… | |
… | |
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 | |
|
|
14 | // some old perls do not have this, try to make it work, no |
|
|
15 | // guarentees, though. if it breaks, you get to keep the pieces. |
|
|
16 | #ifndef UTF8_MAXBYTES |
|
|
17 | # define UTF8_MAXBYTES 13 |
|
|
18 | #endif |
|
|
19 | |
9 | #define F_ASCII 0x00000001 |
20 | #define F_ASCII 0x00000001UL |
|
|
21 | #define F_LATIN1 0x00000002UL |
10 | #define F_UTF8 0x00000002 |
22 | #define F_UTF8 0x00000004UL |
11 | #define F_INDENT 0x00000004 |
23 | #define F_INDENT 0x00000008UL |
12 | #define F_CANONICAL 0x00000008 |
24 | #define F_CANONICAL 0x00000010UL |
13 | #define F_SPACE_BEFORE 0x00000010 |
25 | #define F_SPACE_BEFORE 0x00000020UL |
14 | #define F_SPACE_AFTER 0x00000020 |
26 | #define F_SPACE_AFTER 0x00000040UL |
15 | #define F_JSON_RPC 0x00000040 |
|
|
16 | #define F_ALLOW_NONREF 0x00000080 |
27 | #define F_ALLOW_NONREF 0x00000100UL |
|
|
28 | #define F_SHRINK 0x00000200UL |
|
|
29 | #define F_ALLOW_BLESSED 0x00000400UL |
|
|
30 | #define F_CONV_BLESSED 0x00000800UL |
|
|
31 | #define F_MAXDEPTH 0xf8000000UL |
|
|
32 | #define S_MAXDEPTH 27 |
|
|
33 | #define F_MAXSIZE 0x01f00000UL |
|
|
34 | #define S_MAXSIZE 20 |
|
|
35 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
|
|
36 | |
|
|
37 | #define DEC_DEPTH(flags) (1UL << ((flags & F_MAXDEPTH) >> S_MAXDEPTH)) |
|
|
38 | #define DEC_SIZE(flags) (1UL << ((flags & F_MAXSIZE ) >> S_MAXSIZE )) |
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 |
41 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
20 | |
42 | |
21 | #define INIT_SIZE 32 // initial scalar size to be allocated |
43 | #define INIT_SIZE 32 // initial scalar size to be allocated |
|
|
44 | #define INDENT_STEP 3 // spaces per indentation level |
|
|
45 | |
|
|
46 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
22 | |
47 | |
23 | #define SB do { |
48 | #define SB do { |
24 | #define SE } while (0) |
49 | #define SE } while (0) |
25 | |
50 | |
26 | static HV *json_stash; |
51 | #if __GNUC__ >= 3 |
|
|
52 | # define expect(expr,value) __builtin_expect ((expr),(value)) |
|
|
53 | # define inline inline |
|
|
54 | #else |
|
|
55 | # define expect(expr,value) (expr) |
|
|
56 | # define inline static |
|
|
57 | #endif |
|
|
58 | |
|
|
59 | #define expect_false(expr) expect ((expr) != 0, 0) |
|
|
60 | #define expect_true(expr) expect ((expr) != 0, 1) |
|
|
61 | |
|
|
62 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
|
|
63 | static SV *json_true, *json_false; |
|
|
64 | |
|
|
65 | typedef struct { |
|
|
66 | U32 flags; |
|
|
67 | SV *cb_object; |
|
|
68 | HV *cb_sk_object; |
|
|
69 | } JSON; |
|
|
70 | |
|
|
71 | ///////////////////////////////////////////////////////////////////////////// |
|
|
72 | // utility functions |
|
|
73 | |
|
|
74 | inline void |
|
|
75 | shrink (SV *sv) |
|
|
76 | { |
|
|
77 | sv_utf8_downgrade (sv, 1); |
|
|
78 | if (SvLEN (sv) > SvCUR (sv) + 1) |
|
|
79 | { |
|
|
80 | #ifdef SvPV_shrink_to_cur |
|
|
81 | SvPV_shrink_to_cur (sv); |
|
|
82 | #elif defined (SvPV_renew) |
|
|
83 | SvPV_renew (sv, SvCUR (sv) + 1); |
|
|
84 | #endif |
|
|
85 | } |
|
|
86 | } |
|
|
87 | |
|
|
88 | // decode an utf-8 character and return it, or (UV)-1 in |
|
|
89 | // case of an error. |
|
|
90 | // we special-case "safe" characters from U+80 .. U+7FF, |
|
|
91 | // but use the very good perl function to parse anything else. |
|
|
92 | // note that we never call this function for a ascii codepoints |
|
|
93 | inline UV |
|
|
94 | decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen) |
|
|
95 | { |
|
|
96 | if (expect_false (s[0] > 0xdf || s[0] < 0xc2)) |
|
|
97 | return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY); |
|
|
98 | else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf) |
|
|
99 | { |
|
|
100 | *clen = 2; |
|
|
101 | return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); |
|
|
102 | } |
|
|
103 | else |
|
|
104 | { |
|
|
105 | *clen = (STRLEN)-1; |
|
|
106 | return (UV)-1; |
|
|
107 | } |
|
|
108 | } |
|
|
109 | |
|
|
110 | ///////////////////////////////////////////////////////////////////////////// |
|
|
111 | // encoder |
27 | |
112 | |
28 | // structure used for encoding JSON |
113 | // structure used for encoding JSON |
29 | typedef struct |
114 | typedef struct |
30 | { |
115 | { |
31 | char *cur; |
116 | char *cur; // SvPVX (sv) + current output position |
32 | STRLEN len; // SvLEN (sv) |
|
|
33 | char *end; // SvEND (sv) |
117 | char *end; // SvEND (sv) |
34 | SV *sv; |
118 | SV *sv; // result scalar |
35 | UV flags; |
119 | JSON json; |
36 | int max_recurse; |
120 | U32 indent; // indentation level |
37 | int indent; |
121 | U32 maxdepth; // max. indentation/recursion level |
38 | } enc_t; |
122 | } enc_t; |
39 | |
123 | |
40 | // structure used for decoding JSON |
124 | inline void |
41 | typedef struct |
|
|
42 | { |
|
|
43 | char *cur; |
|
|
44 | char *end; |
|
|
45 | const char *err; |
|
|
46 | UV flags; |
|
|
47 | } dec_t; |
|
|
48 | |
|
|
49 | static UV * |
|
|
50 | SvJSON (SV *sv) |
|
|
51 | { |
|
|
52 | if (!(SvROK (sv) && SvOBJECT (SvRV (sv)) && SvSTASH (SvRV (sv)) == json_stash)) |
|
|
53 | croak ("object is not of type JSON::XS"); |
|
|
54 | |
|
|
55 | return &SvUVX (SvRV (sv)); |
|
|
56 | } |
|
|
57 | |
|
|
58 | ///////////////////////////////////////////////////////////////////////////// |
|
|
59 | |
|
|
60 | static void |
|
|
61 | need (enc_t *enc, STRLEN len) |
125 | need (enc_t *enc, STRLEN len) |
62 | { |
126 | { |
63 | if (enc->cur + len >= enc->end) |
127 | if (expect_false (enc->cur + len >= enc->end)) |
64 | { |
128 | { |
65 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
129 | STRLEN cur = enc->cur - SvPVX (enc->sv); |
66 | SvGROW (enc->sv, cur + len + 1); |
130 | SvGROW (enc->sv, cur + len + 1); |
67 | enc->cur = SvPVX (enc->sv) + cur; |
131 | enc->cur = SvPVX (enc->sv) + cur; |
68 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); |
132 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
69 | } |
133 | } |
70 | } |
134 | } |
71 | |
135 | |
72 | static void |
136 | inline void |
73 | encode_ch (enc_t *enc, char ch) |
137 | encode_ch (enc_t *enc, char ch) |
74 | { |
138 | { |
75 | need (enc, 1); |
139 | need (enc, 1); |
76 | *enc->cur++ = ch; |
140 | *enc->cur++ = ch; |
77 | } |
141 | } |
… | |
… | |
85 | |
149 | |
86 | while (str < end) |
150 | while (str < end) |
87 | { |
151 | { |
88 | unsigned char ch = *(unsigned char *)str; |
152 | unsigned char ch = *(unsigned char *)str; |
89 | |
153 | |
90 | if (ch == '"') |
154 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
91 | { |
155 | { |
|
|
156 | if (expect_false (ch == '"')) // but with slow exceptions |
|
|
157 | { |
92 | need (enc, len += 1); |
158 | need (enc, len += 1); |
93 | *enc->cur++ = '\\'; |
159 | *enc->cur++ = '\\'; |
94 | *enc->cur++ = '"'; |
160 | *enc->cur++ = '"'; |
95 | ++str; |
|
|
96 | } |
161 | } |
97 | else if (ch == '\\') |
162 | else if (expect_false (ch == '\\')) |
98 | { |
163 | { |
99 | need (enc, len += 1); |
164 | need (enc, len += 1); |
100 | *enc->cur++ = '\\'; |
165 | *enc->cur++ = '\\'; |
101 | *enc->cur++ = '\\'; |
166 | *enc->cur++ = '\\'; |
102 | ++str; |
|
|
103 | } |
167 | } |
104 | else if (ch >= 0x20 && ch < 0x80) // most common case |
168 | else |
105 | { |
|
|
106 | *enc->cur++ = ch; |
169 | *enc->cur++ = ch; |
107 | ++str; |
170 | |
108 | } |
|
|
109 | else if (ch == '\015') |
|
|
110 | { |
|
|
111 | need (enc, len += 1); |
|
|
112 | *enc->cur++ = '\\'; |
|
|
113 | *enc->cur++ = 'r'; |
|
|
114 | ++str; |
|
|
115 | } |
|
|
116 | else if (ch == '\012') |
|
|
117 | { |
|
|
118 | need (enc, len += 1); |
|
|
119 | *enc->cur++ = '\\'; |
|
|
120 | *enc->cur++ = 'n'; |
