, HeVAL (he)); |
209 | } |
327 | } |
210 | |
328 | |
211 | // compare hash entries, used when all keys are bytestrings |
329 | // compare hash entries, used when all keys are bytestrings |
212 | static int |
330 | static int |
… | |
… | |
218 | HE *b = *(HE **)b_; |
336 | HE *b = *(HE **)b_; |
219 | |
337 | |
220 | STRLEN la = HeKLEN (a); |
338 | STRLEN la = HeKLEN (a); |
221 | STRLEN lb = HeKLEN (b); |
339 | STRLEN lb = HeKLEN (b); |
222 | |
340 | |
223 | if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) |
341 | if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) |
224 | cmp = la < lb ? -1 : la == lb ? 0 : 1; |
342 | cmp = la - lb; |
225 | |
343 | |
226 | return cmp; |
344 | return cmp; |
227 | } |
345 | } |
228 | |
346 | |
229 | // compare hash entries, used when some keys are sv's or utf-x |
347 | // compare hash entries, used when some keys are sv's or utf-x |
… | |
… | |
236 | static void |
354 | static void |
237 | encode_hv (enc_t *enc, HV *hv) |
355 | encode_hv (enc_t *enc, HV *hv) |
238 | { |
356 | { |
239 | int count, i; |
357 | int count, i; |
240 | |
358 | |
|
|
359 | if (enc->indent >= enc->maxdepth) |
|
|
360 | croak ("data structure too deep (hit recursion limit)"); |
|
|
361 | |
241 | encode_ch (enc, '{'); NL; ++enc->indent; |
362 | encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; |
242 | |
363 | |
243 | if ((count = hv_iterinit (hv))) |
364 | if ((count = hv_iterinit (hv))) |
244 | { |
365 | { |
245 | // for canonical output we have to sort by keys first |
366 | // for canonical output we have to sort by keys first |
246 | // actually, this is mostly due to the stupid so-called |
367 | // actually, this is mostly due to the stupid so-called |
247 | // security workaround added somewhere in 5.8.x. |
368 | // security workaround added somewhere in 5.8.x. |
248 | // that randomises hash orderings |
369 | // that randomises hash orderings |
249 | if (enc->flags & F_CANONICAL) |
370 | if (enc->flags & F_CANONICAL) |
250 | { |
371 | { |
251 | HE *he, *hes [count]; |
372 | HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode |
252 | int fast = 1; |
373 | int fast = 1; |
253 | |
374 | |
254 | i = 0; |
375 | i = 0; |
255 | while ((he = hv_iternext (hv))) |
376 | while ((he = hv_iternext (hv))) |
256 | { |
377 | { |
… | |
… | |
263 | |
384 | |
264 | if (fast) |
385 | if (fast) |
265 | qsort (hes, count, sizeof (HE *), he_cmp_fast); |
386 | qsort (hes, count, sizeof (HE *), he_cmp_fast); |
266 | else |
387 | else |
267 | { |
388 | { |
268 | // hack to disable "use bytes" |
389 | // hack to forcefully disable "use bytes" |
269 | COP *oldcop = PL_curcop, cop; |
390 | COP cop = *PL_curcop; |
270 | cop.op_private = 0; |
391 | cop.op_private = 0; |
|
|
392 | |
|
|
393 | ENTER; |
|
|
394 | SAVETMPS; |
|
|
395 | |
|
|
396 | SAVEVPTR (PL_curcop); |
271 | PL_curcop = &cop; |
397 | PL_curcop = &cop; |
272 | |
398 | |
273 | SAVETMPS; |
|
|
274 | qsort (hes, count, sizeof (HE *), he_cmp_slow); |
399 | qsort (hes, count, sizeof (HE *), he_cmp_slow); |
|
|
400 | |
275 | FREETMPS; |
401 | FREETMPS; |
276 | |
402 | LEAVE; |
277 | PL_curcop = oldcop; |
|
|
278 | } |
403 | } |
279 | |
404 | |
280 | for (i = 0; i < count; ++i) |
405 | for (i = 0; i < count; ++i) |
281 | { |
406 | { |
282 | INDENT; |
407 | encode_indent (enc); |
283 | encode_he (enc, hes [i]); |
408 | encode_he (enc, hes [i]); |
284 | |
409 | |
285 | if (i < count - 1) |
410 | if (i < count - 1) |
286 | COMMA; |
411 | encode_comma (enc); |
287 | } |
412 | } |
288 | |
413 | |
289 | NL; |
414 | encode_nl (enc); |
290 | } |
415 | } |
291 | else |
416 | else |
292 | { |
417 | { |
293 | SV *sv; |
|
|
294 | HE *he = hv_iternext (hv); |
418 | HE *he = hv_iternext (hv); |
295 | |
419 | |
296 | for (;;) |
420 | for (;;) |
297 | { |
421 | { |
298 | INDENT; |
422 | encode_indent (enc); |
299 | encode_he (enc, he); |
423 | encode_he (enc, he); |
300 | |
424 | |
301 | if (!(he = hv_iternext (hv))) |
425 | if (!(he = hv_iternext (hv))) |
302 | break; |
426 | break; |
