… | |
… | |
6 | #include <string.h> |
6 | #include <string.h> |
7 | #include <stdlib.h> |
7 | #include <stdlib.h> |
8 | #include <stdio.h> |
8 | #include <stdio.h> |
9 | #include <limits.h> |
9 | #include <limits.h> |
10 | #include <float.h> |
10 | #include <float.h> |
|
|
11 | #include <inttypes.h> |
11 | |
12 | |
12 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
13 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
13 | # define snprintf _snprintf // C compilers have this in stdio.h |
14 | # define snprintf _snprintf // C compilers have this in stdio.h |
14 | #endif |
15 | #endif |
15 | |
16 | |
16 | // some old perls do not have this, try to make it work, no |
17 | // some old perls do not have this, try to make it work, no |
17 | // guarantees, though. if it breaks, you get to keep the pieces. |
18 | // guarantees, though. if it breaks, you get to keep the pieces. |
18 | #ifndef UTF8_MAXBYTES |
19 | #ifndef UTF8_MAXBYTES |
19 | # define UTF8_MAXBYTES 13 |
20 | # define UTF8_MAXBYTES 13 |
|
|
21 | #endif |
|
|
22 | |
|
|
23 | // compatibility with perl <5.18 |
|
|
24 | #ifndef HvNAMELEN_get |
|
|
25 | # define HvNAMELEN_get(hv) strlen (HvNAME (hv)) |
|
|
26 | #endif |
|
|
27 | #ifndef HvNAMELEN |
|
|
28 | # define HvNAMELEN(hv) HvNAMELEN_get (hv) |
|
|
29 | #endif |
|
|
30 | #ifndef HvNAMEUTF8 |
|
|
31 | # define HvNAMEUTF8(hv) 0 |
20 | #endif |
32 | #endif |
21 | |
33 | |
22 | // three extra for rounding, sign, and end of string |
34 | // three extra for rounding, sign, and end of string |
23 | #define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3) |
35 | #define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3) |
24 | |
36 | |
… | |
… | |
33 | #define F_SHRINK 0x00000200UL |
45 | #define F_SHRINK 0x00000200UL |
34 | #define F_ALLOW_BLESSED 0x00000400UL |
46 | #define F_ALLOW_BLESSED 0x00000400UL |
35 | #define F_CONV_BLESSED 0x00000800UL |
47 | #define F_CONV_BLESSED 0x00000800UL |
36 | #define F_RELAXED 0x00001000UL |
48 | #define F_RELAXED 0x00001000UL |
37 | #define F_ALLOW_UNKNOWN 0x00002000UL |
49 | #define F_ALLOW_UNKNOWN 0x00002000UL |
|
|
50 | #define F_ALLOW_TAGS 0x00004000UL |
38 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
51 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
39 | |
52 | |
40 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
53 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
41 | |
54 | |
42 | #define INIT_SIZE 32 // initial scalar size to be allocated |
55 | #define INIT_SIZE 64 // initial scalar size to be allocated |
43 | #define INDENT_STEP 3 // spaces per indentation level |
56 | #define INDENT_STEP 3 // spaces per indentation level |
44 | |
57 | |
45 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
58 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
46 | |
59 | |
47 | #define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8) |
60 | #define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8) |
… | |
… | |
65 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
78 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
66 | |
79 | |
67 | #define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)" |
80 | #define ERR_NESTING_EXCEEDED "json text or perl structure exceeds maximum nesting level (max_depth set too low?)" |
68 | |
81 | |
69 | #ifdef USE_ITHREADS |
82 | #ifdef USE_ITHREADS |
70 | # define JSON_SLOW 1 |
|
|
71 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
83 | # define JSON_STASH (expect_true (json_stash) ? json_stash : gv_stashpv ("JSON::XS", 1)) |
|
|
84 | # define BOOL_STASH (expect_true (bool_stash) ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1)) |
|
|
85 | # define GET_BOOL(value) (expect_true (bool_ ## value) ? bool_ ## value : get_bool ("Types::Serialiser::" # value)) |
72 | #else |
86 | #else |
73 | # define JSON_SLOW 0 |
|
|
74 | # define JSON_STASH json_stash |
87 | # define JSON_STASH json_stash |
|
|
88 | # define BOOL_STASH bool_stash |
|
|
89 | # define GET_BOOL(value) bool_ ## value |
75 | #endif |
90 | #endif |
76 | |
91 | |
77 | // the amount of HEs to allocate on the stack, when sorting keys |
92 | // the amount of HEs to allocate on the stack, when sorting keys |
78 | #define STACK_HES 64 |
93 | #define STACK_HES 64 |
79 | |
94 | |
80 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
95 | static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean:: |
81 | static SV *json_true, *json_false; |
96 | static SV *bool_false, *bool_true; |
|
|
97 | static SV *sv_json; |
82 | |
98 | |
83 | enum { |
99 | enum { |
84 | INCR_M_WS = 0, // initial whitespace skipping, must be 0 |
100 | INCR_M_WS = 0, // initial whitespace skipping, must be 0 |
|
|
101 | INCR_M_TFN, // inside true/false/null |
|
|
102 | INCR_M_NUM, // inside number |
85 | INCR_M_STR, // inside string |
103 | INCR_M_STR, // inside string |
86 | INCR_M_BS, // inside backslash |
104 | INCR_M_BS, // inside backslash |
87 | INCR_M_C0, // inside comment in initial whitespace sequence |
105 | INCR_M_C0, // inside comment in initial whitespace sequence |
88 | INCR_M_C1, // inside comment in other places |
106 | INCR_M_C1, // inside comment in other places |
89 | INCR_M_JSON // outside anything, count nesting |
107 | INCR_M_JSON // outside anything, count nesting |
… | |
… | |
102 | // for the incremental parser |
120 | // for the incremental parser |
103 | SV *incr_text; // the source text so far |
121 | SV *incr_text; // the source text so far |
104 | STRLEN incr_pos; // the current offset into the text |
122 | STRLEN incr_pos; // the current offset into the text |
105 | int incr_nest; // {[]}-nesting level |
123 | int incr_nest; // {[]}-nesting level |
106 | unsigned char incr_mode; |
124 | unsigned char incr_mode; |
|
|
125 | |
|
|
126 | SV *v_false, *v_true; |
107 | } JSON; |
127 | } JSON; |
108 | |
128 | |
109 | INLINE void |
129 | INLINE void |
110 | json_init (JSON *json) |
130 | json_init (JSON *json) |
111 | { |
131 | { |
112 | Zero (json, 1, JSON); |
132 | static const JSON init = { F_ALLOW_NONREF, 512 }; |
113 | json->max_depth = 512; |
133 | |
|
|
134 | *json = init; |
114 | } |
135 | } |
115 | |
136 | |
116 | ///////////////////////////////////////////////////////////////////////////// |