|
|
121 | ++str; |
171 | ++str; |
122 | } |
172 | } |
123 | else |
173 | else |
124 | { |
174 | { |
125 | STRLEN clen; |
175 | switch (ch) |
126 | UV uch; |
|
|
127 | |
|
|
128 | if (is_utf8) |
|
|
129 | { |
|
|
130 | uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); |
|
|
131 | if (clen < 0) |
|
|
132 | croak ("malformed UTF-8 character in string, cannot convert to JSON"); |
|
|
133 | } |
176 | { |
134 | else |
177 | case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; |
135 | { |
178 | case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; |
136 | uch = ch; |
179 | case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; |
137 | clen = 1; |
180 | case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; |
138 | } |
181 | case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; |
139 | |
182 | |
140 | if (uch < 0x80 || enc->flags & F_ASCII) |
183 | default: |
141 | { |
|
|
142 | if (uch > 0xFFFFUL) |
|
|
143 | { |
184 | { |
|
|
185 | STRLEN clen; |
|
|
186 | UV uch; |
|
|
187 | |
|
|
188 | if (is_utf8) |
|
|
189 | { |
|
|
190 | uch = decode_utf8 (str, end - str, &clen); |
|
|
191 | if (clen == (STRLEN)-1) |
|
|
192 | croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
|
|
193 | } |
|
|
194 | else |
|
|
195 | { |
|
|
196 | uch = ch; |
|
|
197 | clen = 1; |
|
|
198 | } |
|
|
199 | |
|
|
200 | if (uch > 0x10FFFFUL) |
|
|
201 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
|
|
202 | |
|
|
203 | if (uch < 0x80 || enc->json.flags & F_ASCII || (enc->json.flags & F_LATIN1 && uch > 0xFF)) |
|
|
204 | { |
|
|
205 | if (uch > 0xFFFFUL) |
|
|
206 | { |
144 | need (enc, len += 11); |
207 | need (enc, len += 11); |
145 | sprintf (enc->cur, "\\u%04x\\u%04x", |
208 | sprintf (enc->cur, "\\u%04x\\u%04x", |
146 | (uch - 0x10000) / 0x400 + 0xD800, |
209 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
147 | (uch - 0x10000) % 0x400 + 0xDC00); |
210 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
148 | enc->cur += 12; |
211 | enc->cur += 12; |
|
|
212 | } |
|
|
213 | else |
|
|
214 | { |
|
|
215 | static char hexdigit [16] = "0123456789abcdef"; |
|
|
216 | need (enc, len += 5); |
|
|
217 | *enc->cur++ = '\\'; |
|
|
218 | *enc->cur++ = 'u'; |
|
|
219 | *enc->cur++ = hexdigit [ uch >> 12 ]; |
|
|
220 | *enc->cur++ = hexdigit [(uch >> 8) & 15]; |
|
|
221 | *enc->cur++ = hexdigit [(uch >> 4) & 15]; |
|
|
222 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
|
|
223 | } |
|
|
224 | |
|
|
225 | str += clen; |
|
|
226 | } |
|
|
227 | else if (enc->json.flags & F_LATIN1) |
|
|
228 | { |
|
|
229 | *enc->cur++ = uch; |
|
|
230 | str += clen; |
|
|
231 | } |
|
|
232 | else if (is_utf8) |
|
|
233 | { |
|
|
234 | need (enc, len += clen); |
|
|
235 | do |
|
|
236 | { |
|
|
237 | *enc->cur++ = *str++; |
|
|
238 | } |
|
|
239 | while (--clen); |
|
|
240 | } |
|
|
241 | else |
|
|
242 | { |
|
|
243 | need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
|
|
244 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
|
|
245 | ++str; |
|
|
246 | } |
149 | } |
247 | } |
150 | else |
|
|
151 | { |
|
|
152 | static char hexdigit [16] = "0123456789abcdef"; |
|
|
153 | need (enc, len += 5); |
|
|
154 | *enc->cur++ = '\\'; |
|
|
155 | *enc->cur++ = 'u'; |
|
|
156 | *enc->cur++ = hexdigit [ uch >> 12 ]; |
|
|
157 | *enc->cur++ = hexdigit [(uch >> 8) & 15]; |
|
|
158 | *enc->cur++ = hexdigit [(uch >> 4) & 15]; |
|
|
159 | *enc->cur++ = hexdigit [(uch >> 0) & 15]; |
|
|
160 | } |
|
|
161 | |
|
|
162 | str += clen; |
|
|
163 | } |
|
|
164 | else if (is_utf8) |
|
|
165 | { |
|
|
166 | need (enc, len += clen); |
|
|
167 | while (clen--) |
|
|
168 | *enc->cur++ = *str++; |
|
|
169 | } |
|
|
170 | else |
|
|
171 | { |
|
|
172 | need (enc, 10); // never more than 11 bytes needed |
|
|
173 | enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); |
|
|
174 | ++str; |
|
|
175 | } |
248 | } |
176 | } |
249 | } |
177 | |
250 | |
178 | --len; |
251 | --len; |
179 | } |
252 | } |
180 | } |
253 | } |
181 | |
254 | |
182 | #define INDENT SB \ |
255 | inline void |
|
|
256 | encode_indent (enc_t *enc) |
|
|
257 | { |
183 | if (enc->flags & F_INDENT) \ |
258 | if (enc->json.flags & F_INDENT) |
184 | { \ |
259 | { |
185 | int i_; \ |
260 | int spaces = enc->indent * INDENT_STEP; |
186 | need (enc, enc->indent); \ |
261 | |
187 | for (i_ = enc->indent * 3; i_--; )\ |
262 | need (enc, spaces); |
|
|
263 | memset (enc->cur, ' ', spaces); |
|
|
264 | enc->cur += spaces; |
|
|
265 | } |
|
|
266 | } |
|
|
267 | |
|
|
268 | inline void |
|
|
269 | encode_space (enc_t *enc) |
|
|
270 | { |
|
|
271 | need (enc, 1); |
|
|
272 | encode_ch (enc, ' '); |
|
|
273 | } |
|
|
274 | |
|
|
275 | inline void |
|
|
276 | encode_nl (enc_t *enc) |
|
|
277 | { |
|
|
278 | if (enc->json.flags & F_INDENT) |
|
|
279 | { |
|
|
280 | need (enc, 1); |
188 | encode_ch (enc, ' '); \ |
281 | encode_ch (enc, '\n'); |
189 | } \ |
282 | } |
190 | SE |
283 | } |
191 | |
284 | |
192 | #define SPACE SB need (enc, 1); encode_ch (enc, ' '); SE |
285 | inline void |
193 | #define NL SB if (enc->flags & F_INDENT) { need (enc, 1); encode_ch (enc, '\n'); } SE |
286 | encode_comma (enc_t *enc) |
194 | #define COMMA SB \ |
287 | { |
195 | encode_ch (enc, ','); \ |
288 | encode_ch (enc, ','); |
|
|
289 | |
196 | if (enc->flags & F_INDENT) \ |
290 | if (enc->json.flags & F_INDENT) |
197 | NL; \ |
291 | encode_nl (enc); |
198 | else if (enc->flags & F_SPACE_AFTER) \ |
292 | else if (enc->json.flags & F_SPACE_AFTER) |
199 | SPACE; \ |
293 | encode_space (enc); |
200 | SE |
294 | } |
201 | |
295 | |
202 | static void encode_sv (enc_t *enc, SV *sv); |
296 | static void encode_sv (enc_t *enc, SV *sv); |
203 | |
297 | |
204 | static void |
298 | static void |
205 | encode_av (enc_t *enc, AV *av) |
299 | encode_av (enc_t *enc, AV *av) |
206 | { |
300 | { |
207 | int i, len = av_len (av); |
301 | int i, len = av_len (av); |
208 | |
302 | |
209 | encode_ch (enc, '['); NL; |
303 | if (enc->indent >= enc->maxdepth) |
|
|
304 | croak ("data structure too deep (hit recursion limit)"); |
|
|
305 | |
|
|
306 | encode_ch (enc, '['); encode_nl (enc); |
210 | ++enc->indent; |
307 | ++enc->indent; |
211 | |
308 | |
212 | for (i = 0; i <= len; ++i) |
309 | for (i = 0; i <= len; ++i) |
213 | { |
310 | { |
214 | INDENT; |
311 | encode_indent (enc); |
215 | encode_sv (enc, *av_fetch (av, i, 0)); |
312 | encode_sv (enc, *av_fetch (av, i, 0)); |
216 | |
313 | |
217 | if (i < len) |
314 | if (i < len) |
218 | COMMA; |
315 | encode_comma (enc); |
219 | } |
316 | } |
220 | |
317 | |
221 | NL; |
318 | encode_nl (enc); |
222 | |
319 | |
223 | --enc->indent; |
320 | --enc->indent; |
224 | INDENT; encode_ch (enc, ']'); |
321 | encode_indent (enc); encode_ch (enc, ']'); |
225 | } |
322 | } |
226 | |
323 | |
227 | static void |
324 | static void |
228 | encode_he (enc_t *enc, HE *he) |
325 | encode_he (enc_t *enc, HE *he) |
229 | { |
326 | { |
… | |
… | |
243 | else |
340 | else |
244 | encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
341 | encode_str (enc, HeKEY (he), HeKLEN (he), HeKUTF8 (he)); |
245 | |
342 | |
246 | encode_ch (enc, '"'); |
343 | encode_ch (enc, '"'); |
247 | |
344 | |
248 | if (enc->flags & F_SPACE_BEFORE) SPACE; |
345 | if (enc->json.flags & F_SPACE_BEFORE) encode_space (enc); |
249 | encode_ch (enc, ':'); |
346 | encode_ch (enc, ':'); |
250 | if (enc->flags & F_SPACE_AFTER ) SPACE; |
347 | if (enc->json.flags & F_SPACE_AFTER ) encode_space (enc); |
251 | encode_sv (enc, HeVAL (he)); |
348 | encode_sv (enc, HeVAL (he)); |
252 | } |
349 | } |
253 | |
350 | |
254 | // compare hash entries, used when all keys are bytestrings |
351 | // compare hash entries, used when all keys are bytestrings |
255 | static int |