303 | |
427 | |
304 | COMMA; |
428 | encode_comma (enc); |
305 | } |
429 | } |
306 | |
430 | |
307 | NL; |
431 | encode_nl (enc); |
308 | } |
432 | } |
309 | } |
433 | } |
310 | |
434 | |
311 | --enc->indent; INDENT; encode_ch (enc, '}'); |
435 | --enc->indent; encode_indent (enc); encode_ch (enc, '}'); |
|
|
436 | } |
|
|
437 | |
|
|
438 | // encode objects, arrays and special \0=false and \1=true values. |
|
|
439 | static void |
|
|
440 | encode_rv (enc_t *enc, SV *sv) |
|
|
441 | { |
|
|
442 | svtype svt; |
|
|
443 | |
|
|
444 | SvGETMAGIC (sv); |
|
|
445 | svt = SvTYPE (sv); |
|
|
446 | |
|
|
447 | if (svt == SVt_PVHV) |
|
|
448 | encode_hv (enc, (HV *)sv); |
|
|
449 | else if (svt == SVt_PVAV) |
|
|
450 | encode_av (enc, (AV *)sv); |
|
|
451 | else if (svt < SVt_PVAV) |
|
|
452 | { |
|
|
453 | if (SvNIOK (sv) && SvIV (sv) == 0) |
|
|
454 | encode_str (enc, "false", 5, 0); |
|
|
455 | else if (SvNIOK (sv) && SvIV (sv) == 1) |
|
|
456 | encode_str (enc, "true", 4, 0); |
|
|
457 | else |
|
|
458 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
|
|
459 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
460 | } |
|
|
461 | else |
|
|
462 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
|
|
463 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
312 | } |
464 | } |
313 | |
465 | |
314 | static void |
466 | static void |
315 | encode_sv (enc_t *enc, SV *sv) |
467 | encode_sv (enc_t *enc, SV *sv) |
316 | { |
468 | { |
|
|
469 | SvGETMAGIC (sv); |
|
|
470 | |
317 | if (SvPOKp (sv)) |
471 | if (SvPOKp (sv)) |
318 | { |
472 | { |
319 | STRLEN len; |
473 | STRLEN len; |
320 | char *str = SvPV (sv, len); |
474 | char *str = SvPV (sv, len); |
321 | encode_ch (enc, '"'); |
475 | encode_ch (enc, '"'); |
… | |
… | |
335 | SvIsUV(sv) |
489 | SvIsUV(sv) |
336 | ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) |
490 | ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) |
337 | : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); |
491 | : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); |
338 | } |
492 | } |
339 | else if (SvROK (sv)) |
493 | else if (SvROK (sv)) |
340 | { |
494 | encode_rv (enc, SvRV (sv)); |
341 | if (!--enc->max_recurse) |
|
|
342 | croak ("data structure too deep (hit recursion limit)"); |
|
|
343 | |
|
|
344 | sv = SvRV (sv); |
|
|
345 | |
|
|
346 | switch (SvTYPE (sv)) |
|
|
347 | { |
|
|
348 | case SVt_PVAV: encode_av (enc, (AV *)sv); break; |
|
|
349 | case SVt_PVHV: encode_hv (enc, (HV *)sv); break; |
|
|
350 | |
|
|
351 | default: |
|
|
352 | croak ("JSON can only represent references to arrays or hashes"); |
|
|
353 | } |
|
|
354 | } |
|
|
355 | else if (!SvOK (sv)) |
495 | else if (!SvOK (sv)) |
356 | encode_str (enc, "null", 4, 0); |
496 | encode_str (enc, "null", 4, 0); |
357 | else |
497 | else |
358 | croak ("encountered perl type that JSON cannot handle"); |
498 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
|
|
499 | SvPV_nolen (sv), SvFLAGS (sv)); |
359 | } |
500 | } |
360 | |
501 | |
361 | static SV * |
502 | static SV * |
362 | encode_json (SV *scalar, UV flags) |
503 | encode_json (SV *scalar, U32 flags) |
363 | { |
504 | { |
364 | enc_t enc; |
505 | enc_t enc; |
|
|
506 | |
|
|
507 | if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
|
|
508 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
|
|
509 | |
365 | enc.flags = flags; |
510 | enc.flags = flags; |
366 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
511 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
367 | enc.cur = SvPVX (enc.sv); |
512 | enc.cur = SvPVX (enc.sv); |
368 | enc.end = SvEND (enc.sv); |
513 | enc.end = SvEND (enc.sv); |
369 | enc.max_recurse = 0; |
|
|
370 | enc.indent = 0; |
514 | enc.indent = 0; |
|
|
515 | enc.maxdepth = DEC_DEPTH (flags); |
371 | |
516 | |
372 | SvPOK_only (enc.sv); |
517 | SvPOK_only (enc.sv); |
373 | encode_sv (&enc, scalar); |
518 | encode_sv (&enc, scalar); |
374 | |
519 | |
375 | if (!(flags & (F_ASCII | F_UTF8))) |
520 | if (!(flags & (F_ASCII | F_UTF8))) |