137 | ///////////////////////////////////////////////////////////////////////////// |
117 | // utility functions |
138 | // utility functions |
118 | |
139 | |
… | |
… | |
140 | SvPV_renew (sv, SvCUR (sv) + 1); |
161 | SvPV_renew (sv, SvCUR (sv) + 1); |
141 | #endif |
162 | #endif |
142 | } |
163 | } |
143 | } |
164 | } |
144 | |
165 | |
|
|
166 | /* adds two STRLENs together, slow, and with paranoia */ |
|
|
167 | STRLEN |
|
|
168 | strlen_sum (STRLEN l1, STRLEN l2) |
|
|
169 | { |
|
|
170 | size_t sum = l1 + l2; |
|
|
171 | |
|
|
172 | if (sum < (size_t)l2 || sum != (size_t)(STRLEN)sum) |
|
|
173 | croak ("JSON::XS: string size overflow"); |
|
|
174 | |
|
|
175 | return sum; |
|
|
176 | } |
|
|
177 | |
|
|
178 | /* similar to SvGROW, but somewhat safer and guarantees exponential realloc strategy */ |
|
|
179 | static char * |
|
|
180 | json_sv_grow (SV *sv, size_t len1, size_t len2) |
|
|
181 | { |
|
|
182 | len1 = strlen_sum (len1, len2); |
|
|
183 | len1 = strlen_sum (len1, len1 >> 1); |
|
|
184 | |
|
|
185 | if (len1 > 4096 - 24) |
|
|
186 | len1 = (len1 | 4095) - 24; |
|
|
187 | |
|
|
188 | return SvGROW (sv, len1); |
|
|
189 | } |
|
|
190 | |
145 | // decode an utf-8 character and return it, or (UV)-1 in |
191 | // decode a utf-8 character and return it, or (UV)-1 in |
146 | // case of an error. |
192 | // case of an error. |
147 | // we special-case "safe" characters from U+80 .. U+7FF, |
193 | // we special-case "safe" characters from U+80 .. U+7FF, |
148 | // but use the very good perl function to parse anything else. |
194 | // but use the very good perl function to parse anything else. |
149 | // note that we never call this function for a ascii codepoints |
195 | // note that we never call this function for a ascii codepoints |
150 | INLINE UV |
196 | INLINE UV |
… | |
… | |
287 | // a recursion depth of ten gives us >>500 bits |
333 | // a recursion depth of ten gives us >>500 bits |
288 | json_atof_scan1 (s, &accum, &expo, 0, 10); |
334 | json_atof_scan1 (s, &accum, &expo, 0, 10); |
289 | |
335 | |
290 | return neg ? -accum : accum; |
336 | return neg ? -accum : accum; |
291 | } |
337 | } |
|
|
338 | |
|
|
339 | // target of scalar reference is bool? -1 == nope, 0 == false, 1 == true |
|
|
340 | static int |
|
|
341 | ref_bool_type (SV *sv) |
|
|
342 | { |
|
|
343 | svtype svt = SvTYPE (sv); |
|
|
344 | |
|
|
345 | if (svt < SVt_PVAV) |
|
|
346 | { |
|
|
347 | STRLEN len = 0; |
|
|
348 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
349 | |
|
|
350 | if (len == 1) |
|
|
351 | if (*pv == '1') |
|
|
352 | return 1; |
|
|
353 | else if (*pv == '0') |
|
|
354 | return 0; |
|
|
355 | } |
|
|
356 | |
|
|
357 | return -1; |
|
|
358 | } |
|
|
359 | |
|
|
360 | // returns whether scalar is not a reference in the sense of allow_nonref |
|
|
361 | static int |
|
|
362 | json_nonref (SV *scalar) |
|
|
363 | { |
|
|
364 | if (!SvROK (scalar)) |
|
|
365 | return 1; |
|
|
366 | |
|
|
367 | scalar = SvRV (scalar); |
|
|
368 | |
|
|
369 | if (SvTYPE (scalar) >= SVt_PVMG) |
|
|
370 | { |
|
|
371 | if (SvSTASH (scalar) == bool_stash) |
|
|
372 | return 1; |
|
|
373 | |
|
|
374 | if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0) |
|
|
375 | return 1; |
|
|
376 | } |
|
|
377 | |
|
|
378 | return 0; |
|
|
379 | } |
|
|
380 | |
292 | ///////////////////////////////////////////////////////////////////////////// |
381 | ///////////////////////////////////////////////////////////////////////////// |
293 | // encoder |
382 | // encoder |
294 | |
383 | |
295 | // structure used for encoding JSON |
384 | // structure used for encoding JSON |
296 | typedef struct |
385 | typedef struct |
… | |
… | |
304 | } enc_t; |
393 | } enc_t; |
305 | |
394 | |
306 | INLINE void |
395 | INLINE void |
307 | need (enc_t *enc, STRLEN len) |
396 | need (enc_t *enc, STRLEN len) |
308 | { |
397 | { |
309 | if (expect_false (enc->cur + len >= enc->end)) |
398 | if (expect_false ((uintptr_t)(enc->end - enc->cur) < len)) |
310 | { |
399 | { |
311 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
400 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
312 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
401 | char *buf = json_sv_grow (enc->sv, cur, len); |
313 | enc->cur = SvPVX (enc->sv) + cur; |
402 | enc->cur = buf + cur; |
314 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
403 | enc->end = buf + SvLEN (enc->sv) - 1; |
315 | } |
404 | } |
316 | } |
405 | } |
317 | |
406 | |
318 | INLINE void |
407 | INLINE void |
319 | encode_ch (enc_t *enc, char ch) |
408 | encode_ch (enc_t *enc, char ch) |
… | |
… | |
335 | |
424 | |
336 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
425 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
337 | { |
426 | { |
338 | if (expect_false (ch == '"')) // but with slow exceptions |
427 | if (expect_false (ch == '"')) // but with slow exceptions |
339 | { |
428 | { |
340 | need (enc, len += 1); |
429 | need (enc, len + 1); |
341 | *enc->cur++ = '\\'; |
430 | *enc->cur++ = '\\'; |
342 | *enc->cur++ = '"'; |
431 | *enc->cur++ = '"'; |
343 | } |
432 | } |
344 | else if (expect_false (ch == '\\')) |
433 | else if (expect_false (ch == '\\')) |
345 | { |
434 | { |
346 | need (enc, len += 1); |
435 | need (enc, len + 1); |
347 | *enc->cur++ = '\\'; |
436 | *enc->cur++ = '\\'; |
348 | *enc->cur++ = '\\'; |
437 | *enc->cur++ = '\\'; |
349 | } |
438 | } |
350 | else |
439 | else |
351 | *enc->cur++ = ch; |
440 | *enc->cur++ = ch; |
… | |
… | |
354 | } |
443 | } |
355 | else |
444 | else |
356 | { |
445 | { |
357 | switch (ch) |
446 | switch (ch) |
358 | { |
447 | { |
359 | case '\010': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; |
448 | case '\010': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'b'; ++str; break; |
360 | case '\011': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; |
449 | case '\011': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 't'; ++str; break; |
361 | case '\012': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; |
450 | case '\012': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'n'; ++str; break; |
362 | case '\014': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; |
451 | case '\014': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'f'; ++str; break; |