352 | static int |
… | |
… | |
261 | HE *b = *(HE **)b_; |
358 | HE *b = *(HE **)b_; |
262 | |
359 | |
263 | STRLEN la = HeKLEN (a); |
360 | STRLEN la = HeKLEN (a); |
264 | STRLEN lb = HeKLEN (b); |
361 | STRLEN lb = HeKLEN (b); |
265 | |
362 | |
266 | if (!(cmp == memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) |
363 | if (!(cmp = memcmp (HeKEY (a), HeKEY (b), la < lb ? la : lb))) |
267 | cmp = la < lb ? -1 : la == lb ? 0 : 1; |
364 | cmp = la - lb; |
268 | |
365 | |
269 | return cmp; |
366 | return cmp; |
270 | } |
367 | } |
271 | |
368 | |
272 | // compare hash entries, used when some keys are sv's or utf-x |
369 | // compare hash entries, used when some keys are sv's or utf-x |
… | |
… | |
279 | static void |
376 | static void |
280 | encode_hv (enc_t *enc, HV *hv) |
377 | encode_hv (enc_t *enc, HV *hv) |
281 | { |
378 | { |
282 | int count, i; |
379 | int count, i; |
283 | |
380 | |
|
|
381 | if (enc->indent >= enc->maxdepth) |
|
|
382 | croak ("data structure too deep (hit recursion limit)"); |
|
|
383 | |
284 | encode_ch (enc, '{'); NL; ++enc->indent; |
384 | encode_ch (enc, '{'); encode_nl (enc); ++enc->indent; |
285 | |
385 | |
286 | if ((count = hv_iterinit (hv))) |
386 | if ((count = hv_iterinit (hv))) |
287 | { |
387 | { |
288 | // for canonical output we have to sort by keys first |
388 | // for canonical output we have to sort by keys first |
289 | // actually, this is mostly due to the stupid so-called |
389 | // actually, this is mostly due to the stupid so-called |
290 | // security workaround added somewhere in 5.8.x. |
390 | // security workaround added somewhere in 5.8.x. |
291 | // that randomises hash orderings |
391 | // that randomises hash orderings |
292 | if (enc->flags & F_CANONICAL) |
392 | if (enc->json.flags & F_CANONICAL) |
293 | { |
393 | { |
294 | HE *he, *hes [count]; |
|
|
295 | int fast = 1; |
394 | int fast = 1; |
|
|
395 | HE *he; |
|
|
396 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
|
|
397 | HE **hes = _alloca (count * sizeof (HE)); |
|
|
398 | #else |
|
|
399 | HE *hes [count]; // if your compiler dies here, you need to enable C99 mode |
|
|
400 | #endif |
296 | |
401 | |
297 | i = 0; |
402 | i = 0; |
298 | while ((he = hv_iternext (hv))) |
403 | while ((he = hv_iternext (hv))) |
299 | { |
404 | { |
300 | hes [i++] = he; |
405 | hes [i++] = he; |
… | |
… | |
306 | |
411 | |
307 | if (fast) |
412 | if (fast) |
308 | qsort (hes, count, sizeof (HE *), he_cmp_fast); |
413 | qsort (hes, count, sizeof (HE *), he_cmp_fast); |
309 | else |
414 | else |
310 | { |
415 | { |
311 | // hack to disable "use bytes" |
416 | // hack to forcefully disable "use bytes" |
312 | COP *oldcop = PL_curcop, cop; |
417 | COP cop = *PL_curcop; |
313 | cop.op_private = 0; |
418 | cop.op_private = 0; |
|
|
419 | |
|
|
420 | ENTER; |
|
|
421 | SAVETMPS; |
|
|
422 | |
|
|
423 | SAVEVPTR (PL_curcop); |
314 | PL_curcop = &cop; |
424 | PL_curcop = &cop; |
315 | |
425 | |
316 | SAVETMPS; |
|
|
317 | qsort (hes, count, sizeof (HE *), he_cmp_slow); |
426 | qsort (hes, count, sizeof (HE *), he_cmp_slow); |
|
|
427 | |
318 | FREETMPS; |
428 | FREETMPS; |
319 | |
429 | LEAVE; |
320 | PL_curcop = oldcop; |
|
|
321 | } |
430 | } |
322 | |
431 | |
323 | for (i = 0; i < count; ++i) |
432 | for (i = 0; i < count; ++i) |
324 | { |
433 | { |
325 | INDENT; |
434 | encode_indent (enc); |
326 | encode_he (enc, hes [i]); |
435 | encode_he (enc, hes [i]); |
327 | |
436 | |
328 | if (i < count - 1) |
437 | if (i < count - 1) |
329 | COMMA; |
438 | encode_comma (enc); |
330 | } |
|
|
331 | |
|
|
332 | NL; |
439 | } |
|
|
440 | |
|
|
441 | encode_nl (enc); |
333 | } |
442 | } |
334 | else |
443 | else |
335 | { |
444 | { |
336 | SV *sv; |
|
|
337 | HE *he = hv_iternext (hv); |
445 | HE *he = hv_iternext (hv); |
338 | |
446 | |
339 | for (;;) |
447 | for (;;) |
340 | { |
448 | { |
341 | INDENT; |
449 | encode_indent (enc); |
342 | encode_he (enc, he); |
450 | encode_he (enc, he); |
343 | |
451 | |
344 | if (!(he = hv_iternext (hv))) |
452 | if (!(he = hv_iternext (hv))) |
345 | break; |
453 | break; |
346 | |
454 | |
347 | COMMA; |
455 | encode_comma (enc); |
348 | } |
|
|
349 | |
|
|
350 | NL; |
456 | } |
|
|
457 | |
|
|
458 | encode_nl (enc); |
351 | } |
459 | } |
352 | } |
460 | } |
353 | |
461 | |
354 | --enc->indent; INDENT; encode_ch (enc, '}'); |
462 | --enc->indent; encode_indent (enc); encode_ch (enc, '}'); |
|
|
463 | } |
|
|
464 | |
|
|
465 | // encode objects, arrays and special \0=false and \1=true values. |
|
|
466 | static void |
|
|
467 | encode_rv (enc_t *enc, SV *sv) |
|
|
468 | { |
|
|
469 | svtype svt; |
|
|
470 | |
|
|
471 | SvGETMAGIC (sv); |
|
|
472 | svt = SvTYPE (sv); |
|
|
473 | |
|
|
474 | if (expect_false (SvOBJECT (sv))) |
|
|
475 | { |
|
|
476 | if (SvSTASH (sv) == json_boolean_stash) |
|
|
477 | { |
|
|
478 | if (SvIV (sv)) |
|
|
479 | encode_str (enc, "true", 4, 0); |
|
|
480 | else |
|
|
481 | encode_str (enc, "false", 5, 0); |
|
|
482 | } |
|
|
483 | else |
|
|
484 | { |
|
|
485 | #if 0 |
|
|
486 | if (0 && sv_derived_from (rv, "JSON::Literal")) |
|
|
487 | { |
|
|
488 | // not yet |
|
|
489 | } |
|
|
490 | #endif |
|
|
491 | if (enc->json.flags & F_CONV_BLESSED) |
|
|
492 | { |
|
|
493 | // we re-bless the reference to get overload and other niceties right |
|
|
494 | GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 1); |
|
|
495 | |
|
|
496 | if (to_json) |
|
|
497 | { |
|
|
498 | dSP; ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
499 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); |
|
|
500 | |
|
|
501 | // calling with G_SCALAR ensures that we always get a 1 reutrn value |
|
|
502 | // check anyways. |
|
|
503 | PUTBACK; |
|
|
504 | assert (1 == call_sv ((SV *)GvCV (to_json), G_SCALAR)); |
|
|
505 | SPAGAIN; |
|
|
506 | |
|
|
507 | encode_sv (enc, POPs); |
|
|
508 | |
|
|
509 | FREETMPS; LEAVE; |
|
|
510 | } |
|
|
511 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
512 | encode_str (enc, "null", 4, 0); |
|
|
513 | else |
|
|
514 | croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it", |
|
|
515 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
516 | } |
|
|
517 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
518 | encode_str (enc, "null", 4, 0); |
|
|
519 | else |
|
|
520 | croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", |
|
|
521 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
522 | } |
|
|
523 | } |
|
|
524 | else if (svt == SVt_PVHV) |
|
|
525 | encode_hv (enc, (HV *)sv); |
|
|
526 | else if (svt == SVt_PVAV) |
|
|
527 | encode_av (enc, (AV *)sv); |
|
|
528 | else if (svt < SVt_PVAV) |
|
|
529 | { |
|
|
530 | STRLEN len = 0; |
|
|
531 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
532 | |
|
|
533 | if (len == 1 && *pv == '1') |
|
|
534 | encode_str (enc, "true", 4, 0); |
|
|
535 | else if (len == 1 && *pv == '0') |
|
|
536 | encode_str (enc, "false", 5, 0); |
|
|
537 | else |
|
|
538 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
|
|
539 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
540 | } |
|
|
541 | else |
|
|
542 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
|
|
543 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
355 | } |
544 | } |
356 | |
545 | |
357 | static void |
546 | static void |
358 | encode_sv (enc_t *enc, SV *sv) |
547 | encode_sv (enc_t *enc, SV *sv) |
359 | { |
548 | { |
… | |
… | |
367 | encode_str (enc, str, len, SvUTF8 (sv)); |
556 | encode_str (enc, str, len, SvUTF8 (sv)); |
368 | encode_ch (enc, '"'); |
557 | encode_ch (enc, '"'); |
369 | } |
558 | } |
370 | else if (SvNOKp (sv)) |
559 | else if (SvNOKp (sv)) |
371 | { |
560 | { |
|
|
561 | // trust that perl will do the right thing w.r.t. JSON syntax. |