376 | SvUTF8_on (enc.sv); |
521 | SvUTF8_on (enc.sv); |
377 | |
522 | |
378 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
523 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
|
|
524 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
|
|
525 | |
|
|
526 | if (enc.flags & F_SHRINK) |
|
|
527 | shrink (enc.sv); |
|
|
528 | |
379 | return enc.sv; |
529 | return enc.sv; |
380 | } |
530 | } |
381 | |
531 | |
382 | ///////////////////////////////////////////////////////////////////////////// |
532 | ///////////////////////////////////////////////////////////////////////////// |
|
|
533 | // decoder |
383 | |
534 | |
384 | #define WS \ |
535 | // structure used for decoding JSON |
|
|
536 | typedef struct |
|
|
537 | { |
|
|
538 | char *cur; // current parser pointer |
|
|
539 | char *end; // end of input string |
|
|
540 | const char *err; // parse error, if != 0 |
|
|
541 | U32 flags; // F_* |
|
|
542 | U32 depth; // recursion depth |
|
|
543 | U32 maxdepth; // recursion depth limit |
|
|
544 | } dec_t; |
|
|
545 | |
|
|
546 | static void |
|
|
547 | decode_ws (dec_t *dec) |
|
|
548 | { |
385 | for (;;) \ |
549 | for (;;) |
386 | { \ |
550 | { |
387 | char ch = *dec->cur; \ |
551 | char ch = *dec->cur; |
|
|
552 | |
388 | if (ch > 0x20 \ |
553 | if (ch > 0x20 |
389 | || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ |
554 | || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) |
390 | break; \ |
555 | break; |
|
|
556 | |
391 | ++dec->cur; \ |
557 | ++dec->cur; |
392 | } |
558 | } |
|
|
559 | } |
393 | |
560 | |
394 | #define ERR(reason) SB dec->err = reason; goto fail; SE |
561 | #define ERR(reason) SB dec->err = reason; goto fail; SE |
|
|
562 | |
395 | #define EXPECT_CH(ch) SB \ |
563 | #define EXPECT_CH(ch) SB \ |
396 | if (*dec->cur != ch) \ |
564 | if (*dec->cur != ch) \ |
397 | ERR (# ch " expected"); \ |
565 | ERR (# ch " expected"); \ |
398 | ++dec->cur; \ |
566 | ++dec->cur; \ |
399 | SE |
567 | SE |
400 | |
568 | |
|
|
569 | #define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") |
|
|
570 | #define DEC_DEC_DEPTH --dec->depth |
|
|
571 | |
401 | static SV *decode_sv (dec_t *dec); |
572 | static SV *decode_sv (dec_t *dec); |
402 | |
|
|
403 | #define APPEND_CH(ch) SB \ |
|
|
404 | SvGROW (sv, cur + 1 + 1); \ |
|
|
405 | SvPVX (sv)[cur++] = (ch); \ |
|
|
406 | SE |
|
|
407 | |
573 | |
408 | static signed char decode_hexdigit[256]; |
574 | static signed char decode_hexdigit[256]; |
409 | |
575 | |
410 | static UV |
576 | static UV |
411 | decode_4hex (dec_t *dec) |
577 | decode_4hex (dec_t *dec) |
412 | { |
578 | { |
413 | signed char d1, d2, d3, d4; |
579 | signed char d1, d2, d3, d4; |
|
|
580 | unsigned char *cur = (unsigned char *)dec->cur; |
414 | |
581 | |
415 | d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; |
|
|
416 | if (d1 < 0) ERR ("four hexadecimal digits expected"); |
582 | d1 = decode_hexdigit [cur [0]]; if (d1 < 0) ERR ("four hexadecimal digits expected"); |
417 | d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; |
|
|
418 | if (d2 < 0) ERR ("four hexadecimal digits expected"); |
583 | d2 = decode_hexdigit [cur [1]]; if (d2 < 0) ERR ("four hexadecimal digits expected"); |
419 | d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]]; |
|
|
420 | if (d3 < 0) ERR ("four hexadecimal digits expected"); |
584 | d3 = decode_hexdigit [cur [2]]; if (d3 < 0) ERR ("four hexadecimal digits expected"); |
421 | d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]]; |
|
|
422 | if (d4 < 0) ERR ("four hexadecimal digits expected"); |
585 | d4 = decode_hexdigit [cur [3]]; if (d4 < 0) ERR ("four hexadecimal digits expected"); |
423 | |
586 | |
424 | dec->cur += 4; |
587 | dec->cur += 4; |
425 | |
588 | |
426 | return ((UV)d1) << 12 |
589 | return ((UV)d1) << 12 |
427 | | ((UV)d2) << 8 |
590 | | ((UV)d2) << 8 |
… | |
… | |
433 | } |
596 | } |
434 | |
597 | |
435 | static SV * |
598 | static SV * |
436 | decode_str (dec_t *dec) |
599 | decode_str (dec_t *dec) |
437 | { |