363 | case '\015': need (enc, len += 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; |
452 | case '\015': need (enc, len + 1); *enc->cur++ = '\\'; *enc->cur++ = 'r'; ++str; break; |
364 | |
453 | |
365 | default: |
454 | default: |
366 | { |
455 | { |
367 | STRLEN clen; |
456 | STRLEN clen; |
368 | UV uch; |
457 | UV uch; |
… | |
… | |
384 | if (uch >= 0x10000UL) |
473 | if (uch >= 0x10000UL) |
385 | { |
474 | { |
386 | if (uch >= 0x110000UL) |
475 | if (uch >= 0x110000UL) |
387 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
476 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
388 | |
477 | |
389 | need (enc, len += 11); |
478 | need (enc, len + 11); |
390 | sprintf (enc->cur, "\\u%04x\\u%04x", |
479 | sprintf (enc->cur, "\\u%04x\\u%04x", |
391 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
480 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
392 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
481 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
393 | enc->cur += 12; |
482 | enc->cur += 12; |
394 | } |
483 | } |
395 | else |
484 | else |
396 | { |
485 | { |
397 | need (enc, len += 5); |
486 | need (enc, len + 5); |
398 | *enc->cur++ = '\\'; |
487 | *enc->cur++ = '\\'; |
399 | *enc->cur++ = 'u'; |
488 | *enc->cur++ = 'u'; |
400 | *enc->cur++ = PL_hexdigit [ uch >> 12 ]; |
489 | *enc->cur++ = PL_hexdigit [ uch >> 12 ]; |
401 | *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; |
490 | *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; |
402 | *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; |
491 | *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; |
… | |
… | |
410 | *enc->cur++ = uch; |
499 | *enc->cur++ = uch; |
411 | str += clen; |
500 | str += clen; |
412 | } |
501 | } |
413 | else if (is_utf8) |
502 | else if (is_utf8) |
414 | { |
503 | { |
415 | need (enc, len += clen); |
504 | need (enc, len + clen); |
416 | do |
505 | do |
417 | { |
506 | { |
418 | *enc->cur++ = *str++; |
507 | *enc->cur++ = *str++; |
419 | } |
508 | } |
420 | while (--clen); |
509 | while (--clen); |
421 | } |
510 | } |
422 | else |
511 | else |
423 | { |
512 | { |
424 | need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
513 | need (enc, len + UTF8_MAXBYTES - 1); // never more than 11 bytes needed |
425 | enc->cur = encode_utf8 (enc->cur, uch); |
514 | enc->cur = encode_utf8 (enc->cur, uch); |
426 | ++str; |
515 | ++str; |
427 | } |
516 | } |
428 | } |
517 | } |
429 | } |
518 | } |
… | |
… | |
682 | // encode objects, arrays and special \0=false and \1=true values. |
771 | // encode objects, arrays and special \0=false and \1=true values. |
683 | static void |
772 | static void |
684 | encode_rv (enc_t *enc, SV *sv) |
773 | encode_rv (enc_t *enc, SV *sv) |
685 | { |
774 | { |
686 | svtype svt; |
775 | svtype svt; |
|
|
776 | GV *method; |
687 | |
777 | |
688 | SvGETMAGIC (sv); |
778 | SvGETMAGIC (sv); |
689 | svt = SvTYPE (sv); |
779 | svt = SvTYPE (sv); |
690 | |
780 | |
691 | if (expect_false (SvOBJECT (sv))) |
781 | if (expect_false (SvOBJECT (sv))) |
692 | { |
782 | { |
693 | HV *stash = !JSON_SLOW || json_boolean_stash |
783 | HV *stash = SvSTASH (sv); |
694 | ? json_boolean_stash |
|
|
695 | : gv_stashpv ("JSON::XS::Boolean", 1); |
|
|
696 | |
784 | |
697 | if (SvSTASH (sv) == stash) |
785 | if (stash == bool_stash) |
698 | { |
786 | { |
699 | if (SvIV (sv)) |
|
|
700 | encode_str (enc, "true", 4, 0); |
787 | if (SvIV (sv)) encode_str (enc, "true" , 4, 0); |
701 | else |
|
|
702 | encode_str (enc, "false", 5, 0); |
788 | else encode_str (enc, "false", 5, 0); |
703 | } |
789 | } |
|
|
790 | else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0))) |
|
|
791 | { |
|
|
792 | int count; |
|
|
793 | dSP; |
|
|
794 | |
|
|
795 | ENTER; SAVETMPS; |
|
|
796 | PUSHMARK (SP); |
|
|
797 | EXTEND (SP, 2); |
|
|
798 | // we re-bless the reference to get overload and other niceties right |
|
|
799 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
|
|
800 | PUSHs (sv_json); |
|
|
801 | |
|
|
802 | PUTBACK; |
|
|
803 | count = call_sv ((SV *)GvCV (method), G_ARRAY); |
|
|
804 | SPAGAIN; |
|
|
805 | |
|
|
806 | // catch this surprisingly common error |
|
|
807 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
808 | croak ("%s::FREEZE method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
809 | |
|
|
810 | encode_ch (enc, '('); |
|
|
811 | encode_ch (enc, '"'); |
|
|
812 | encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash)); |
|
|
813 | encode_ch (enc, '"'); |
|
|
814 | encode_ch (enc, ')'); |
|
|
815 | encode_ch (enc, '['); |
|
|
816 | |
|
|
817 | if (count) |
|
|
818 | { |
|
|
819 | int i; |
|
|
820 | |
|
|
821 | for (i = 0; i < count - 1; ++i) |
|
|
822 | { |
|
|
823 | encode_sv (enc, SP[i + 1 - count]); |
|
|
824 | encode_ch (enc, ','); |
|
|
825 | } |
|
|
826 | |
|
|
827 | encode_sv (enc, TOPs); |
|
|
828 | SP -= count; |
|
|
829 | } |
|
|
830 | |
|
|
831 | encode_ch (enc, ']'); |
|
|
832 | |
|
|
833 | FREETMPS; LEAVE; |
|
|
834 | } |
|
|
835 | else if ((enc->json.flags & F_CONV_BLESSED) && (method = gv_fetchmethod_autoload (stash, "TO_JSON", 0))) |
|
|
836 | { |
|
|
837 | dSP; |
|
|
838 | |
|
|
839 | ENTER; SAVETMPS; |
|
|
840 | PUSHMARK (SP); |
|
|
841 | // we re-bless the reference to get overload and other niceties right |
|
|
842 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
|
|
843 | |
|
|
844 | // calling with G_SCALAR ensures that we always get a 1 return value |
|
|
845 | PUTBACK; |
|
|
846 | call_sv ((SV *)GvCV (method), G_SCALAR); |
|
|
847 | SPAGAIN; |
|
|
848 | |
|
|
849 | // catch this surprisingly common error |
|
|
850 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
851 | croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
852 | |
|
|
853 | sv = POPs; |
|
|
854 | PUTBACK; |
|
|
855 | |
|
|
856 | encode_sv (enc, sv); |
|
|
857 | |
|
|
858 | FREETMPS; LEAVE; |
|
|
859 | } |
|
|
860 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
861 | encode_str (enc, "null", 4, 0); |
704 | else |
862 | else |
705 | { |
|
|
706 | #if 0 |
|
|
707 | if (0 && sv_derived_from (rv, "JSON::Literal")) |
|
|
708 | { |
|
|
709 | // not yet |
|
|
710 | } |
|
|
711 | #endif |
|
|
712 | if (enc->json.flags & F_CONV_BLESSED) |
|
|
713 | { |
|
|
714 | // we re-bless the reference to get overload and other niceties right |