372 | need (enc, NV_DIG + 32); |
562 | need (enc, NV_DIG + 32); |
373 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
563 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
374 | enc->cur += strlen (enc->cur); |
564 | enc->cur += strlen (enc->cur); |
375 | } |
565 | } |
376 | else if (SvIOKp (sv)) |
566 | else if (SvIOKp (sv)) |
377 | { |
567 | { |
378 | need (enc, 64); |
568 | // we assume we can always read an IV as a UV |
|
|
569 | if (SvUV (sv) & ~(UV)0x7fff) |
|
|
570 | { |
|
|
571 | // large integer, use the (rather slow) snprintf way. |
|
|
572 | need (enc, sizeof (UV) * 3); |
379 | enc->cur += |
573 | enc->cur += |
380 | SvIsUV(sv) |
574 | SvIsUV(sv) |
381 | ? snprintf (enc->cur, 64, "%"UVuf, (UV)SvUVX (sv)) |
575 | ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv)) |
382 | : snprintf (enc->cur, 64, "%"IVdf, (IV)SvIVX (sv)); |
576 | : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv)); |
|
|
577 | } |
|
|
578 | else |
|
|
579 | { |
|
|
580 | // optimise the "small number case" |
|
|
581 | // code will likely be branchless and use only a single multiplication |
|
|
582 | I32 i = SvIV (sv); |
|
|
583 | U32 u; |
|
|
584 | char digit, nz = 0; |
|
|
585 | |
|
|
586 | need (enc, 6); |
|
|
587 | |
|
|
588 | *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0; |
|
|
589 | u = i < 0 ? -i : i; |
|
|
590 | |
|
|
591 | // convert to 4.28 fixed-point representation |
|
|
592 | u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits |
|
|
593 | |
|
|
594 | // now output digit by digit, each time masking out the integer part |
|
|
595 | // and multiplying by 5 while moving the decimal point one to the right, |
|
|
596 | // resulting in a net multiplication by 10. |
|
|
597 | // we always write the digit to memory but conditionally increment |
|
|
598 | // the pointer, to ease the usage of conditional move instructions. |
|
|
599 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; |
|
|
600 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; |
|
|
601 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; |
|
|
602 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; |
|
|
603 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
|
|
604 | } |
383 | } |
605 | } |
384 | else if (SvROK (sv)) |
606 | else if (SvROK (sv)) |
385 | { |
607 | encode_rv (enc, SvRV (sv)); |
386 | if (!--enc->max_recurse) |
|
|
387 | croak ("data structure too deep (hit recursion limit)"); |
|
|
388 | |
|
|
389 | sv = SvRV (sv); |
|
|
390 | |
|
|
391 | switch (SvTYPE (sv)) |
|
|
392 | { |
|
|
393 | case SVt_PVAV: encode_av (enc, (AV *)sv); break; |
|
|
394 | case SVt_PVHV: encode_hv (enc, (HV *)sv); break; |
|
|
395 | |
|
|
396 | default: |
|
|
397 | croak ("JSON can only represent references to arrays or hashes"); |
|
|
398 | } |
|
|
399 | } |
|
|
400 | else if (!SvOK (sv)) |
608 | else if (!SvOK (sv)) |
401 | encode_str (enc, "null", 4, 0); |
609 | encode_str (enc, "null", 4, 0); |
402 | else |
610 | else |
403 | croak ("encountered perl type that JSON cannot handle"); |
611 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
|
|
612 | SvPV_nolen (sv), SvFLAGS (sv)); |
404 | } |
613 | } |
405 | |
614 | |
406 | static SV * |
615 | static SV * |
407 | encode_json (SV *scalar, UV flags) |
616 | encode_json (SV *scalar, JSON *json) |
408 | { |
617 | { |
409 | if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
|
|
410 | croak ("hash- or arraref required (not a simple scalar, use allow_nonref to allow this)"); |
|
|
411 | |
|
|
412 | enc_t enc; |
618 | enc_t enc; |
413 | enc.flags = flags; |
619 | |
|
|
620 | if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
|
|
621 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
|
|
622 | |
|
|
623 | enc.json = *json; |
414 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
624 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
415 | enc.cur = SvPVX (enc.sv); |
625 | enc.cur = SvPVX (enc.sv); |
416 | enc.end = SvEND (enc.sv); |
626 | enc.end = SvEND (enc.sv); |
417 | enc.max_recurse = 0; |
|
|
418 | enc.indent = 0; |
627 | enc.indent = 0; |
|
|
628 | enc.maxdepth = DEC_DEPTH (enc.json.flags); |
419 | |
629 | |
420 | SvPOK_only (enc.sv); |
630 | SvPOK_only (enc.sv); |
421 | encode_sv (&enc, scalar); |
631 | encode_sv (&enc, scalar); |
422 | |
632 | |
|
|
633 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
|
|
634 | *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings |
|
|
635 | |
423 | if (!(flags & (F_ASCII | F_UTF8))) |
636 | if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) |
424 | SvUTF8_on (enc.sv); |
637 | SvUTF8_on (enc.sv); |
425 | |
638 | |
426 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
639 | if (enc.json.flags & F_SHRINK) |
|
|
640 | shrink (enc.sv); |
|
|
641 | |
427 | return enc.sv; |
642 | return enc.sv; |
428 | } |
643 | } |
429 | |
644 | |
430 | ///////////////////////////////////////////////////////////////////////////// |
645 | ///////////////////////////////////////////////////////////////////////////// |
|
|
646 | // decoder |
431 | |
647 | |
432 | #define WS \ |
648 | // structure used for decoding JSON |
|
|
649 | typedef struct |
|
|
650 | { |
|
|
651 | char *cur; // current parser pointer |
|
|
652 | char *end; // end of input string |
|
|
653 | const char *err; // parse error, if != 0 |
|
|
654 | JSON json; |
|
|
655 | U32 depth; // recursion depth |
|
|
656 | U32 maxdepth; // recursion depth limit |
|
|
657 | } dec_t; |
|
|
658 | |
|
|
659 | inline void |
|
|
660 | decode_ws (dec_t *dec) |
|
|
661 | { |
433 | for (;;) \ |
662 | for (;;) |
434 | { \ |
663 | { |
435 | char ch = *dec->cur; \ |
664 | char ch = *dec->cur; |
|
|
665 | |
436 | if (ch > 0x20 \ |
666 | if (ch > 0x20 |
437 | || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) \ |
667 | || (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09)) |
438 | break; \ |
668 | break; |
|
|
669 | |
439 | ++dec->cur; \ |
670 | ++dec->cur; |
440 | } |
671 | } |
|
|
672 | } |
441 | |
673 | |
442 | #define ERR(reason) SB dec->err = reason; goto fail; SE |
674 | #define ERR(reason) SB dec->err = reason; goto fail; SE |
|
|
675 | |
443 | #define EXPECT_CH(ch) SB \ |
676 | #define EXPECT_CH(ch) SB \ |
444 | if (*dec->cur != ch) \ |
677 | if (*dec->cur != ch) \ |
445 | ERR (# ch " expected"); \ |
678 | ERR (# ch " expected"); \ |
446 | ++dec->cur; \ |
679 | ++dec->cur; \ |
447 | SE |
680 | SE |
448 | |
681 | |
|
|
682 | #define DEC_INC_DEPTH if (++dec->depth > dec->maxdepth) ERR ("json datastructure exceeds maximum nesting level (set a higher max_depth)") |
|
|
683 | #define DEC_DEC_DEPTH --dec->depth |
|
|
684 | |
449 | static SV *decode_sv (dec_t *dec); |
685 | static SV *decode_sv (dec_t *dec); |
450 | |
686 | |
451 | static signed char decode_hexdigit[256]; |
687 | static signed char decode_hexdigit[256]; |
452 | |
688 | |
453 | static UV |
689 | static UV |
454 | decode_4hex (dec_t *dec) |
690 | decode_4hex (dec_t *dec) |
455 | { |
691 | { |
456 | signed char d1, d2, d3, d4; |
692 | signed char d1, d2, d3, d4; |
|
|
693 | unsigned char *cur = (unsigned char *)dec->cur; |
457 | |
694 | |
458 | d1 = decode_hexdigit [((unsigned char *)dec->cur) [0]]; |
695 | d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected"); |
459 | if (d1 < 0) ERR ("four hexadecimal digits expected"); |
696 | d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected"); |
460 | d2 = decode_hexdigit [((unsigned char *)dec->cur) [1]]; |
697 | d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected"); |
461 | if (d2 < 0) ERR ("four hexadecimal digits expected"); |
698 | d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected"); |
462 | d3 = decode_hexdigit [((unsigned char *)dec->cur) [2]]; |
|
|