600 | { |
438 | SV *sv = NEWSV (0,2); |
601 | SV *sv = 0; |
439 | STRLEN cur = 0; |
|
|
440 | int utf8 = 0; |
602 | int utf8 = 0; |
441 | |
603 | |
442 | for (;;) |
604 | do |
443 | { |
605 | { |
444 | unsigned char ch = *(unsigned char *)dec->cur; |
606 | char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; |
|
|
607 | char *cur = buf; |
445 | |
608 | |
446 | if (ch == '"') |
609 | do |
447 | break; |
|
|
448 | else if (ch == '\\') |
|
|
449 | { |
610 | { |
450 | switch (*++dec->cur) |
611 | unsigned char ch = *(unsigned char *)dec->cur++; |
|
|
612 | |
|
|
613 | if (ch == '"') |
451 | { |
614 | { |
452 | case '\\': |
615 | --dec->cur; |
453 | case '/': |
616 | break; |
454 | case '"': APPEND_CH (*dec->cur++); break; |
617 | } |
455 | |
618 | else if (ch == '\\') |
456 | case 'b': APPEND_CH ('\010'); ++dec->cur; break; |
619 | { |
457 | case 't': APPEND_CH ('\011'); ++dec->cur; break; |
620 | switch (*dec->cur) |
458 | case 'n': APPEND_CH ('\012'); ++dec->cur; break; |
|
|
459 | case 'f': APPEND_CH ('\014'); ++dec->cur; break; |
|
|
460 | case 'r': APPEND_CH ('\015'); ++dec->cur; break; |
|
|
461 | |
|
|
462 | case 'u': |
|
|
463 | { |
621 | { |
464 | UV lo, hi; |
622 | case '\\': |
465 | ++dec->cur; |
623 | case '/': |
|
|
624 | case '"': *cur++ = *dec->cur++; break; |
466 | |
625 | |
467 | hi = decode_4hex (dec); |
626 | case 'b': ++dec->cur; *cur++ = '\010'; break; |
468 | if (hi == (UV)-1) |
627 | case 't': ++dec->cur; *cur++ = '\011'; break; |
469 | goto fail; |
628 | case 'n': ++dec->cur; *cur++ = '\012'; break; |
|
|
629 | case 'f': ++dec->cur; *cur++ = '\014'; break; |
|
|
630 | case 'r': ++dec->cur; *cur++ = '\015'; break; |
470 | |
631 | |
471 | // possibly a surrogate pair |
632 | case 'u': |
472 | if (hi >= 0xd800 && hi < 0xdc00) |
|
|
473 | { |
633 | { |
474 | if (dec->cur [0] != '\\' || dec->cur [1] != 'u') |
634 | UV lo, hi; |
475 | ERR ("illegal surrogate character"); |
|
|
476 | |
|
|
477 | dec->cur += 2; |
635 | ++dec->cur; |
478 | |
636 | |
479 | lo = decode_4hex (dec); |
637 | hi = decode_4hex (dec); |
480 | if (lo == (UV)-1) |
638 | if (hi == (UV)-1) |
481 | goto fail; |
639 | goto fail; |
482 | |
640 | |
|
|
641 | // possibly a surrogate pair |
|
|
642 | if (hi >= 0xd800) |
|
|
643 | if (hi < 0xdc00) |
|
|
644 | { |
|
|
645 | if (dec->cur [0] != '\\' || dec->cur [1] != 'u') |
|
|
646 | ERR ("missing low surrogate character in surrogate pair"); |
|
|
647 | |
|
|
648 | dec->cur += 2; |
|
|
649 | |
|
|
650 | lo = decode_4hex (dec); |
|
|
651 | if (lo == (UV)-1) |
|
|
652 | goto fail; |
|
|
653 | |
483 | if (lo < 0xdc00 || lo >= 0xe000) |
654 | if (lo < 0xdc00 || lo >= 0xe000) |
484 | ERR ("surrogate pair expected"); |
655 | ERR ("surrogate pair expected"); |
485 | |
656 | |
486 | hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; |
657 | hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; |
|
|
658 | } |
|
|
659 | else if (hi < 0xe000) |
|
|
660 | ERR ("missing high surrogate character in surrogate pair"); |
|
|
661 | |
|
|
662 | if (hi >= 0x80) |
|
|
663 | { |
|
|
664 | utf8 = 1; |
|
|
665 | |
|
|
666 | cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); |
|
|
667 | } |
|
|
668 | else |
|
|
669 | *cur++ = hi; |
487 | } |
670 | } |
488 | else if (lo >= 0xdc00 && lo < 0xe000) |
|
|
489 | ERR ("illegal surrogate character"); |
|
|
490 | |
|
|
491 | if (hi >= 0x80) |
|
|
492 | { |
671 | break; |
493 | utf8 = 1; |
|
|
494 | |
672 | |
495 | SvGROW (sv, cur + 4 + 1); // at most 4 bytes for 21 bits |
|
|
496 | cur = (char *)uvuni_to_utf8_flags (SvPVX (sv) + cur, hi, 0) - SvPVX (sv); |
|
|
497 | } |
|
|
498 | else |
673 | default: |
499 | APPEND_CH (hi); |
674 | --dec->cur; |
|
|
675 | ERR ("illegal backslash escape sequence in string"); |
500 | } |
676 | } |
|
|
677 | } |
|
|
678 | else if (ch >= 0x20 && ch <= 0x7f) |
|
|
679 | *cur++ = ch; |
|
|
680 | else if (ch >= 0x80) |
|
|
681 | { |
|
|
682 | STRLEN clen; |
|
|
683 | UV uch; |
|
|
684 | |
|
|
685 | --dec->cur; |
|
|
686 | |
|
|