|
|
715 | GV *to_json = gv_fetchmethod_autoload (SvSTASH (sv), "TO_JSON", 0); |
|
|
716 | |
|
|
717 | if (to_json) |
|
|
718 | { |
|
|
719 | dSP; |
|
|
720 | |
|
|
721 | ENTER; SAVETMPS; PUSHMARK (SP); |
|
|
722 | XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), SvSTASH (sv))); |
|
|
723 | |
|
|
724 | // calling with G_SCALAR ensures that we always get a 1 return value |
|
|
725 | PUTBACK; |
|
|
726 | call_sv ((SV *)GvCV (to_json), G_SCALAR); |
|
|
727 | SPAGAIN; |
|
|
728 | |
|
|
729 | // catch this surprisingly common error |
|
|
730 | if (SvROK (TOPs) && SvRV (TOPs) == sv) |
|
|
731 | croak ("%s::TO_JSON method returned same object as was passed instead of a new one", HvNAME (SvSTASH (sv))); |
|
|
732 | |
|
|
733 | sv = POPs; |
|
|
734 | PUTBACK; |
|
|
735 | |
|
|
736 | encode_sv (enc, sv); |
|
|
737 | |
|
|
738 | FREETMPS; LEAVE; |
|
|
739 | } |
|
|
740 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
741 | encode_str (enc, "null", 4, 0); |
|
|
742 | else |
|
|
743 | croak ("encountered object '%s', but neither allow_blessed enabled nor TO_JSON method available on it", |
|
|
744 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
|
|
745 | } |
|
|
746 | else if (enc->json.flags & F_ALLOW_BLESSED) |
|
|
747 | encode_str (enc, "null", 4, 0); |
|
|
748 | else |
|
|
749 | croak ("encountered object '%s', but neither allow_blessed nor convert_blessed settings are enabled", |
863 | croak ("encountered object '%s', but neither allow_blessed, convert_blessed nor allow_tags settings are enabled (or TO_JSON/FREEZE method missing)", |
750 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
864 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
751 | } |
|
|
752 | } |
865 | } |
753 | else if (svt == SVt_PVHV) |
866 | else if (svt == SVt_PVHV) |
754 | encode_hv (enc, (HV *)sv); |
867 | encode_hv (enc, (HV *)sv); |
755 | else if (svt == SVt_PVAV) |
868 | else if (svt == SVt_PVAV) |
756 | encode_av (enc, (AV *)sv); |
869 | encode_av (enc, (AV *)sv); |
757 | else if (svt < SVt_PVAV) |
870 | else if (svt < SVt_PVAV) |
758 | { |
871 | { |
759 | STRLEN len = 0; |
872 | int bool_type = ref_bool_type (sv); |
760 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
761 | |
873 | |
762 | if (len == 1 && *pv == '1') |
874 | if (bool_type == 1) |
763 | encode_str (enc, "true", 4, 0); |
875 | encode_str (enc, "true", 4, 0); |
764 | else if (len == 1 && *pv == '0') |
876 | else if (bool_type == 0) |
765 | encode_str (enc, "false", 5, 0); |
877 | encode_str (enc, "false", 5, 0); |
766 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
878 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
767 | encode_str (enc, "null", 4, 0); |
879 | encode_str (enc, "null", 4, 0); |
768 | else |
880 | else |
769 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
881 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
… | |
… | |
852 | static SV * |
964 | static SV * |
853 | encode_json (SV *scalar, JSON *json) |
965 | encode_json (SV *scalar, JSON *json) |
854 | { |
966 | { |
855 | enc_t enc; |
967 | enc_t enc; |
856 | |
968 | |
857 | if (!(json->flags & F_ALLOW_NONREF) && !SvROK (scalar)) |
969 | if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar)) |
858 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
970 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
859 | |
971 | |
860 | enc.json = *json; |
972 | enc.json = *json; |
861 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
973 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
862 | enc.cur = SvPVX (enc.sv); |
974 | enc.cur = SvPVX (enc.sv); |
… | |
… | |
1069 | *cur++ = *dec_cur++; |
1181 | *cur++ = *dec_cur++; |
1070 | while (--clen); |
1182 | while (--clen); |
1071 | |
1183 | |
1072 | utf8 = 1; |
1184 | utf8 = 1; |
1073 | } |
1185 | } |
|
|
1186 | else if (ch == '\t' && dec->json.flags & F_RELAXED) |
|
|
1187 | *cur++ = ch; |
1074 | else |
1188 | else |
1075 | { |
1189 | { |
1076 | --dec_cur; |
1190 | --dec_cur; |
1077 | |
1191 | |
1078 | if (!ch) |
1192 | if (!ch) |
… | |
… | |
1088 | |
1202 | |
1089 | if (sv) |
1203 | if (sv) |
1090 | { |
1204 | { |
1091 | STRLEN cur = SvCUR (sv); |
1205 | STRLEN cur = SvCUR (sv); |
1092 | |
1206 | |
1093 | if (SvLEN (sv) <= cur + len) |
1207 | if (SvLEN (sv) - cur <= len) |
1094 | SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
1208 | json_sv_grow (sv, cur, len); |
1095 | |
1209 | |
1096 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
1210 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
1097 | SvCUR_set (sv, SvCUR (sv) + len); |
1211 | SvCUR_set (sv, SvCUR (sv) + len); |
1098 | } |
1212 | } |
1099 | else |
1213 | else |
… | |
… | |
1390 | |
1504 | |
1391 | DEC_DEC_DEPTH; |
1505 | DEC_DEC_DEPTH; |
1392 | sv = newRV_noinc ((SV *)hv); |
1506 | sv = newRV_noinc ((SV *)hv); |
1393 | |
1507 | |
1394 | // check filter callbacks |
1508 | // check filter callbacks |
1395 | if (dec->json.flags & F_HOOK) |
1509 | if (expect_false (dec->json.flags & F_HOOK)) |
1396 | { |
1510 | { |
1397 | if (dec->json.cb_sk_object && HvKEYS (hv) == 1) |
1511 | if (dec->json.cb_sk_object && HvKEYS (hv) == 1) |
1398 | { |
1512 | { |
1399 | HE *cb, *he; |
1513 | HE *cb, *he; |
1400 | |
1514 | |
1401 | hv_iterinit (hv); |
1515 | hv_iterinit (hv); |
1402 | he = hv_iternext (hv); |
1516 | he = hv_iternext (hv); |
1403 | hv_iterinit (hv); |
1517 | hv_iterinit (hv); |
1404 | |
1518 | |
1405 | // the next line creates a mortal sv each time its called. |
1519 | // the next line creates a mortal sv each time it's called. |
1406 | // might want to optimise this for common cases. |
1520 | // might want to optimise this for common cases. |
1407 | cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); |
1521 | cb = hv_fetch_ent (dec->json.cb_sk_object, hv_iterkeysv (he), 0, 0); |
1408 | |
1522 | |
1409 | if (cb) |
1523 | if (cb) |
1410 | { |
1524 | { |
1411 | dSP; |
1525 | dSP; |
1412 | int count; |
1526 | int count; |
1413 | |
1527 | |
1414 | ENTER; SAVETMPS; PUSHMARK (SP); |
1528 | ENTER; SAVETMPS; |
|
|
1529 | PUSHMARK (SP); |
1415 | XPUSHs (HeVAL (he)); |
1530 | XPUSHs (HeVAL (he)); |
1416 | sv_2mortal (sv); |
1531 | sv_2mortal (sv); |
1417 | |
1532 | |
1418 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1533 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1419 | |