463 | if (d3 < 0) ERR ("four hexadecimal digits expected"); |
|
|
464 | d4 = decode_hexdigit [((unsigned char *)dec->cur) [3]]; |
|
|
465 | if (d4 < 0) ERR ("four hexadecimal digits expected"); |
|
|
466 | |
699 | |
467 | dec->cur += 4; |
700 | dec->cur += 4; |
468 | |
701 | |
469 | return ((UV)d1) << 12 |
702 | return ((UV)d1) << 12 |
470 | | ((UV)d2) << 8 |
703 | | ((UV)d2) << 8 |
… | |
… | |
473 | |
706 | |
474 | fail: |
707 | fail: |
475 | return (UV)-1; |
708 | return (UV)-1; |
476 | } |
709 | } |
477 | |
710 | |
478 | #define APPEND_GROW(n) SB \ |
|
|
479 | if (cur + (n) >= end) \ |
|
|
480 | { \ |
|
|
481 | STRLEN ofs = cur - SvPVX (sv); \ |
|
|
482 | SvGROW (sv, ofs + (n) + 1); \ |
|
|
483 | cur = SvPVX (sv) + ofs; \ |
|
|
484 | end = SvEND (sv); \ |
|
|
485 | } \ |
|
|
486 | SE |
|
|
487 | |
|
|
488 | #define APPEND_CH(ch) SB \ |
|
|
489 | APPEND_GROW (1); \ |
|
|
490 | *cur++ = (ch); \ |
|
|
491 | SE |
|
|
492 | |
|
|
493 | static SV * |
711 | static SV * |
494 | decode_str (dec_t *dec) |
712 | decode_str (dec_t *dec) |
495 | { |
713 | { |
496 | SV *sv = NEWSV (0,2); |
714 | SV *sv = 0; |
497 | int utf8 = 0; |
715 | int utf8 = 0; |
498 | char *cur = SvPVX (sv); |
716 | char *dec_cur = dec->cur; |
499 | char *end = SvEND (sv); |
|
|
500 | |
717 | |
501 | for (;;) |
718 | do |
502 | { |
719 | { |
503 | unsigned char ch = *(unsigned char *)dec->cur; |
720 | char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; |
|
|
721 | char *cur = buf; |
504 | |
722 | |
505 | if (ch == '"') |
723 | do |
506 | break; |
|
|
507 | else if (ch == '\\') |
|
|
508 | { |
724 | { |
509 | switch (*++dec->cur) |
725 | unsigned char ch = *(unsigned char *)dec_cur++; |
|
|
726 | |
|
|
727 | if (expect_false (ch == '"')) |
|
|
728 | { |
|
|
729 | --dec_cur; |
|
|
730 | break; |
510 | { |
731 | } |
511 | case '\\': |
732 | else if (expect_false (ch == '\\')) |
512 | case '/': |
733 | { |
513 | case '"': APPEND_CH (*dec->cur++); break; |
734 | switch (*dec_cur) |
514 | |
|
|
515 | case 'b': APPEND_CH ('\010'); ++dec->cur; break; |
|
|
516 | case 't': APPEND_CH ('\011'); ++dec->cur; break; |
|
|
517 | case 'n': APPEND_CH ('\012'); ++dec->cur; break; |
|
|
518 | case 'f': APPEND_CH ('\014'); ++dec->cur; break; |
|
|
519 | case 'r': APPEND_CH ('\015'); ++dec->cur; break; |
|
|
520 | |
|
|
521 | case 'u': |
|
|
522 | { |
735 | { |
523 | UV lo, hi; |
736 | case '\\': |
524 | ++dec->cur; |
737 | case '/': |
|
|
738 | case '"': *cur++ = *dec_cur++; break; |
525 | |
739 | |
526 | hi = decode_4hex (dec); |
740 | case 'b': ++dec_cur; *cur++ = '\010'; break; |
527 | if (hi == (UV)-1) |
741 | case 't': ++dec_cur; *cur++ = '\011'; break; |
528 | goto fail; |
742 | case 'n': ++dec_cur; *cur++ = '\012'; break; |
|
|
743 | case 'f': ++dec_cur; *cur++ = '\014'; break; |
|
|
744 | case 'r': ++dec_cur; *cur++ = '\015'; break; |
529 | |
745 | |
530 | // possibly a surrogate pair |
746 | case 'u': |
531 | if (hi >= 0xd800 && hi < 0xdc00) |
|
|
532 | { |
747 | { |
533 | if (dec->cur [0] != '\\' || dec->cur [1] != 'u') |
748 | UV lo, hi; |
534 | ERR ("illegal surrogate character"); |
749 | ++dec_cur; |
535 | |
750 | |
536 | dec->cur += 2; |
751 | dec->cur = dec_cur; |
537 | |
|
|
538 | lo = decode_4hex (dec); |
752 | hi = decode_4hex (dec); |
|
|
753 | dec_cur = dec->cur; |
539 | if (lo == (UV)-1) |
754 | if (hi == (UV)-1) |
540 | goto fail; |
755 | goto fail; |
541 | |
756 | |
|
|
757 | // possibly a surrogate pair |
|
|
758 | if (hi >= 0xd800) |
|
|
759 | if (hi < 0xdc00) |
|
|
760 | { |
|
|
761 | if (dec_cur [0] != '\\' || dec_cur [1] != 'u') |
|
|
762 | ERR ("missing low surrogate character in surrogate pair"); |
|
|
763 | |
|
|
764 | dec_cur += 2; |
|
|
765 | |
|
|
766 | dec->cur = dec_cur; |
|
|
767 | lo = decode_4hex (dec); |
|
|
768 | dec_cur = dec->cur; |
|
|
769 | if (lo == (UV)-1) |
|
|
770 | goto fail; |
|
|
771 | |
542 | if (lo < 0xdc00 || lo >= 0xe000) |
772 | if (lo < 0xdc00 || lo >= 0xe000) |
543 | ERR ("surrogate pair expected"); |
773 | ERR ("surrogate pair expected"); |
544 | |
774 | |
545 | hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; |
775 | hi = (hi - 0xD800) * 0x400 + (lo - 0xDC00) + 0x10000; |
|
|
776 | } |
|
|
777 | else if (hi < 0xe000) |
|
|
778 | ERR ("missing high surrogate character in surrogate pair"); |
|
|
779 | |
|
|
780 | if (hi >= 0x80) |
|
|
781 | { |
|
|
782 | utf8 = 1; |
|
|
783 | |
|
|
784 | cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); |
|
|
785 | } |
|
|
786 | else |
|
|
787 | *cur++ = hi; |
546 | } |
788 | } |
547 | else if (lo >= 0xdc00 && lo < 0xe000) |
|
|
548 | ERR ("illegal surrogate character"); |
|
|
549 | |
|
|
550 | if (hi >= 0x80) |
|
|
551 | { |
789 | break; |
552 | utf8 = 1; |
|
|
553 | |
790 | |
554 | APPEND_GROW (4); // at most 4 bytes for 21 bits |
|
|
555 | cur = (char *)uvuni_to_utf8_flags (cur, hi, 0); |
|
|
556 | } |
|
|
557 | else |
791 | default: |
558 | APPEND_CH (hi); |
792 | --dec_cur; |
|
|
793 | ERR ("illegal backslash escape sequence in string"); |
559 | } |
794 | } |
|
|
795 | } |
|
|
796 | else if (expect_true (ch >= 0x20 && ch <= 0x7f)) |
|
|
797 | *cur++ = ch; |
|
|
798 | else if (ch >= 0x80) |
|
|
799 | { |
|
|
800 | STRLEN clen; |
|
|
801 | UV uch; |
|
|
802 | |
|
|
803 | --dec_cur; |
|
|
804 | |
|
|
805 | uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
|
|
806 | if (clen == (STRLEN)-1) |
|
|
807 | ERR ("malformed UTF-8 character in JSON string"); |
|
|
808 | |
|
|
809 | do |
|
|
810 | *cur++ = *dec_cur++; |
|
|
811 | while (--clen); |
|
|
812 | |
|
|
813 | utf8 = 1; |
|
|
814 | } |
|
|
815 | else |
|
|
816 | { |
|
|
817 | --dec_cur; |
|
|
818 | |
|
|
819 | if (!ch) |
|
|
820 | ERR ("unexpected end of string while parsing JSON string"); |
560 | break; |
821 | else |
|
|
822 | ERR ("invalid character encountered while parsing JSON string"); |
561 | } |
823 | } |
562 | } |
824 | } |
563 | else if (ch >= 0x20 && ch <= 0x7f) |
825 | while (cur < buf + SHORT_STRING_LEN); |
564 | APPEND_CH (*dec->cur++); |
826 | |
565 | else if (ch >= 0x80) |
827 | { |
|
|
828 | STRLEN len = cur - buf; |
|
|
829 | |
|
|
830 | if (sv) |
566 | { |
831 | { |
567 | STRLEN clen; |
832 | SvGROW (sv, SvCUR (sv) + len + 1); |
568 | UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY); |
833 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
569 | if (clen < 0) |
834 | SvCUR_set (sv, SvCUR (sv) + len); |
570 | ERR ("malformed UTF-8 character in string, cannot convert to JSON"); |
|
|
571 | |
|
|
572 | APPEND_GROW (clen); |
|
|
573 | memcpy (cur, dec->cur, clen); |
|
|
574 | cur += clen; |
|
|
575 | dec->cur += clen; |
|
|
576 | } |
835 | } |
577 | else |
836 | else |
578 | ERR ("invalid character encountered"); |
837 | sv = newSVpvn (buf, len); |
579 | } |
838 | } |
|
|
839 | } |
|
|
840 | while (*dec_cur != '"'); |
580 | |
841 | |
581 | ++dec->cur; |
842 | ++dec_cur; |
582 | |
843 | |
583 | SvCUR_set (sv, cur - SvPVX (sv)); |
844 | if (sv) |
584 | |
845 | { |
585 | SvPOK_only (sv); |
846 | SvPOK_only (sv); |
586 | *SvEND (sv) = 0; |
847 | *SvEND (sv) = 0; |
587 | |
848 | |
588 | if (utf8) |
849 | if (utf8) |
589 | SvUTF8_on (sv); |
850 | SvUTF8_on (sv); |
|
|
851 | } |
|
|
852 | else |
|
|
853 | sv = newSVpvn ("", 0); |
590 | |
854 | |
|
|
855 | dec->cur = dec_cur; |
591 | return sv; |
856 | return sv; |
592 | |
857 | |
593 | fail: |
858 | fail: |
594 | SvREFCNT_dec (sv); |
859 | dec->cur = dec_cur; |
595 | return 0; |
860 | return 0; |
596 | } |
861 | } |
597 | |
862 | |
598 | static SV * |
863 | static SV * |
599 | decode_num (dec_t *dec) |
864 | decode_num (dec_t *dec) |
… | |
… | |
609 | { |
874 | { |
610 | ++dec->cur; |
875 | ++dec->cur; |
611 | if (*dec->cur >= '0' && *dec->cur <= '9') |