687 | uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); |
|
|
688 | if (clen == (STRLEN)-1) |
|
|
689 | ERR ("malformed UTF-8 character in JSON string"); |
|
|
690 | |
|
|
691 | do |
|
|
692 | *cur++ = *dec->cur++; |
|
|
693 | while (--clen); |
|
|
694 | |
|
|
695 | utf8 = 1; |
|
|
696 | } |
|
|
697 | else |
|
|
698 | { |
|
|
699 | --dec->cur; |
|
|
700 | |
|
|
701 | if (!ch) |
|
|
702 | ERR ("unexpected end of string while parsing JSON string"); |
501 | break; |
703 | else |
|
|
704 | ERR ("invalid character encountered while parsing JSON string"); |
502 | } |
705 | } |
503 | } |
706 | } |
504 | else if (ch >= 0x20 && ch <= 0x7f) |
707 | while (cur < buf + SHORT_STRING_LEN); |
505 | APPEND_CH (*dec->cur++); |
708 | |
506 | else if (ch >= 0x80) |
709 | { |
|
|
710 | STRLEN len = cur - buf; |
|
|
711 | |
|
|
712 | if (sv) |
507 | { |
713 | { |
508 | STRLEN clen; |
714 | SvGROW (sv, SvCUR (sv) + len + 1); |
509 | UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); |
715 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
510 | if (clen < 0) |
716 | SvCUR_set (sv, SvCUR (sv) + len); |
511 | ERR ("malformed UTF-8 character in string, cannot convert to JSON"); |
|
|
512 | |
|
|
513 | SvGROW (sv, cur + clen + 1); // at most 4 bytes for 21 bits |
|
|
514 | memcpy (SvPVX (sv) + cur, dec->cur, clen); |
|
|
515 | dec->cur += clen; |
|
|
516 | } |
717 | } |
517 | else |
718 | else |
518 | ERR ("invalid character encountered"); |
719 | sv = newSVpvn (buf, len); |
519 | } |
720 | } |
|
|
721 | } |
|
|
722 | while (*dec->cur != '"'); |
520 | |
723 | |
521 | ++dec->cur; |
724 | ++dec->cur; |
522 | |
725 | |
|
|
726 | if (sv) |
|
|
727 | { |
523 | SvPOK_only (sv); |
728 | SvPOK_only (sv); |
524 | |
|
|
525 | SvCUR_set (sv, cur); |
|
|
526 | *SvEND (sv) = 0; |
729 | *SvEND (sv) = 0; |
527 | |
730 | |
528 | if (utf8) |
731 | if (utf8) |
529 | SvUTF8_on (sv); |
732 | SvUTF8_on (sv); |
|
|
733 | } |
|
|
734 | else |
|
|
735 | sv = newSVpvn ("", 0); |
530 | |
736 | |
531 | return sv; |
737 | return sv; |
532 | |
738 | |
533 | fail: |
739 | fail: |
534 | SvREFCNT_dec (sv); |
|
|
535 | return 0; |
740 | return 0; |
536 | } |
741 | } |
537 | |
742 | |
538 | static SV * |
743 | static SV * |
539 | decode_num (dec_t *dec) |
744 | decode_num (dec_t *dec) |
… | |
… | |
549 | { |
754 | { |
550 | ++dec->cur; |
755 | ++dec->cur; |
551 | if (*dec->cur >= '0' && *dec->cur <= '9') |
756 | if (*dec->cur >= '0' && *dec->cur <= '9') |
552 | ERR ("malformed number (leading zero must not be followed by another digit)"); |
757 | ERR ("malformed number (leading zero must not be followed by another digit)"); |
553 | } |
758 | } |
554 | |
759 | else if (*dec->cur < '0' || *dec->cur > '9') |
555 | // int |
760 | ERR ("malformed number (no digits after initial minus)"); |
|
|
761 | else |
|
|
762 | do |
|
|
763 | { |
|
|
764 | ++dec->cur; |
|
|
765 | } |
556 | while (*dec->cur >= '0' && *dec->cur <= '9') |
766 | while (*dec->cur >= '0' && *dec->cur <= '9'); |
557 | ++dec->cur; |
|
|
558 | |
767 | |
559 | // [frac] |
768 | // [frac] |
560 | if (*dec->cur == '.') |
769 | if (*dec->cur == '.') |
561 | { |
770 | { |
562 | is_nv = 1; |
771 | ++dec->cur; |
|
|
772 | |
|
|
773 | if (*dec->cur < '0' || *dec->cur > '9') |
|
|
774 | ERR ("malformed number (no digits after decimal point)"); |
563 | |
775 | |
564 | do |
776 | do |
565 | { |
777 | { |
566 | ++dec->cur; |
778 | ++dec->cur; |
567 | } |
779 | } |
568 | while (*dec->cur >= '0' && *dec->cur <= '9'); |
780 | while (*dec->cur >= '0' && *dec->cur <= '9'); |
|
|
781 | |
|
|
782 | is_nv = 1; |
569 | } |
783 | } |
570 | |
784 | |
571 | // [exp] |
785 | // [exp] |
572 | if (*dec->cur == 'e' || *dec->cur == 'E') |
786 | if (*dec->cur == 'e' || *dec->cur == 'E') |
573 | { |
787 | { |
574 | is_nv = 1; |
|
|
575 | |
|
|
576 | ++dec->cur; |
788 | ++dec->cur; |
|
|
789 | |
577 | if (*dec->cur == '-' || *dec->cur == '+') |
790 | if (*dec->cur == '-' || *dec->cur == '+') |
578 | ++dec->cur; |
791 | ++dec->cur; |
579 | |
792 | |
|
|