1534 | |
… | |
… | |
1421 | { |
1536 | { |
1422 | sv = newSVsv (POPs); |
1537 | sv = newSVsv (POPs); |
1423 | FREETMPS; LEAVE; |
1538 | FREETMPS; LEAVE; |
1424 | return sv; |
1539 | return sv; |
1425 | } |
1540 | } |
|
|
1541 | else if (count) |
|
|
1542 | croak ("filter_json_single_key_object callbacks must not return more than one scalar"); |
1426 | |
1543 | |
1427 | SvREFCNT_inc (sv); |
1544 | SvREFCNT_inc (sv); |
1428 | FREETMPS; LEAVE; |
1545 | FREETMPS; LEAVE; |
1429 | } |
1546 | } |
1430 | } |
1547 | } |
… | |
… | |
1432 | if (dec->json.cb_object) |
1549 | if (dec->json.cb_object) |
1433 | { |
1550 | { |
1434 | dSP; |
1551 | dSP; |
1435 | int count; |
1552 | int count; |
1436 | |
1553 | |
1437 | ENTER; SAVETMPS; PUSHMARK (SP); |
1554 | ENTER; SAVETMPS; |
|
|
1555 | PUSHMARK (SP); |
1438 | XPUSHs (sv_2mortal (sv)); |
1556 | XPUSHs (sv_2mortal (sv)); |
1439 | |
1557 | |
1440 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
1558 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
1441 | |
1559 | |
1442 | if (count == 1) |
1560 | if (count == 1) |
1443 | { |
|
|
1444 | sv = newSVsv (POPs); |
1561 | sv = newSVsv (POPs); |
1445 | FREETMPS; LEAVE; |
1562 | else if (count == 0) |
1446 | return sv; |
|
|
1447 | } |
|
|
1448 | |
|
|
1449 | SvREFCNT_inc (sv); |
1563 | SvREFCNT_inc (sv); |
|
|
1564 | else |
|
|
1565 | croak ("filter_json_object callbacks must not return more than one scalar"); |
|
|
1566 | |
1450 | FREETMPS; LEAVE; |
1567 | FREETMPS; LEAVE; |
1451 | } |
1568 | } |
1452 | } |
1569 | } |
1453 | |
1570 | |
1454 | return sv; |
1571 | return sv; |
… | |
… | |
1458 | DEC_DEC_DEPTH; |
1575 | DEC_DEC_DEPTH; |
1459 | return 0; |
1576 | return 0; |
1460 | } |
1577 | } |
1461 | |
1578 | |
1462 | static SV * |
1579 | static SV * |
|
|
1580 | decode_tag (dec_t *dec) |
|
|
1581 | { |
|
|
1582 | SV *tag = 0; |
|
|
1583 | SV *val = 0; |
|
|
1584 | |
|
|
1585 | if (!(dec->json.flags & F_ALLOW_TAGS)) |
|
|
1586 | ERR ("malformed JSON string, neither array, object, number, string or atom"); |
|
|
1587 | |
|
|
1588 | ++dec->cur; |
|
|
1589 | |
|
|
1590 | decode_ws (dec); |
|
|
1591 | |
|
|
1592 | tag = decode_sv (dec); |
|
|
1593 | if (!tag) |
|
|
1594 | goto fail; |
|
|
1595 | |
|
|
1596 | if (!SvPOK (tag)) |
|
|
1597 | ERR ("malformed JSON string, (tag) must be a string"); |
|
|
1598 | |
|
|
1599 | decode_ws (dec); |
|
|
1600 | |
|
|
1601 | if (*dec->cur != ')') |
|
|
1602 | ERR (") expected after tag"); |
|
|
1603 | |
|
|
1604 | ++dec->cur; |
|
|
1605 | |
|
|
1606 | decode_ws (dec); |
|
|
1607 | |
|
|
1608 | val = decode_sv (dec); |
|
|
1609 | if (!val) |
|
|
1610 | goto fail; |
|
|
1611 | |
|
|
1612 | if (!SvROK (val) || SvTYPE (SvRV (val)) != SVt_PVAV) |
|
|
1613 | ERR ("malformed JSON string, tag value must be an array"); |
|
|
1614 | |
|
|
1615 | { |
|
|
1616 | AV *av = (AV *)SvRV (val); |
|
|
1617 | int i, len = av_len (av) + 1; |
|
|
1618 | HV *stash = gv_stashsv (tag, 0); |
|
|
1619 | SV *sv; |
|
|
1620 | |
|
|
1621 | if (!stash) |
|
|
1622 | ERR ("cannot decode perl-object (package does not exist)"); |
|
|
1623 | |
|
|
1624 | GV *method = gv_fetchmethod_autoload (stash, "THAW", 0); |
|
|
1625 | |
|
|
1626 | if (!method) |
|
|
1627 | ERR ("cannot decode perl-object (package does not have a THAW method)"); |
|
|
1628 | |
|
|
1629 | dSP; |
|
|
1630 | |
|
|
1631 | ENTER; SAVETMPS; |
|
|
1632 | PUSHMARK (SP); |
|
|
1633 | EXTEND (SP, len + 2); |
|
|
1634 | // we re-bless the reference to get overload and other niceties right |
|
|
1635 | PUSHs (tag); |
|
|
1636 | PUSHs (sv_json); |
|
|
1637 | |
|
|
1638 | for (i = 0; i < len; ++i) |
|
|
1639 | PUSHs (*av_fetch (av, i, 1)); |
|
|
1640 | |
|
|
1641 | PUTBACK; |
|
|
1642 | call_sv ((SV *)GvCV (method), G_SCALAR); |
|
|
1643 | SPAGAIN; |
|
|
1644 | |
|
|
1645 | SvREFCNT_dec (tag); |
|
|
1646 | SvREFCNT_dec (val); |
|
|
1647 | sv = SvREFCNT_inc (POPs); |
|
|
1648 | |
|
|
1649 | PUTBACK; |
|
|
1650 | |
|
|
1651 | FREETMPS; LEAVE; |
|
|
1652 | |
|
|
1653 | return sv; |
|
|
1654 | } |
|
|
1655 | |
|
|
1656 | fail: |
|
|
1657 | SvREFCNT_dec (tag); |
|
|
1658 | SvREFCNT_dec (val); |
|
|
1659 | return 0; |
|
|
1660 | } |
|
|
1661 | |
|
|
1662 | static SV * |
1463 | decode_sv (dec_t *dec) |
1663 | decode_sv (dec_t *dec) |
1464 | { |
1664 | { |
1465 | // the beauty of JSON: you need exactly one character lookahead |
1665 | // the beauty of JSON: you need exactly one character lookahead |
1466 | // to parse everything. |
1666 | // to parse everything. |
1467 | switch (*dec->cur) |
1667 | switch (*dec->cur) |
1468 | { |
1668 | { |
1469 | case '"': ++dec->cur; return decode_str (dec); |
1669 | case '"': ++dec->cur; return decode_str (dec); |
1470 | case '[': ++dec->cur; return decode_av (dec); |
1670 | case '[': ++dec->cur; return decode_av (dec); |
1471 | case '{': ++dec->cur; return decode_hv (dec); |
1671 | case '{': ++dec->cur; return decode_hv (dec); |
|
|
1672 | case '(': return decode_tag (dec); |
1472 | |
1673 | |
1473 | case '-': |
1674 | case '-': |
1474 | case '0': case '1': case '2': case '3': case '4': |
1675 | case '0': case '1': case '2': case '3': case '4': |
1475 | case '5': case '6': case '7': case '8': case '9': |
1676 | case '5': case '6': case '7': case '8': case '9': |
1476 | return decode_num (dec); |
1677 | return decode_num (dec); |
1477 | |
1678 | |
|
|
1679 | case 'f': |
|
|
1680 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
|
|
1681 | { |
|
|
1682 | dec->cur += 5; |
|
|
1683 | |
|
|
1684 | if (expect_false (!dec->json.v_false)) |
|
|
1685 | dec->json.v_false = GET_BOOL (false); |
|
|
1686 | |
|
|
1687 | return newSVsv (dec->json.v_false); |
|
|
1688 | } |
|
|
1689 | else |
|
|
1690 | ERR ("'false' expected"); |
|
|
1691 | |
|
|
1692 | break; |
|
|
1693 | |
1478 | case 't': |
1694 | case 't': |
1479 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1695 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1480 | { |
1696 | { |
1481 | dec->cur += 4; |
1697 | dec->cur += 4; |
1482 | #if JSON_SLOW |
1698 | |
1483 | json_true = get_bool ("JSON::XS::true"); |
1699 | if (expect_false (!dec->json.v_true)) |
1484 | #endif |
1700 | dec->json.v_true = GET_BOOL (true); |
|
|
1701 | |
1485 | return newSVsv (json_true); |