876 | if (*dec->cur >= '0' && *dec->cur <= '9') |
612 | ERR ("malformed number (leading zero must not be followed by another digit)"); |
877 | ERR ("malformed number (leading zero must not be followed by another digit)"); |
613 | } |
878 | } |
614 | |
879 | else if (*dec->cur < '0' || *dec->cur > '9') |
615 | // int |
880 | ERR ("malformed number (no digits after initial minus)"); |
|
|
881 | else |
|
|
882 | do |
|
|
883 | { |
|
|
884 | ++dec->cur; |
|
|
885 | } |
616 | while (*dec->cur >= '0' && *dec->cur <= '9') |
886 | while (*dec->cur >= '0' && *dec->cur <= '9'); |
617 | ++dec->cur; |
|
|
618 | |
887 | |
619 | // [frac] |
888 | // [frac] |
620 | if (*dec->cur == '.') |
889 | if (*dec->cur == '.') |
621 | { |
890 | { |
622 | is_nv = 1; |
891 | ++dec->cur; |
|
|
892 | |
|
|
893 | if (*dec->cur < '0' || *dec->cur > '9') |
|
|
894 | ERR ("malformed number (no digits after decimal point)"); |
623 | |
895 | |
624 | do |
896 | do |
625 | { |
897 | { |
626 | ++dec->cur; |
898 | ++dec->cur; |
627 | } |
899 | } |
628 | while (*dec->cur >= '0' && *dec->cur <= '9'); |
900 | while (*dec->cur >= '0' && *dec->cur <= '9'); |
|
|
901 | |
|
|
902 | is_nv = 1; |
629 | } |
903 | } |
630 | |
904 | |
631 | // [exp] |
905 | // [exp] |
632 | if (*dec->cur == 'e' || *dec->cur == 'E') |
906 | if (*dec->cur == 'e' || *dec->cur == 'E') |
633 | { |
907 | { |
634 | is_nv = 1; |
|
|
635 | |
|
|
636 | ++dec->cur; |
908 | ++dec->cur; |
|
|
909 | |
637 | if (*dec->cur == '-' || *dec->cur == '+') |
910 | if (*dec->cur == '-' || *dec->cur == '+') |
638 | ++dec->cur; |
911 | ++dec->cur; |
639 | |
912 | |
|
|
913 | if (*dec->cur < '0' || *dec->cur > '9') |
|
|
914 | ERR ("malformed number (no digits after exp sign)"); |
|
|
915 | |
|
|
916 | do |
|
|
917 | { |
|
|
918 | ++dec->cur; |
|
|
919 | } |
640 | while (*dec->cur >= '0' && *dec->cur <= '9') |
920 | while (*dec->cur >= '0' && *dec->cur <= '9'); |
641 | ++dec->cur; |
921 | |
|
|
922 | is_nv = 1; |
642 | } |
923 | } |
643 | |
924 | |
644 | if (!is_nv) |
925 | if (!is_nv) |
645 | { |
926 | { |
646 | UV uv; |
927 | // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so |
647 | int numtype = grok_number (start, dec->cur - start, &uv); |
928 | if (*start == '-') |
648 | if (numtype & IS_NUMBER_IN_UV) |
929 | switch (dec->cur - start) |
649 | if (numtype & IS_NUMBER_NEG) |
|
|
650 | { |
930 | { |
651 | if (uv < (UV)IV_MIN) |
931 | case 2: return newSViv (-( start [1] - '0' * 1)); |
652 | return newSViv (-(IV)uv); |
932 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
|
|
933 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
|
|
934 | case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
653 | } |
935 | } |
|
|
936 | else |
|
|
937 | switch (dec->cur - start) |
|
|
938 | { |
|
|
939 | case 1: return newSViv ( start [0] - '0' * 1); |
|
|
940 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
|
|
941 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
|
|
942 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
|
|
943 | } |
|
|
944 | |
|
|
945 | { |
|
|
946 | UV uv; |
|
|
947 | int numtype = grok_number (start, dec->cur - start, &uv); |
|
|
948 | if (numtype & IS_NUMBER_IN_UV) |
|
|
949 | if (numtype & IS_NUMBER_NEG) |
|
|
950 | { |
|
|
951 | if (uv < (UV)IV_MIN) |
|
|
952 | return newSViv (-(IV)uv); |
|
|
953 | } |
654 | else |
954 | else |
655 | return newSVuv (uv); |
955 | return newSVuv (uv); |
|
|
956 | |
|
|
957 | // here would likely be the place for bigint support |
656 | } |
958 | } |
|
|
959 | } |
657 | |
960 | |
|
|
961 | // if we ever support bigint or bigfloat, this is the place for bigfloat |
658 | return newSVnv (Atof (start)); |
962 | return newSVnv (Atof (start)); |
659 | |
963 | |
660 | fail: |
964 | fail: |
661 | return 0; |
965 | return 0; |
662 | } |
966 | } |
… | |
… | |
664 | static SV * |
968 | static SV * |
665 | decode_av (dec_t *dec) |
969 | decode_av (dec_t *dec) |
666 | { |
970 | { |
667 | AV *av = newAV (); |
971 | AV *av = newAV (); |
668 | |
972 | |
|
|
973 | DEC_INC_DEPTH; |
|
|
974 | decode_ws (dec); |
|
|
975 | |
|
|
976 | if (*dec->cur == ']') |
|
|
977 | ++dec->cur; |
|
|
978 | else |
669 | for (;;) |
979 | for (;;) |
670 | { |
980 | { |
671 | SV *value; |
981 | SV *value; |
672 | |
982 | |
673 | value = decode_sv (dec); |
983 | value = decode_sv (dec); |
674 | if (!value) |
984 | if (!value) |
675 | goto fail; |
985 | goto fail; |
676 | |
986 | |
677 | av_push (av, value); |
987 | av_push (av, value); |
678 | |
988 | |
679 | WS; |
989 | decode_ws (dec); |
680 | |
990 | |
681 | if (*dec->cur == ']') |
991 | if (*dec->cur == ']') |
682 | { |
992 | { |
683 | ++dec->cur; |
993 | ++dec->cur; |
684 | break; |
994 | break; |
|
|
995 | } |
685 | } |
996 | |
686 | |
|
|
687 | if (*dec->cur != ',') |
997 | if (*dec->cur != ',') |
688 | ERR (", or ] expected while parsing array"); |
998 | ERR (", or ] expected while parsing array"); |
689 | |
999 | |
690 | ++dec->cur; |
1000 | ++dec->cur; |
691 | } |
1001 | } |
692 | |
1002 | |
|
|
1003 | DEC_DEC_DEPTH; |
693 | return newRV_noinc ((SV *)av); |
1004 | return newRV_noinc ((SV *)av); |
694 | |
1005 | |
695 | fail: |
1006 | fail: |
696 | SvREFCNT_dec (av); |
1007 | SvREFCNT_dec (av); |
|
|
1008 | DEC_DEC_DEPTH; |
697 | return 0; |
1009 | return 0; |
698 | } |
1010 | } |
699 | |
1011 | |
700 | static SV * |
1012 | static SV * |
701 | decode_hv (dec_t *dec) |
1013 | decode_hv (dec_t *dec) |
702 | { |
1014 | { |
|
|
1015 | SV *sv; |
703 | HV *hv = newHV (); |
1016 | HV *hv = newHV (); |
704 | |
1017 | |
|
|
1018 | DEC_INC_DEPTH; |
|
|
1019 | decode_ws (dec); |
|
|
1020 | |
|
|
1021 | if (*dec->cur == '}') |
|
|
1022 | ++dec->cur; |
|
|
1023 | else |
705 | for (;;) |
1024 | for (;;) |
706 | { |
1025 | { |
707 | SV *key, *value; |
1026 | decode_ws (dec); EXPECT_CH ('"'); |
708 | |
1027 | |
709 | WS; EXPECT_CH ('"'); |
1028 | // heuristic: assume that |
710 | |
1029 | // a) decode_str + hv_store_ent are abysmally slow. |
711 | key = decode_str (dec); |
1030 | // b) most hash keys are short, simple ascii text. |
712 | if (!key) |
1031 | // => try to "fast-match" such strings to avoid |
713 | goto fail; |
1032 | // the overhead of decode_str + hv_store_ent. |
714 | |
|
|
715 | WS; EXPECT_CH (':'); |
|
|
716 | |
|
|
717 | value = decode_sv (dec); |
|
|
718 | if (!value) |
|
|
719 | { |
1033 | { |
|
|
1034 | SV *value; |
|
|
1035 | char *p = dec->cur; |
|
|
1036 | char *e = p + 24; // only try up to 24 bytes |
|
|
1037 | |
|
|
1038 | for (;;) |
|
|
1039 | { |
|
|
1040 | // the >= 0x80 is true on most architectures |
|
|
1041 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
|
|
1042 | { |
|
|
1043 | // slow path, back up and use decode_str |
|
|
1044 | SV *key = decode_str (dec); |
|
|
1045 | if (!key) |
|
|
1046 | goto fail; |
|
|
1047 | |
|
|
1048 | decode_ws (dec); EXPECT_CH (':'); |
|
|
1049 | |
|
|
1050 | value = decode_sv (dec); |
|
|
1051 | if (!value) |
|
|
1052 | { |
|
|
1053 | SvREFCNT_dec (key); |
|
|
1054 | goto fail; |
|
|
1055 | } |
|
|
1056 | |
|
|
1057 | hv_store_ent (hv, key, value, 0); |
720 | SvREFCNT_dec (key); |
1058 | SvREFCNT_dec (key); |
|
|
1059 | |
|
|
1060 | break; |
|
|
1061 | } |
|
|
1062 | else if (*p == '"') |
|
|
1063 | { |
|
|
1064 | // fast path, got a simple key |
|
|
1065 | char *key = dec->cur; |
|
|
1066 | int len = p - key; |
|
|
1067 | dec->cur = p + 1; |
|
|
1068 | |
|
|
1069 | decode_ws (dec); EXPECT_CH (':'); |
|
|
1070 | |
|
|
1071 | value = decode_sv (dec); |
|
|
1072 | if (!value) |
721 | goto fail; |
1073 | goto fail; |
|
|
1074 | |
|
|
1075 | hv_store (hv, key, len, value, 0); |
|
|
1076 | |
|
|
1077 | break; |
|
|
1078 | } |
|
|
1079 | |
|
|
1080 | ++p; |
|
|
1081 | } |
722 | } |
1082 | } |
723 | |
1083 | |
724 | //TODO: optimise |