793 | if (*dec->cur < '0' || *dec->cur > '9') |
|
|
794 | ERR ("malformed number (no digits after exp sign)"); |
|
|
795 | |
|
|
796 | do |
|
|
797 | { |
|
|
798 | ++dec->cur; |
|
|
799 | } |
580 | while (*dec->cur >= '0' && *dec->cur <= '9') |
800 | while (*dec->cur >= '0' && *dec->cur <= '9'); |
581 | ++dec->cur; |
801 | |
|
|
802 | is_nv = 1; |
582 | } |
803 | } |
583 | |
804 | |
584 | if (!is_nv) |
805 | if (!is_nv) |
585 | { |
806 | { |
586 | UV uv; |
807 | UV uv; |
… | |
… | |
604 | static SV * |
825 | static SV * |
605 | decode_av (dec_t *dec) |
826 | decode_av (dec_t *dec) |
606 | { |
827 | { |
607 | AV *av = newAV (); |
828 | AV *av = newAV (); |
608 | |
829 | |
|
|
830 | DEC_INC_DEPTH; |
|
|
831 | decode_ws (dec); |
|
|
832 | |
|
|
833 | if (*dec->cur == ']') |
|
|
834 | ++dec->cur; |
|
|
835 | else |
609 | for (;;) |
836 | for (;;) |
610 | { |
837 | { |
611 | SV *value; |
838 | SV *value; |
612 | |
839 | |
613 | value = decode_sv (dec); |
840 | value = decode_sv (dec); |
614 | if (!value) |
841 | if (!value) |
615 | goto fail; |
842 | goto fail; |
616 | |
843 | |
617 | av_push (av, value); |
844 | av_push (av, value); |
618 | |
845 | |
619 | WS; |
846 | decode_ws (dec); |
620 | |
847 | |
621 | if (*dec->cur == ']') |
848 | if (*dec->cur == ']') |
622 | { |
849 | { |
623 | ++dec->cur; |
850 | ++dec->cur; |
624 | break; |
851 | break; |
|
|
852 | } |
625 | } |
853 | |
626 | |
|
|
627 | if (*dec->cur != ',') |
854 | if (*dec->cur != ',') |
628 | ERR (", or ] expected while parsing array"); |
855 | ERR (", or ] expected while parsing array"); |
629 | |
856 | |
630 | ++dec->cur; |
857 | ++dec->cur; |
631 | } |
858 | } |
632 | |
859 | |
|
|
860 | DEC_DEC_DEPTH; |
633 | return newRV_noinc ((SV *)av); |
861 | return newRV_noinc ((SV *)av); |
634 | |
862 | |
635 | fail: |
863 | fail: |
636 | SvREFCNT_dec (av); |
864 | SvREFCNT_dec (av); |
|
|
865 | DEC_DEC_DEPTH; |
637 | return 0; |
866 | return 0; |
638 | } |
867 | } |
639 | |
868 | |
640 | static SV * |
869 | static SV * |
641 | decode_hv (dec_t *dec) |
870 | decode_hv (dec_t *dec) |
642 | { |
871 | { |
643 | HV *hv = newHV (); |
872 | HV *hv = newHV (); |
644 | |
873 | |
|
|
874 | DEC_INC_DEPTH; |
|
|
875 | decode_ws (dec); |
|
|
876 | |
|
|
877 | if (*dec->cur == '}') |
|
|
878 | ++dec->cur; |
|
|
879 | else |
645 | for (;;) |
880 | for (;;) |
646 | { |
881 | { |
647 | SV *key, *value; |
882 | SV *key, *value; |
648 | |
883 | |
649 | WS; EXPECT_CH ('"'); |
884 | decode_ws (dec); EXPECT_CH ('"'); |
650 | |
885 | |
651 | key = decode_str (dec); |
886 | key = decode_str (dec); |
652 | if (!key) |
887 | if (!key) |
653 | goto fail; |
|
|
654 | |
|
|
655 | WS; EXPECT_CH (':'); |
|
|
656 | |
|
|
657 | value = decode_sv (dec); |
|
|
658 | if (!value) |
|
|
659 | { |
|
|
660 | SvREFCNT_dec (key); |
|
|
661 | goto fail; |
888 | goto fail; |
|
|
889 | |
|
|
890 | decode_ws (dec); EXPECT_CH (':'); |
|
|
891 | |
|
|
892 | value = decode_sv (dec); |
|
|
893 | if (!value) |
|
|
894 | { |
|
|
895 | SvREFCNT_dec (key); |
|
|
896 | goto fail; |
662 | } |
897 | } |
663 | |
898 | |
664 | //TODO: optimise |
|
|
665 | hv_store_ent (hv, key, value, 0); |
899 | hv_store_ent (hv, key, value, 0); |
|
|
900 | SvREFCNT_dec (key); |
666 | |
901 | |
667 | WS; |
902 | decode_ws (dec); |
668 | |
903 | |
669 | if (*dec->cur == '}') |
904 | if (*dec->cur == '}') |
670 | { |
905 | { |
671 | ++dec->cur; |
906 | ++dec->cur; |
672 | break; |
907 | break; |
673 | } |
908 | } |
674 | |
909 | |
675 | if (*dec->cur != ',') |
910 | if (*dec->cur != ',') |
676 | ERR (", or } expected while parsing object/hash"); |
911 | ERR (", or } expected while parsing object/hash"); |
677 | |
912 | |
678 | ++dec->cur; |
913 | ++dec->cur; |
679 | } |
914 | } |
680 | |
915 | |
|
|
916 | DEC_DEC_DEPTH; |
681 | return newRV_noinc ((SV *)hv); |
917 | return newRV_noinc ((SV *)hv); |
682 | |
918 | |
683 | fail: |
919 | fail: |
684 | SvREFCNT_dec (hv); |
920 | SvREFCNT_dec (hv); |
|
|
921 | DEC_DEC_DEPTH; |
685 | return 0; |
922 | return 0; |