1702 | return newSVsv (dec->json.v_true); |
1486 | } |
1703 | } |
1487 | else |
1704 | else |
1488 | ERR ("'true' expected"); |
1705 | ERR ("'true' expected"); |
1489 | |
|
|
1490 | break; |
|
|
1491 | |
|
|
1492 | case 'f': |
|
|
1493 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
|
|
1494 | { |
|
|
1495 | dec->cur += 5; |
|
|
1496 | #if JSON_SLOW |
|
|
1497 | json_false = get_bool ("JSON::XS::false"); |
|
|
1498 | #endif |
|
|
1499 | return newSVsv (json_false); |
|
|
1500 | } |
|
|
1501 | else |
|
|
1502 | ERR ("'false' expected"); |
|
|
1503 | |
1706 | |
1504 | break; |
1707 | break; |
1505 | |
1708 | |
1506 | case 'n': |
1709 | case 'n': |
1507 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
1710 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
… | |
… | |
1513 | ERR ("'null' expected"); |
1716 | ERR ("'null' expected"); |
1514 | |
1717 | |
1515 | break; |
1718 | break; |
1516 | |
1719 | |
1517 | default: |
1720 | default: |
1518 | ERR ("malformed JSON string, neither array, object, number, string or atom"); |
1721 | ERR ("malformed JSON string, neither tag, array, object, number, string or atom"); |
1519 | break; |
1722 | break; |
1520 | } |
1723 | } |
1521 | |
1724 | |
1522 | fail: |
1725 | fail: |
1523 | return 0; |
1726 | return 0; |
1524 | } |
1727 | } |
1525 | |
1728 | |
1526 | static SV * |
1729 | static SV * |
1527 | decode_json (SV *string, JSON *json, char **offset_return) |
1730 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1528 | { |
1731 | { |
1529 | dec_t dec; |
1732 | dec_t dec; |
1530 | SV *sv; |
1733 | SV *sv; |
1531 | |
1734 | |
1532 | /* work around bugs in 5.10 where manipulating magic values |
1735 | /* work around bugs in 5.10 where manipulating magic values |
1533 | * makes perl ignore the magic in subsequent accesses. |
1736 | * makes perl ignore the magic in subsequent accesses. |
1534 | * also make a copy of non-PV values, to get them into a clean |
1737 | * also make a copy of non-PV values, to get them into a clean |
1535 | * state (SvPV should do that, but it's buggy, see below). |
1738 | * state (SvPV should do that, but it's buggy, see below). |
|
|
1739 | * |
|
|
1740 | * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16), |
|
|
1741 | * as reported by Reini Urban. |
1536 | */ |
1742 | */ |
1537 | /*SvGETMAGIC (string);*/ |
1743 | /*SvGETMAGIC (string);*/ |
1538 | if (SvMAGICAL (string) || !SvPOK (string)) |
1744 | if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string)) |
1539 | string = sv_2mortal (newSVsv (string)); |
1745 | string = sv_2mortal (newSVsv (string)); |
1540 | |
1746 | |
1541 | SvUPGRADE (string, SVt_PV); |
1747 | SvUPGRADE (string, SVt_PV); |
1542 | |
1748 | |
1543 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
1749 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
… | |
… | |
1582 | |
1788 | |
1583 | decode_ws (&dec); |
1789 | decode_ws (&dec); |
1584 | sv = decode_sv (&dec); |
1790 | sv = decode_sv (&dec); |
1585 | |
1791 | |
1586 | if (offset_return) |
1792 | if (offset_return) |
1587 | *offset_return = dec.cur; |
1793 | *offset_return = dec.cur - SvPVX (string); |
1588 | |
1794 | else if (sv) |
1589 | if (!(offset_return || !sv)) |
|
|
1590 | { |
1795 | { |
1591 | // check for trailing garbage |
1796 | // check for trailing garbage |
1592 | decode_ws (&dec); |
1797 | decode_ws (&dec); |
1593 | |
1798 | |
1594 | if (*dec.cur) |
1799 | if (dec.cur != dec.end) |
1595 | { |
1800 | { |
1596 | dec.err = "garbage after JSON object"; |
1801 | dec.err = "garbage after JSON object"; |
1597 | SvREFCNT_dec (sv); |
1802 | SvREFCNT_dec (sv); |
1598 | sv = 0; |
1803 | sv = 0; |
1599 | } |
1804 | } |
… | |
… | |
1618 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1823 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1619 | } |
1824 | } |
1620 | |
1825 | |
1621 | sv = sv_2mortal (sv); |
1826 | sv = sv_2mortal (sv); |
1622 | |
1827 | |
1623 | if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1828 | if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv)) |
1624 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1829 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1625 | |
1830 | |
1626 | return sv; |
1831 | return sv; |
1627 | } |
1832 | } |
1628 | |
1833 | |
… | |
… | |
1637 | // the state machine here is a bit convoluted and could be simplified a lot |
1842 | // the state machine here is a bit convoluted and could be simplified a lot |
1638 | // but this would make it slower, so... |
1843 | // but this would make it slower, so... |
1639 | |
1844 | |
1640 | for (;;) |
1845 | for (;;) |
1641 | { |
1846 | { |
1642 | //printf ("loop pod %d *p<%c><%s>, mode %d nest %d\n", p - SvPVX (self->incr_text), *p, p, self->incr_mode, self->incr_nest);//D |
|
|
1643 | switch (self->incr_mode) |
1847 | switch (self->incr_mode) |
1644 | { |
1848 | { |
|
|
1849 | // reached end of a scalar, see if we are inside a nested structure or not |
|
|
1850 | end_of_scalar: |
|
|
1851 | self->incr_mode = INCR_M_JSON; |
|
|
1852 | |
|
|
1853 | if (self->incr_nest) // end of a scalar inside array, object or tag |
|
|
1854 | goto incr_m_json; |
|
|
1855 | else // end of scalar outside structure, json text ends here |
|
|
1856 | goto interrupt; |
|
|
1857 | |
1645 | // only used for initial whitespace skipping |
1858 | // only used for initial whitespace skipping |
1646 | case INCR_M_WS: |
1859 | case INCR_M_WS: |
1647 | for (;;) |
1860 | for (;;) |
1648 | { |
1861 | { |
1649 | if (*p > 0x20) |
1862 | if (*p > 0x20) |
… | |
… | |
1691 | ++p; |
1904 | ++p; |
1692 | } |
1905 | } |
1693 | |
1906 | |
1694 | break; |
1907 | break; |
1695 | |
1908 | |
|
|
1909 | // inside true/false/null |
|
|
1910 | case INCR_M_TFN: |
|
|
1911 | incr_m_tfn: |
|
|
1912 | for (;;) |
|
|
1913 | switch (*p++) |
|
|
1914 | { |
|
|
1915 | case 'r': case 'u': case 'e': // tRUE, falsE, nUll |
|
|
1916 | case 'a': case 'l': case 's': // fALSe, nuLL |
|
|
1917 | // allowed |
|
|
1918 | break; |
|
|
1919 | |
|
|
1920 | default: |
|
|
1921 | --p; |
|
|
1922 | goto end_of_scalar; |
|
|
1923 | } |
|
|
1924 | |
|
|
1925 | // inside a number |
|
|
1926 | case INCR_M_NUM: |
|
|
1927 | incr_m_num: |
|
|
1928 | for (;;) |
|
|
1929 | switch (*p++) |
|
|
1930 | { |
|
|