1084 | decode_ws (dec); |
725 | hv_store_ent (hv, key, value, 0); |
|
|
726 | |
1085 | |
727 | WS; |
|
|
728 | |
|
|
729 | if (*dec->cur == '}') |
1086 | if (*dec->cur == '}') |
|
|
1087 | { |
|
|
1088 | ++dec->cur; |
|
|
1089 | break; |
|
|
1090 | } |
|
|
1091 | |
|
|
1092 | if (*dec->cur != ',') |
|
|
1093 | ERR (", or } expected while parsing object/hash"); |
|
|
1094 | |
|
|
1095 | ++dec->cur; |
|
|
1096 | } |
|
|
1097 | |
|
|
1098 | DEC_DEC_DEPTH; |
|
|
1099 | sv = newRV_noinc ((SV *)hv); |
|
|
1100 | |
|
|
1101 | // check filter callbacks |
|
|
1102 | if (dec->json.flags & F_HOOK) |
|
|
1103 | { |
|
|
1104 | if (dec->json.cb_sk_object && HvKEYS (hv) == 1) |
730 | { |
1105 | { |
731 | ++dec->cur; |
1106 | HE *cb, *he; |
732 | break; |
1107 | |
|
|
1108 | hv_iterinit (hv); |
|
|
1109 | he = hv_iternext (hv); |
|
|
1110 | hv_iterinit (hv); |
|
|
1111 | |
|
|
1112 | // the next line creates a mortal sv each time its called. |
|
|
1113 | // might want to optimise this for common cases. |
|
|
1114 | cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); |
|
|
1115 | |
|
|
1116 | if (cb) |
|
|
1117 | { |
|
|
1118 | int count; |
|
|
1119 | ENTER; SAVETMPS; |
|
|
1120 | |
|
|
1121 | dSP; PUSHMARK (SP); |
|
|
1122 | XPUSHs (HeVAL (he)); |
|
|
1123 | |
|
|
1124 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
|
|
1125 | |
|
|
1126 | if (count == 1) |
|
|
1127 | { |
|
|
1128 | sv = newSVsv (POPs); |
|
|
1129 | FREETMPS; LEAVE; |
|
|
1130 | return sv; |
|
|
1131 | } |
|
|
1132 | |
|
|
1133 | FREETMPS; LEAVE; |
|
|
1134 | } |
733 | } |
1135 | } |
734 | |
1136 | |
735 | if (*dec->cur != ',') |
1137 | if (dec->json.cb_object) |
736 | ERR (", or } expected while parsing object/hash"); |
1138 | { |
|
|
1139 | int count; |
|
|
1140 | ENTER; SAVETMPS; |
737 | |
1141 | |
738 | ++dec->cur; |
1142 | dSP; ENTER; SAVETMPS; PUSHMARK (SP); |
739 | } |
1143 | XPUSHs (sv_2mortal (sv)); |
740 | |
1144 | |
741 | return newRV_noinc ((SV *)hv); |
1145 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
|
|
1146 | |
|
|
1147 | if (count == 1) |
|
|
1148 | { |
|
|
1149 | sv = newSVsv (POPs); |
|
|
1150 | FREETMPS; LEAVE; |
|
|
1151 | return sv; |
|
|
1152 | } |
|
|
1153 | |
|
|
1154 | SvREFCNT_inc (sv); |
|
|
1155 | FREETMPS; LEAVE; |
|
|
1156 | } |
|
|
1157 | } |
|
|
1158 | |
|
|
1159 | return sv; |
742 | |
1160 | |
743 | fail: |
1161 | fail: |
744 | SvREFCNT_dec (hv); |
1162 | SvREFCNT_dec (hv); |
|
|
1163 | DEC_DEC_DEPTH; |
745 | return 0; |
1164 | return 0; |
746 | } |
1165 | } |
747 | |
1166 | |
748 | static SV * |
1167 | static SV * |
749 | decode_sv (dec_t *dec) |
1168 | decode_sv (dec_t *dec) |
750 | { |
1169 | { |
751 | WS; |
1170 | decode_ws (dec); |
|
|
1171 | |
|
|
1172 | // the beauty of JSON: you need exactly one character lookahead |
|
|
1173 | // to parse anything. |
752 | switch (*dec->cur) |
1174 | switch (*dec->cur) |
753 | { |
1175 | { |
754 | case '"': ++dec->cur; return decode_str (dec); |
1176 | case '"': ++dec->cur; return decode_str (dec); |
755 | case '[': ++dec->cur; return decode_av (dec); |
1177 | case '[': ++dec->cur; return decode_av (dec); |
756 | case '{': ++dec->cur; return decode_hv (dec); |
1178 | case '{': ++dec->cur; return decode_hv (dec); |
… | |
… | |
762 | |
1184 | |
763 | case 't': |
1185 | case 't': |
764 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1186 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
765 | { |
1187 | { |
766 | dec->cur += 4; |
1188 | dec->cur += 4; |
767 | return newSViv (1); |
1189 | return SvREFCNT_inc (json_true); |
768 | } |
1190 | } |
769 | else |
1191 | else |
770 | ERR ("'true' expected"); |
1192 | ERR ("'true' expected"); |
771 | |
1193 | |
772 | break; |
1194 | break; |
773 | |
1195 | |
774 | case 'f': |
1196 | case 'f': |
775 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1197 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
776 | { |
1198 | { |
777 | dec->cur += 5; |
1199 | dec->cur += 5; |
778 | return newSViv (0); |
1200 | return SvREFCNT_inc (json_false); |
779 | } |
1201 | } |
780 | else |
1202 | else |
781 | ERR ("'false' expected"); |
1203 | ERR ("'false' expected"); |
782 | |
1204 | |
783 | break; |
1205 | break; |
784 | |
1206 | |
785 | case 'n': |
1207 | case 'n': |
786 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
1208 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
787 | { |
1209 | { |
788 | dec->cur += 4; |
1210 | dec->cur += 4; |
789 | return newSViv (1); |
1211 | return newSVsv (&PL_sv_undef); |
790 | } |
1212 | } |
791 | else |
1213 | else |
792 | ERR ("'null' expected"); |
1214 | ERR ("'null' expected"); |
793 | |
1215 | |
794 | break; |
1216 | break; |
795 | |
1217 | |
796 | default: |
1218 | default: |
797 | ERR ("malformed json string"); |
1219 | ERR ("malformed JSON string, neither array, object, number, string or atom"); |
798 | break; |
1220 | break; |
799 | } |
1221 | } |
800 | |
1222 | |
801 | fail: |
1223 | fail: |
802 | return 0; |
1224 | return 0; |
803 | } |
1225 | } |
804 | |
1226 | |
805 | static SV * |
1227 | static SV * |
806 | decode_json (SV *string, UV flags) |
1228 | decode_json (SV *string, JSON *json, UV *offset_return) |
807 | { |
1229 | { |
|
|
1230 | dec_t dec; |
|
|
1231 | UV offset; |
808 | SV *sv; |
1232 | SV *sv; |
809 | |
1233 | |
|
|
1234 | SvGETMAGIC (string); |
|
|
1235 | SvUPGRADE (string, SVt_PV); |
|
|
1236 | |
|
|
1237 | if (json->flags & F_MAXSIZE && SvCUR (string) > DEC_SIZE (json->flags)) |
|
|
1238 | croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", |
|
|
1239 | (unsigned long)SvCUR (string), (unsigned long)DEC_SIZE (json->flags)); |
|
|
1240 | |
810 | if (!(flags & F_UTF8)) |
1241 | if (json->flags & F_UTF8) |
|
|
1242 | sv_utf8_downgrade (string, 0); |
|
|
1243 | else |
811 | sv_utf8_upgrade (string); |
1244 | sv_utf8_upgrade (string); |
812 | |
1245 | |
813 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1246 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
814 | |
1247 | |
815 | dec_t dec; |
1248 | dec.json = *json; |
816 | dec.flags = flags; |
|
|
817 | dec.cur = SvPVX (string); |
1249 | dec.cur = SvPVX (string); |
818 | dec.end = SvEND (string); |
1250 | dec.end = SvEND (string); |
819 | dec.err = 0; |
1251 | dec.err = 0; |
|
|
1252 | dec.depth = 0; |
|
|
1253 | dec.maxdepth = DEC_DEPTH (dec.json.flags); |
820 | |
1254 | |
821 | *dec.end = 1; // invalid anywhere |
1255 | if (dec.json.cb_object || dec.json.cb_sk_object) |
|
|
1256 | dec.json.flags |= F_HOOK; |
|
|
1257 | |
|
|
1258 | *dec.end = 0; // this should basically be a nop, too, but make sure it's there |
822 | sv = decode_sv (&dec); |
1259 | sv = decode_sv (&dec); |
823 | *dec.end = 0; |
1260 | |
|
|
1261 | if (!(offset_return || !sv)) |
|
|
1262 | { |
|
|
1263 | // check for trailing garbage |
|
|
1264 | decode_ws (&dec); |
|
|
1265 | |
|
|
1266 | if (*dec.cur) |
|
|
1267 | { |
|
|
1268 | dec.err = "garbage after JSON object"; |
|
|
1269 | SvREFCNT_dec (sv); |
|
|
1270 | sv = 0; |
|
|
1271 | } |
|
|
1272 | } |
|
|
1273 | |
|
|
1274 | if (offset_return || !sv) |
|
|
1275 | { |
|
|
1276 | offset = dec.json.flags & F_UTF8 |
|
|
1277 | ? dec.cur - SvPVX (string) |
|
|
1278 | : utf8_distance (dec.cur, SvPVX (string)); |
|
|
1279 | |
|
|
1280 | if (offset_return) |
|
|
1281 | *offset_return = offset; |
|
|
1282 | } |
824 | |
1283 | |
825 | if (!sv) |
1284 | if (!sv) |
826 | { |
1285 | { |
827 | IV offset = utf8_distance (dec.cur, SvPVX (string)); |
|
|
828 | SV *uni = sv_newmortal (); |
1286 | SV *uni = sv_newmortal (); |
829 | |
1287 | |
|
|
1288 | // horrible hack to silence warning inside pv_uni_display |
|
|
1289 | COP cop = *PL_curcop; |
|
|
1290 | cop.cop_warnings = pWARN_NONE; |
|
|
1291 | ENTER; |
|
|