686 | } |
923 | } |
687 | |
924 | |
688 | static SV * |
925 | static SV * |
689 | decode_sv (dec_t *dec) |
926 | decode_sv (dec_t *dec) |
690 | { |
927 | { |
691 | WS; |
928 | decode_ws (dec); |
692 | switch (*dec->cur) |
929 | switch (*dec->cur) |
693 | { |
930 | { |
694 | case '"': ++dec->cur; return decode_str (dec); |
931 | case '"': ++dec->cur; return decode_str (dec); |
695 | case '[': ++dec->cur; return decode_av (dec); |
932 | case '[': ++dec->cur; return decode_av (dec); |
696 | case '{': ++dec->cur; return decode_hv (dec); |
933 | case '{': ++dec->cur; return decode_hv (dec); |
… | |
… | |
724 | |
961 | |
725 | case 'n': |
962 | case 'n': |
726 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
963 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
727 | { |
964 | { |
728 | dec->cur += 4; |
965 | dec->cur += 4; |
729 | return newSViv (1); |
966 | return newSVsv (&PL_sv_undef); |
730 | } |
967 | } |
731 | else |
968 | else |
732 | ERR ("'null' expected"); |
969 | ERR ("'null' expected"); |
733 | |
970 | |
734 | break; |
971 | break; |
735 | |
972 | |
736 | default: |
973 | default: |
737 | ERR ("malformed json string"); |
974 | ERR ("malformed JSON string, neither array, object, number, string or atom"); |
738 | break; |
975 | break; |
739 | } |
976 | } |
740 | |
977 | |
741 | fail: |
978 | fail: |
742 | return 0; |
979 | return 0; |
743 | } |
980 | } |
744 | |
981 | |
745 | static SV * |
982 | static SV * |
746 | decode_json (SV *string, UV flags) |
983 | decode_json (SV *string, U32 flags) |
747 | { |
984 | { |
|
|
985 | dec_t dec; |
748 | SV *sv; |
986 | SV *sv; |
749 | |
987 | |
|
|
988 | SvUPGRADE (string, SVt_PV); |
|
|
989 | |
750 | if (!(flags & F_UTF8)) |
990 | if (flags & F_UTF8) |
|
|
991 | sv_utf8_downgrade (string, 0); |
|
|
992 | else |
751 | sv_utf8_upgrade (string); |
993 | sv_utf8_upgrade (string); |
752 | |
994 | |
753 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
995 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
754 | |
996 | |
755 | dec_t dec; |
|
|
756 | dec.flags = flags; |
997 | dec.flags = flags; |
757 | dec.cur = SvPVX (string); |
998 | dec.cur = SvPVX (string); |
758 | dec.end = SvEND (string); |
999 | dec.end = SvEND (string); |
759 | dec.err = 0; |
1000 | dec.err = 0; |
|
|
1001 | dec.depth = 0; |
|
|
1002 | dec.maxdepth = DEC_DEPTH (dec.flags); |
760 | |
1003 | |
761 | *dec.end = 1; // invalid anywhere |
1004 | *dec.end = 0; // this should basically be a nop, too, but make sure its there |
762 | sv = decode_sv (&dec); |
1005 | sv = decode_sv (&dec); |
763 | *dec.end = 0; |
|
|
764 | |
1006 | |
765 | if (!sv) |
1007 | if (!sv) |
766 | { |
1008 | { |
|
|
1009 | IV offset = dec.flags & F_UTF8 |
|
|
1010 | ? dec.cur - SvPVX (string) |
767 | IV offset = utf8_distance (dec.cur, SvPVX (string)); |
1011 | : utf8_distance (dec.cur, SvPVX (string)); |
768 | SV *uni = sv_newmortal (); |
1012 | SV *uni = sv_newmortal (); |
769 | |
1013 | |
|
|
1014 | // horrible hack to silence warning inside pv_uni_display |
|
|
1015 | COP cop = *PL_curcop; |
|
|
1016 | cop.cop_warnings = pWARN_NONE; |
|
|
1017 | ENTER; |
|
|
1018 | SAVEVPTR (PL_curcop); |
|
|
1019 | PL_curcop = &cop; |
770 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1020 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
|
|
1021 | LEAVE; |
|
|
1022 | |
771 | croak ("%s, at character %d (%s)", |
1023 | croak ("%s, at character offset %d [\"%s\"]", |
772 | dec.err, |
1024 | dec.err, |
773 | (int)offset, |
1025 | (int)offset, |
774 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1026 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
775 | } |
1027 | } |
776 | |
1028 | |
777 | sv_dump (sv);//D |
|
|
778 | return sv_2mortal (sv); |
1029 | sv = sv_2mortal (sv); |
|
|
1030 | |
|
|
1031 | if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
|
|
1032 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
|
|
1033 | |
|
|
1034 | return sv; |
779 | } |
1035 | } |
|
|
1036 | |
|
|