1931 | case 'e': case 'E': case '.': case '+': |
|
|
1932 | case '-': |
|
|
1933 | case '0': case '1': case '2': case '3': case '4': |
|
|
1934 | case '5': case '6': case '7': case '8': case '9': |
|
|
1935 | // allowed |
|
|
1936 | break; |
|
|
1937 | |
|
|
1938 | default: |
|
|
1939 | --p; |
|
|
1940 | goto end_of_scalar; |
|
|
1941 | } |
|
|
1942 | |
1696 | // inside a string |
1943 | // inside a string |
1697 | case INCR_M_STR: |
1944 | case INCR_M_STR: |
1698 | incr_m_str: |
1945 | incr_m_str: |
1699 | for (;;) |
1946 | for (;;) |
1700 | { |
1947 | { |
1701 | if (*p == '"') |
1948 | if (*p == '"') |
1702 | { |
1949 | { |
1703 | ++p; |
1950 | ++p; |
1704 | self->incr_mode = INCR_M_JSON; |
|
|
1705 | |
|
|
1706 | if (!self->incr_nest) |
|
|
1707 | goto interrupt; |
|
|
1708 | |
|
|
1709 | goto incr_m_json; |
1951 | goto end_of_scalar; |
1710 | } |
1952 | } |
1711 | else if (*p == '\\') |
1953 | else if (*p == '\\') |
1712 | { |
1954 | { |
1713 | ++p; // "virtually" consumes character after \ |
1955 | ++p; // "virtually" consumes character after \ |
1714 | |
1956 | |
… | |
… | |
1744 | --p; // do not eat the whitespace, let the next round do it |
1986 | --p; // do not eat the whitespace, let the next round do it |
1745 | goto interrupt; |
1987 | goto interrupt; |
1746 | } |
1988 | } |
1747 | break; |
1989 | break; |
1748 | |
1990 | |
|
|
1991 | // the following three blocks handle scalars. this makes the parser |
|
|
1992 | // more strict than required inside arrays or objects, and could |
|
|
1993 | // be moved to a special case on the toplevel (except strings) |
|
|
1994 | case 't': |
|
|
1995 | case 'f': |
|
|
1996 | case 'n': |
|
|
1997 | self->incr_mode = INCR_M_TFN; |
|
|
1998 | goto incr_m_tfn; |
|
|
1999 | |
|
|
2000 | case '-': |
|
|
2001 | case '0': case '1': case '2': case '3': case '4': |
|
|
2002 | case '5': case '6': case '7': case '8': case '9': |
|
|
2003 | self->incr_mode = INCR_M_NUM; |
|
|
2004 | goto incr_m_num; |
|
|
2005 | |
1749 | case '"': |
2006 | case '"': |
1750 | self->incr_mode = INCR_M_STR; |
2007 | self->incr_mode = INCR_M_STR; |
1751 | goto incr_m_str; |
2008 | goto incr_m_str; |
1752 | |
2009 | |
1753 | case '[': |
2010 | case '[': |
1754 | case '{': |
2011 | case '{': |
|
|
2012 | case '(': |
1755 | if (++self->incr_nest > self->max_depth) |
2013 | if (++self->incr_nest > self->max_depth) |
1756 | croak (ERR_NESTING_EXCEEDED); |
2014 | croak (ERR_NESTING_EXCEEDED); |
1757 | break; |
2015 | break; |
1758 | |
2016 | |
1759 | case ']': |
2017 | case ']': |
1760 | case '}': |
2018 | case '}': |
1761 | if (--self->incr_nest <= 0) |
2019 | if (--self->incr_nest <= 0) |
1762 | goto interrupt; |
2020 | goto interrupt; |
|
|
2021 | break; |
|
|
2022 | |
|
|
2023 | case ')': |
|
|
2024 | --self->incr_nest; |
1763 | break; |
2025 | break; |
1764 | |
2026 | |
1765 | case '#': |
2027 | case '#': |
1766 | self->incr_mode = INCR_M_C1; |
2028 | self->incr_mode = INCR_M_C1; |
1767 | goto incr_m_c; |
2029 | goto incr_m_c; |
… | |
… | |
1793 | i >= '0' && i <= '9' ? i - '0' |
2055 | i >= '0' && i <= '9' ? i - '0' |
1794 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
2056 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1795 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
2057 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1796 | : -1; |
2058 | : -1; |
1797 | |
2059 | |
1798 | json_stash = gv_stashpv ("JSON::XS" , 1); |
2060 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1799 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
2061 | bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); |
1800 | |
|
|
1801 | json_true = get_bool ("JSON::XS::true"); |
|
|
1802 | json_false = get_bool ("JSON::XS::false"); |
2062 | bool_false = get_bool ("Types::Serialiser::false"); |
|
|
2063 | bool_true = get_bool ("Types::Serialiser::true"); |
|
|
2064 | |
|
|
2065 | sv_json = newSVpv ("JSON", 0); |
|
|
2066 | SvREADONLY_on (sv_json); |
1803 | |
2067 | |
1804 | CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ |
2068 | CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ |
1805 | } |
2069 | } |
1806 | |
2070 | |
1807 | PROTOTYPES: DISABLE |
2071 | PROTOTYPES: DISABLE |
1808 | |
2072 | |
1809 | void CLONE (...) |
2073 | void CLONE (...) |
1810 | CODE: |
2074 | CODE: |
|
|
2075 | // as long as these writes are atomic, the race should not matter |
|
|
2076 | // as existing threads either already use 0, or use the old value, |
|
|
2077 | // which is sitll correct for the initial thread. |
1811 | json_stash = 0; |
2078 | json_stash = 0; |
1812 | json_boolean_stash = 0; |
2079 | bool_stash = 0; |
|
|
2080 | bool_false = 0; |
|
|
2081 | bool_true = 0; |
1813 | |
2082 | |
1814 | void new (char *klass) |
2083 | void new (char *klass) |
1815 | PPCODE: |
2084 | PPCODE: |
1816 | { |
2085 | { |
1817 | SV *pv = NEWSV (0, sizeof (JSON)); |
2086 | SV *pv = NEWSV (0, sizeof (JSON)); |
… | |
… | |
1820 | XPUSHs (sv_2mortal (sv_bless ( |
2089 | XPUSHs (sv_2mortal (sv_bless ( |
1821 | newRV_noinc (pv), |
2090 | newRV_noinc (pv), |
1822 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
2091 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
1823 | ))); |
2092 | ))); |
1824 | } |
2093 | } |
|
|
2094 | |
|
|
2095 | void boolean_values (JSON *self, SV *v_false = 0, SV *v_true = 0) |
|
|
2096 | PPCODE: |
|
|
2097 | self->v_false = newSVsv (v_false); |
|
|
2098 | self->v_true = newSVsv (v_true); |
|
|
2099 | XPUSHs (ST (0)); |
|
|
2100 | |
|
|
2101 | void get_boolean_values (JSON *self) |
|
|
2102 | PPCODE: |
|
|
2103 | if (self->v_false && self->v_true) |
|
|
2104 | { |
|
|
2105 | EXTEND (SP, 2); |
|
|
2106 | PUSHs (self->v_false); |
|
|
2107 | PUSHs (self->v_true); |
|
|
2108 | } |
1825 | |
2109 | |
1826 | void ascii (JSON *self, int enable = 1) |
2110 | void ascii (JSON *self, int enable = 1) |
1827 | ALIAS: |
2111 | ALIAS: |
1828 | ascii = F_ASCII |
2112 | ascii = F_ASCII |
1829 | latin1 = F_LATIN1 |
2113 | latin1 = F_LATIN1 |
… | |
… | |
1837 | shrink = F_SHRINK |
2121 | shrink = F_SHRINK |
1838 | allow_blessed = F_ALLOW_BLESSED |
2122 | allow_blessed = F_ALLOW_BLESSED |
1839 | convert_blessed = F_CONV_BLESSED |
2123 | convert_blessed = F_CONV_BLESSED |
1840 | relaxed = F_RELAXED |
2124 | relaxed = F_RELAXED |