1292 | SAVEVPTR (PL_curcop); |
|
|
1293 | PL_curcop = &cop; |
830 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
1294 | pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); |
|
|
1295 | LEAVE; |
|
|
1296 | |
831 | croak ("%s, at character %d (%s)", |
1297 | croak ("%s, at character offset %d [\"%s\"]", |
832 | dec.err, |
1298 | dec.err, |
833 | (int)offset, |
1299 | (int)offset, |
834 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1300 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
835 | } |
1301 | } |
836 | |
1302 | |
837 | sv = sv_2mortal (sv); |
1303 | sv = sv_2mortal (sv); |
838 | |
1304 | |
839 | if (!(dec.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1305 | if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
840 | croak ("JSON object or array expected (but number, string, true, false or null found, use allow_nonref to allow this)"); |
1306 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
841 | |
1307 | |
842 | return sv; |
1308 | return sv; |
843 | } |
1309 | } |
844 | |
1310 | |
|
|
1311 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1312 | // XS interface functions |
|
|
1313 | |
845 | MODULE = JSON::XS PACKAGE = JSON::XS |
1314 | MODULE = JSON::XS PACKAGE = JSON::XS |
846 | |
1315 | |
847 | BOOT: |
1316 | BOOT: |
848 | { |
1317 | { |
849 | int i; |
1318 | int i; |
850 | |
1319 | |
851 | memset (decode_hexdigit, 0xff, 256); |
|
|
852 | for (i = 10; i--; ) |
1320 | for (i = 0; i < 256; ++i) |
853 | decode_hexdigit ['0' + i] = i; |
1321 | decode_hexdigit [i] = |
|
|
1322 | i >= '0' && i <= '9' ? i - '0' |
|
|
1323 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
|
|
1324 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
|
|
1325 | : -1; |
854 | |
1326 | |
855 | for (i = 7; i--; ) |
1327 | json_stash = gv_stashpv ("JSON::XS" , 1); |
|
|
1328 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
|
|
1329 | |
|
|
1330 | json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); |
|
|
1331 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
|
|
1332 | } |
|
|
1333 | |
|
|
1334 | PROTOTYPES: DISABLE |
|
|
1335 | |
|
|
1336 | void new (char *klass) |
|
|
1337 | PPCODE: |
|
|
1338 | { |
|
|
1339 | SV *pv = NEWSV (0, sizeof (JSON)); |
|
|
1340 | SvPOK_only (pv); |
|
|
1341 | Zero (SvPVX (pv), 1, JSON); |
|
|
1342 | ((JSON *)SvPVX (pv))->flags = F_DEFAULT; |
|
|
1343 | XPUSHs (sv_2mortal (sv_bless (newRV_noinc (pv), json_stash))); |
|
|
1344 | } |
|
|
1345 | |
|
|
1346 | void ascii (JSON *self, int enable = 1) |
|
|
1347 | ALIAS: |
|
|
1348 | ascii = F_ASCII |
|
|
1349 | latin1 = F_LATIN1 |
|
|
1350 | utf8 = F_UTF8 |
|
|
1351 | indent = F_INDENT |
|
|
1352 | canonical = F_CANONICAL |
|
|
1353 | space_before = F_SPACE_BEFORE |
|
|
1354 | space_after = F_SPACE_AFTER |
|
|
1355 | pretty = F_PRETTY |
|
|
1356 | allow_nonref = F_ALLOW_NONREF |
|
|
1357 | shrink = F_SHRINK |
|
|
1358 | allow_blessed = F_ALLOW_BLESSED |
|
|
1359 | convert_blessed = F_CONV_BLESSED |
|
|
1360 | PPCODE: |
|
|
1361 | { |
|
|
1362 | if (enable) |
|
|
1363 | self->flags |= ix; |
|
|
1364 | else |
|
|
1365 | self->flags &= ~ix; |
|
|
1366 | |
|
|
1367 | XPUSHs (ST (0)); |
|
|
1368 | } |
|
|
1369 | |
|
|
1370 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
|
|
1371 | PPCODE: |
|
|
1372 | { |
|
|
1373 | UV log2 = 0; |
|
|
1374 | |
|
|
1375 | if (max_depth > 0x80000000UL) max_depth = 0x80000000UL; |
|
|
1376 | |
|
|
1377 | while ((1UL << log2) < max_depth) |
|
|
1378 | ++log2; |
|
|
1379 | |
|
|
1380 | self->flags = self->flags & ~F_MAXDEPTH | (log2 << S_MAXDEPTH); |
|
|
1381 | |
|
|
1382 | XPUSHs (ST (0)); |
|
|
1383 | } |
|
|
1384 | |
|
|
1385 | void max_size (JSON *self, UV max_size = 0) |
|
|
1386 | PPCODE: |
|
|
1387 | { |
|
|
1388 | UV log2 = 0; |
|
|
1389 | |
|
|
1390 | if (max_size > 0x80000000UL) max_size = 0x80000000UL; |
|
|
1391 | if (max_size == 1) max_size = 2; |
|
|
1392 | |
|
|
1393 | while ((1UL << log2) < max_size) |
|
|
1394 | ++log2; |
|
|
1395 | |
|
|
1396 | self->flags = self->flags & ~F_MAXSIZE | (log2 << S_MAXSIZE); |
|
|
1397 | |
|
|
1398 | XPUSHs (ST (0)); |
|
|
1399 | } |
|
|
1400 | |
|
|
1401 | void filter_json_object (JSON *self, SV *cb = &PL_sv_undef) |
|
|
1402 | PPCODE: |
|
|
1403 | { |
|
|
1404 | SvREFCNT_dec (self->cb_object); |
|
|
1405 | self->cb_object = SvOK (cb) ? newSVsv (cb) : 0; |
|
|
1406 | |
|
|
1407 | XPUSHs (ST (0)); |
|
|
1408 | } |
|
|
1409 | |
|
|
1410 | void filter_json_single_key_object (JSON *self, SV *key, SV *cb = &PL_sv_undef) |
|
|
1411 | PPCODE: |
|
|
1412 | { |
|
|
1413 | if (!self->cb_sk_object) |
|
|
1414 | self->cb_sk_object = newHV (); |
|
|
1415 | |
|
|
1416 | if (SvOK (cb)) |
|
|
1417 | hv_store_ent (self->cb_sk_object, key, newSVsv (cb), 0); |
|
|
1418 | else |
856 | { |
1419 | { |
857 | decode_hexdigit ['a' + i] = 10 + i; |
1420 | hv_delete_ent (self->cb_sk_object, key, G_DISCARD, 0); |
858 | decode_hexdigit ['A' + i] = 10 + i; |
1421 | |
|
|
1422 | if (!HvKEYS (self->cb_sk_object)) |
|
|
1423 | { |
|
|
1424 | SvREFCNT_dec (self->cb_sk_object); |
|
|
1425 | self->cb_sk_object = 0; |
|
|
1426 | } |
859 | } |
1427 | } |
860 | |
1428 | |
861 | json_stash = gv_stashpv ("JSON::XS", 1); |
1429 | XPUSHs (ST (0)); |
862 | } |
1430 | } |
863 | |
1431 | |
864 | PROTOTYPES: DISABLE |
1432 | void encode (JSON *self, SV *scalar) |
|
|
1433 | PPCODE: |
|
|
1434 | XPUSHs (encode_json (scalar, self)); |
865 | |
1435 | |
866 | SV *new (char *dummy) |
1436 | void decode (JSON *self, SV *jsonstr) |
|
|
1437 | PPCODE: |
|
|
1438 | XPUSHs (decode_json (jsonstr, self, 0)); |
|
|
1439 | |
|
|
1440 | void decode_prefix (JSON *self, SV *jsonstr) |
|
|
1441 | PPCODE: |
|
|
1442 | { |
|
|
1443 | UV offset; |
|
|
1444 | EXTEND (SP, 2); |
|
|
1445 | PUSHs (decode_json (jsonstr, self, &offset)); |
|
|
1446 | PUSHs (sv_2mortal (newSVuv (offset))); |
|
|
1447 | } |
|
|
1448 | |
|
|
1449 | void DESTROY (JSON *self) |
867 | CODE: |
1450 | CODE: |
868 | RETVAL = sv_bless (newRV_noinc (newSVuv (F_DEFAULT)), json_stash); |
1451 | SvREFCNT_dec (self->cb_sk_object); |
869 | OUTPUT: |
1452 | SvREFCNT_dec (self->cb_object); |
870 | RETVAL |
|
|
871 | |
|
|
872 | SV *ascii (SV *self, int enable) |
|
|
873 | ALIAS: |
|
|
874 | ascii = F_ASCII |
|
|
875 | utf8 = F_UTF8 |
|
|
876 | indent = F_INDENT |
|
|
877 | canonical = F_CANONICAL |
|
|
878 | space_before = F_SPACE_BEFORE |
|
|
879 | space_after = F_SPACE_AFTER |
|
|
880 | json_rpc = F_JSON_RPC |
|
|
881 | pretty = F_PRETTY |
|
|
882 | allow_nonref = F_ALLOW_NONREF |
|
|
883 | CODE: |
|
|
884 | { |
|
|
885 | UV *uv = SvJSON (self); |
|
|
886 | if (enable) |
|
|
887 | *uv |= ix; |
|
|
888 | else |
|
|
889 | *uv &= ~ix; |
|
|
890 | |
|
|
891 | RETVAL = newSVsv (self); |
|
|
892 | } |
|
|
893 | OUTPUT: |
|
|
894 | RETVAL |
|
|
895 | |
|
|
896 | void encode (SV *self, SV *scalar) |
|
|
897 | PPCODE: |
|
|
898 | XPUSHs (encode_json (scalar, *SvJSON (self))); |
|
|
899 | |
|
|
900 | void decode (SV *self, SV *jsonstr) |
|
|
901 | PPCODE: |
|
|
902 | XPUSHs (decode_json (jsonstr, *SvJSON (self))); |
|
|
903 | |
1453 | |
904 | PROTOTYPES: ENABLE |
1454 | PROTOTYPES: ENABLE |
905 | |
1455 | |
906 | void to_json (SV *scalar) |
1456 | void to_json (SV *scalar) |
907 | PPCODE: |
1457 | PPCODE: |
|
|
1458 | { |
|
|
1459 | JSON json = { F_DEFAULT | F_UTF8 }; |
908 | XPUSHs (encode_json (scalar, F_UTF8)); |
1460 | XPUSHs (encode_json (scalar, &json)); |
|
|
1461 | } |
909 | |
1462 | |
910 | void from_json (SV *jsonstr) |
1463 | void from_json (SV *jsonstr) |
911 | PPCODE: |
1464 | PPCODE: |
|
|
1465 | { |
|
|
1466 | JSON json = { F_DEFAULT | F_UTF8 }; |
912 | XPUSHs (decode_json (jsonstr, F_UTF8)); |
1467 | XPUSHs (decode_json (jsonstr, &json, 0)); |
|
|
1468 | } |
913 | |
1469 | |