1037 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1038 | // XS interface functions |
780 | |
1039 | |
781 | MODULE = JSON::XS PACKAGE = JSON::XS |
1040 | MODULE = JSON::XS PACKAGE = JSON::XS |
782 | |
1041 | |
783 | BOOT: |
1042 | BOOT: |
784 | { |
1043 | { |
785 | int i; |
1044 | int i; |
786 | |
1045 | |
787 | memset (decode_hexdigit, 0xff, 256); |
1046 | memset (decode_hexdigit, 0xff, 256); |
|
|
1047 | |
788 | for (i = 10; i--; ) |
1048 | for (i = 0; i < 256; ++i) |
789 | decode_hexdigit ['0' + i] = i; |
1049 | decode_hexdigit [i] = |
790 | |
1050 | i >= '0' && i <= '9' ? i - '0' |
791 | for (i = 6; --i; ) |
1051 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
792 | { |
1052 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
793 | decode_hexdigit ['a' + i] = 10 + i; |
1053 | : -1; |
794 | decode_hexdigit ['A' + i] = 10 + i; |
|
|
795 | } |
|
|
796 | |
1054 | |
797 | json_stash = gv_stashpv ("JSON::XS", 1); |
1055 | json_stash = gv_stashpv ("JSON::XS", 1); |
798 | } |
1056 | } |
|
|
1057 | |
|
|
1058 | PROTOTYPES: DISABLE |
799 | |
1059 | |
800 | SV *new (char *dummy) |
1060 | SV *new (char *dummy) |
801 | CODE: |
1061 | CODE: |
802 | RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); |
1062 | RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); |
803 | OUTPUT: |
1063 | OUTPUT: |
804 | RETVAL |
1064 | RETVAL |
805 | |
1065 | |
806 | SV *ascii (SV *self, int enable) |
1066 | SV *ascii (SV *self, int enable = 1) |
807 | ALIAS: |
1067 | ALIAS: |
808 | ascii = F_ASCII |
1068 | ascii = F_ASCII |
809 | utf8 = F_UTF8 |
1069 | utf8 = F_UTF8 |
810 | indent = F_INDENT |
1070 | indent = F_INDENT |
811 | canonical = F_CANONICAL |
1071 | canonical = F_CANONICAL |
812 | space_before = F_SPACE_BEFORE |
1072 | space_before = F_SPACE_BEFORE |
813 | space_after = F_SPACE_AFTER |
1073 | space_after = F_SPACE_AFTER |
814 | json_rpc = F_JSON_RPC |
|
|
815 | pretty = F_PRETTY |
1074 | pretty = F_PRETTY |
|
|
1075 | allow_nonref = F_ALLOW_NONREF |
|
|
1076 | shrink = F_SHRINK |
816 | CODE: |
1077 | CODE: |
817 | { |
1078 | { |
818 | UV *uv = SvJSON (self); |
1079 | UV *uv = SvJSON (self); |
819 | if (enable) |
1080 | if (enable) |
820 | *uv |= ix; |
1081 | *uv |= ix; |
… | |
… | |
824 | RETVAL = newSVsv (self); |
1085 | RETVAL = newSVsv (self); |
825 | } |
1086 | } |
826 | OUTPUT: |
1087 | OUTPUT: |
827 | RETVAL |
1088 | RETVAL |
828 | |
1089 | |
|
|
1090 | SV *max_depth (SV *self, UV max_depth = 0x80000000UL) |
|
|
1091 | CODE: |
|
|
1092 | { |
|
|
1093 | UV *uv = SvJSON (self); |
|
|
1094 | UV log2 = 0; |
|
|
1095 | |
|
|
1096 | if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
|
|
1097 | |
|
|
1098 | while ((1UL << log2) < max_depth) |
|
|
1099 | ++log2; |
|
|
1100 | |
|
|
1101 | *uv = *uv & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
|
|
1102 | |
|
|
1103 | RETVAL = newSVsv (self); |
|
|
1104 | } |
|
|
1105 | OUTPUT: |
|
|
1106 | RETVAL |
|
|
1107 | |
829 | void encode (SV *self, SV *scalar) |
1108 | void encode (SV *self, SV *scalar) |
830 | PPCODE: |
1109 | PPCODE: |
831 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
1110 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
832 | |
1111 | |
833 | void decode (SV *self, SV *jsonstr) |
1112 | void decode (SV *self, SV *jsonstr) |
834 | PPCODE: |
1113 | PPCODE: |
835 | XPUSHs (decode_json (jsonstr, *SvJSON (self))); |
1114 | XPUSHs (decode_json (jsonstr, *SvJSON (self))); |
836 | |
1115 | |
|
|
1116 | PROTOTYPES: ENABLE |
|
|
1117 | |
837 | void to_json (SV *scalar) |
1118 | void to_json (SV *scalar) |
|
|
1119 | ALIAS: |
|
|
1120 | objToJson = 0 |
838 | PPCODE: |
1121 | PPCODE: |
839 | XPUSHs (encode_json (scalar, F_UTF8)); |
1122 | XPUSHs (encode_json (scalar, F_DEFAULT | F_UTF8)); |
840 | |
1123 | |
841 | void from_json (SV *jsonstr) |
1124 | void from_json (SV *jsonstr) |
|
|
1125 | ALIAS: |
|
|
1126 | jsonToObj = 0 |
842 | PPCODE: |
1127 | PPCODE: |
843 | XPUSHs (decode_json (jsonstr, F_UTF8)); |
1128 | XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); |
844 | |
1129 | |