1841 | allow_unknown = F_ALLOW_UNKNOWN |
2125 | allow_unknown = F_ALLOW_UNKNOWN |
|
|
2126 | allow_tags = F_ALLOW_TAGS |
1842 | PPCODE: |
2127 | PPCODE: |
1843 | { |
2128 | { |
1844 | if (enable) |
2129 | if (enable) |
1845 | self->flags |= ix; |
2130 | self->flags |= ix; |
1846 | else |
2131 | else |
… | |
… | |
1862 | get_shrink = F_SHRINK |
2147 | get_shrink = F_SHRINK |
1863 | get_allow_blessed = F_ALLOW_BLESSED |
2148 | get_allow_blessed = F_ALLOW_BLESSED |
1864 | get_convert_blessed = F_CONV_BLESSED |
2149 | get_convert_blessed = F_CONV_BLESSED |
1865 | get_relaxed = F_RELAXED |
2150 | get_relaxed = F_RELAXED |
1866 | get_allow_unknown = F_ALLOW_UNKNOWN |
2151 | get_allow_unknown = F_ALLOW_UNKNOWN |
|
|
2152 | get_allow_tags = F_ALLOW_TAGS |
1867 | PPCODE: |
2153 | PPCODE: |
1868 | XPUSHs (boolSV (self->flags & ix)); |
2154 | XPUSHs (boolSV (self->flags & ix)); |
1869 | |
2155 | |
1870 | void max_depth (JSON *self, U32 max_depth = 0x80000000UL) |
2156 | void max_depth (JSON *self, U32 max_depth = 0x80000000UL) |
1871 | PPCODE: |
2157 | PPCODE: |
… | |
… | |
1920 | XPUSHs (ST (0)); |
2206 | XPUSHs (ST (0)); |
1921 | } |
2207 | } |
1922 | |
2208 | |
1923 | void encode (JSON *self, SV *scalar) |
2209 | void encode (JSON *self, SV *scalar) |
1924 | PPCODE: |
2210 | PPCODE: |
1925 | PUTBACK; scalar = encode_json (scalar, self); SPAGAIN; |
2211 | PUTBACK; XPUSHs (encode_json (scalar, self)); |
1926 | XPUSHs (scalar); |
|
|
1927 | |
2212 | |
1928 | void decode (JSON *self, SV *jsonstr) |
2213 | void decode (JSON *self, SV *jsonstr) |
1929 | PPCODE: |
2214 | PPCODE: |
1930 | PUTBACK; jsonstr = decode_json (jsonstr, self, 0); SPAGAIN; |
2215 | PUTBACK; XPUSHs (decode_json (jsonstr, self, 0)); |
1931 | XPUSHs (jsonstr); |
|
|
1932 | |
2216 | |
1933 | void decode_prefix (JSON *self, SV *jsonstr) |
2217 | void decode_prefix (JSON *self, SV *jsonstr) |
1934 | PPCODE: |
2218 | PPCODE: |
1935 | { |
2219 | { |
1936 | SV *sv; |
2220 | SV *sv; |
1937 | char *offset; |
2221 | STRLEN offset; |
1938 | PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; |
2222 | PUTBACK; sv = decode_json (jsonstr, self, &offset); |
1939 | EXTEND (SP, 2); |
2223 | EXTEND (SP, 2); |
1940 | PUSHs (sv); |
2224 | PUSHs (sv); |
1941 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); |
2225 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset)))); |
1942 | } |
2226 | } |
1943 | |
2227 | |
1944 | void incr_parse (JSON *self, SV *jsonstr = 0) |
2228 | void incr_parse (JSON *self, SV *jsonstr = 0) |
1945 | PPCODE: |
2229 | PPCODE: |
1946 | { |
2230 | { |
… | |
… | |
1980 | { |
2264 | { |
1981 | STRLEN len; |
2265 | STRLEN len; |
1982 | const char *str = SvPV (jsonstr, len); |
2266 | const char *str = SvPV (jsonstr, len); |
1983 | STRLEN cur = SvCUR (self->incr_text); |
2267 | STRLEN cur = SvCUR (self->incr_text); |
1984 | |
2268 | |
1985 | if (SvLEN (self->incr_text) <= cur + len) |
2269 | if (SvLEN (self->incr_text) - cur <= len) |
1986 | SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
2270 | json_sv_grow (self->incr_text, cur, len); |
1987 | |
2271 | |
1988 | Move (str, SvEND (self->incr_text), len, char); |
2272 | Move (str, SvEND (self->incr_text), len, char); |
1989 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
2273 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
1990 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
2274 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
1991 | } |
2275 | } |
… | |
… | |
1993 | |
2277 | |
1994 | if (GIMME_V != G_VOID) |
2278 | if (GIMME_V != G_VOID) |
1995 | do |
2279 | do |
1996 | { |
2280 | { |
1997 | SV *sv; |
2281 | SV *sv; |
1998 | char *offset; |
2282 | STRLEN offset; |
1999 | |
2283 | |
2000 | if (!INCR_DONE (self)) |
2284 | if (!INCR_DONE (self)) |
2001 | { |
2285 | { |
2002 | incr_parse (self); |
2286 | incr_parse (self); |
2003 | |
2287 | |
… | |
… | |
2016 | |
2300 | |
2017 | break; |
2301 | break; |
2018 | } |
2302 | } |
2019 | } |
2303 | } |
2020 | |
2304 | |
2021 | PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; |
2305 | PUTBACK; sv = decode_json (self->incr_text, self, &offset); |
2022 | XPUSHs (sv); |
2306 | XPUSHs (sv); |
2023 | |
2307 | |
2024 | self->incr_pos -= offset - SvPVX (self->incr_text); |
2308 | self->incr_pos -= offset; |
2025 | self->incr_nest = 0; |
2309 | self->incr_nest = 0; |
2026 | self->incr_mode = 0; |
2310 | self->incr_mode = 0; |
2027 | |
2311 | |
2028 | sv_chop (self->incr_text, offset); |
2312 | sv_chop (self->incr_text, SvPVX (self->incr_text) + offset); |
2029 | } |
2313 | } |
2030 | while (GIMME_V == G_ARRAY); |
2314 | while (GIMME_V == G_ARRAY); |
2031 | } |
2315 | } |
2032 | |
2316 | |
2033 | SV *incr_text (JSON *self) |
2317 | SV *incr_text (JSON *self) |
… | |
… | |
2064 | self->incr_mode = 0; |
2348 | self->incr_mode = 0; |
2065 | } |
2349 | } |
2066 | |
2350 | |
2067 | void DESTROY (JSON *self) |
2351 | void DESTROY (JSON *self) |
2068 | CODE: |
2352 | CODE: |
|
|
2353 | SvREFCNT_dec (self->v_false); |
|
|
2354 | SvREFCNT_dec (self->v_true); |
2069 | SvREFCNT_dec (self->cb_sk_object); |
2355 | SvREFCNT_dec (self->cb_sk_object); |
2070 | SvREFCNT_dec (self->cb_object); |
2356 | SvREFCNT_dec (self->cb_object); |
2071 | SvREFCNT_dec (self->incr_text); |
2357 | SvREFCNT_dec (self->incr_text); |
2072 | |
2358 | |
2073 | PROTOTYPES: ENABLE |
2359 | PROTOTYPES: ENABLE |
… | |
… | |
2076 | PPCODE: |
2362 | PPCODE: |
2077 | { |
2363 | { |
2078 | JSON json; |
2364 | JSON json; |
2079 | json_init (&json); |
2365 | json_init (&json); |
2080 | json.flags |= F_UTF8; |
2366 | json.flags |= F_UTF8; |
2081 | PUTBACK; scalar = encode_json (scalar, &json); SPAGAIN; |
2367 | PUTBACK; XPUSHs (encode_json (scalar, &json)); |
2082 | XPUSHs (scalar); |
|
|
2083 | } |
2368 | } |
2084 | |
2369 | |
2085 | void decode_json (SV *jsonstr) |
2370 | void decode_json (SV *jsonstr) |
2086 | PPCODE: |
2371 | PPCODE: |
2087 | { |
2372 | { |
2088 | JSON json; |
2373 | JSON json; |
2089 | json_init (&json); |
2374 | json_init (&json); |
2090 | json.flags |= F_UTF8; |
2375 | json.flags |= F_UTF8; |
2091 | PUTBACK; jsonstr = decode_json (jsonstr, &json, 0); SPAGAIN; |
2376 | PUTBACK; XPUSHs (decode_json (jsonstr, &json, 0)); |
2092 | XPUSHs (jsonstr); |
|
|
2093 | } |
2377 | } |
2094 | |
2378 | |