… | |
… | |
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 | |
… | |
… | |
49 | #define F_ALLOW_TAGS 0x00004000UL |
50 | #define F_ALLOW_TAGS 0x00004000UL |
50 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
51 | #define F_HOOK 0x00080000UL // some hooks exist, so slow-path processing |
51 | |
52 | |
52 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
53 | #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER |
53 | |
54 | |
54 | #define INIT_SIZE 32 // initial scalar size to be allocated |
55 | #define INIT_SIZE 64 // initial scalar size to be allocated |
55 | #define INDENT_STEP 3 // spaces per indentation level |
56 | #define INDENT_STEP 3 // spaces per indentation level |
56 | |
57 | |
57 | #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 |
58 | |
59 | |
59 | #define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8) |
60 | #define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8) |
… | |
… | |
77 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
78 | <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg))) |
78 | |
79 | |
79 | #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?)" |
80 | |
81 | |
81 | #ifdef USE_ITHREADS |
82 | #ifdef USE_ITHREADS |
82 | # define JSON_SLOW 1 |
|
|
83 | # 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 (bool_stash ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 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)) |
85 | #else |
86 | #else |
86 | # define JSON_SLOW 0 |
|
|
87 | # define JSON_STASH json_stash |
87 | # define JSON_STASH json_stash |
88 | # define BOOL_STASH bool_stash |
88 | # define BOOL_STASH bool_stash |
|
|
89 | # define GET_BOOL(value) bool_ ## value |
89 | #endif |
90 | #endif |
90 | |
91 | |
91 | // 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 |
92 | #define STACK_HES 64 |
93 | #define STACK_HES 64 |
93 | |
94 | |
94 | static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean:: |
95 | static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean:: |
95 | static SV *bool_true, *bool_false, *sv_json; |
96 | static SV *bool_false, *bool_true; |
|
|
97 | static SV *sv_json; |
96 | |
98 | |
97 | enum { |
99 | enum { |
98 | 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 |
99 | INCR_M_STR, // inside string |
103 | INCR_M_STR, // inside string |
100 | INCR_M_BS, // inside backslash |
104 | INCR_M_BS, // inside backslash |
101 | INCR_M_C0, // inside comment in initial whitespace sequence |
105 | INCR_M_C0, // inside comment in initial whitespace sequence |
102 | INCR_M_C1, // inside comment in other places |
106 | INCR_M_C1, // inside comment in other places |
103 | INCR_M_JSON // outside anything, count nesting |
107 | INCR_M_JSON // outside anything, count nesting |
… | |
… | |
116 | // for the incremental parser |
120 | // for the incremental parser |
117 | SV *incr_text; // the source text so far |
121 | SV *incr_text; // the source text so far |
118 | STRLEN incr_pos; // the current offset into the text |
122 | STRLEN incr_pos; // the current offset into the text |
119 | int incr_nest; // {[]}-nesting level |
123 | int incr_nest; // {[]}-nesting level |
120 | unsigned char incr_mode; |
124 | unsigned char incr_mode; |
|
|
125 | |
|
|
126 | SV *v_false, *v_true; |
121 | } JSON; |
127 | } JSON; |
122 | |
128 | |
123 | INLINE void |
129 | INLINE void |
124 | json_init (JSON *json) |
130 | json_init (JSON *json) |
125 | { |
131 | { |
126 | Zero (json, 1, JSON); |
132 | static const JSON init = { F_ALLOW_NONREF, 512 }; |
127 | json->max_depth = 512; |
133 | |
|
|
134 | *json = init; |
128 | } |
135 | } |
129 | |
136 | |
130 | ///////////////////////////////////////////////////////////////////////////// |
137 | ///////////////////////////////////////////////////////////////////////////// |
131 | // utility functions |
138 | // utility functions |
132 | |
139 | |
… | |
… | |
154 | SvPV_renew (sv, SvCUR (sv) + 1); |
161 | SvPV_renew (sv, SvCUR (sv) + 1); |
155 | #endif |
162 | #endif |
156 | } |
163 | } |
157 | } |
164 | } |
158 | |
165 | |
|
|
166 | /* adds two STRLENs together, slow, and with paranoia */ |
|
|
167 | static 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 | |
159 | // 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 |
160 | // case of an error. |
192 | // case of an error. |
161 | // we special-case "safe" characters from U+80 .. U+7FF, |
193 | // we special-case "safe" characters from U+80 .. U+7FF, |
162 | // but use the very good perl function to parse anything else. |
194 | // but use the very good perl function to parse anything else. |
163 | // note that we never call this function for a ascii codepoints |
195 | // note that we never call this function for a ascii codepoints |
164 | INLINE UV |
196 | INLINE UV |
… | |
… | |
318 | if (len == 1) |
350 | if (len == 1) |
319 | if (*pv == '1') |
351 | if (*pv == '1') |
320 | return 1; |
352 | return 1; |
321 | else if (*pv == '0') |
353 | else if (*pv == '0') |
322 | return 0; |
354 | return 0; |
323 | |
|
|
324 | } |
355 | } |
325 | |
356 | |
326 | return -1; |
357 | return -1; |
327 | } |
358 | } |
328 | |
359 | |
… | |
… | |
333 | if (!SvROK (scalar)) |
364 | if (!SvROK (scalar)) |
334 | return 1; |
365 | return 1; |
335 | |
366 | |
336 | scalar = SvRV (scalar); |
367 | scalar = SvRV (scalar); |
337 | |
368 | |
|
|
369 | if (SvTYPE (scalar) >= SVt_PVMG) |
|
|
370 | { |
338 | if (SvSTASH (scalar) == bool_stash) |
371 | if (SvSTASH (scalar) == bool_stash) |
339 | return 1; |
372 | return 1; |
340 | |
373 | |
341 | if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0) |
374 | if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0) |
342 | return 1; |
375 | return 1; |
|
|
376 | } |
343 | |
377 | |
344 | return 0; |
378 | return 0; |
345 | } |
379 | } |
346 | |
380 | |
347 | ///////////////////////////////////////////////////////////////////////////// |
381 | ///////////////////////////////////////////////////////////////////////////// |
… | |
… | |
359 | } enc_t; |
393 | } enc_t; |
360 | |
394 | |
361 | INLINE void |
395 | INLINE void |
362 | need (enc_t *enc, STRLEN len) |
396 | need (enc_t *enc, STRLEN len) |
363 | { |
397 | { |
364 | if (expect_false (enc->cur + len >= enc->end)) |
398 | if (expect_false ((uintptr_t)(enc->end - enc->cur) < len)) |
365 | { |
399 | { |
366 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
400 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
367 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
401 | char *buf = json_sv_grow (enc->sv, cur, len); |
368 | enc->cur = SvPVX (enc->sv) + cur; |
402 | enc->cur = buf + cur; |
369 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
403 | enc->end = buf + SvLEN (enc->sv) - 1; |
370 | } |
404 | } |
371 | } |
405 | } |
372 | |
406 | |
373 | INLINE void |
407 | INLINE void |
374 | encode_ch (enc_t *enc, char ch) |
408 | encode_ch (enc_t *enc, char ch) |
… | |
… | |
390 | |
424 | |
391 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
425 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
392 | { |
426 | { |
393 | if (expect_false (ch == '"')) // but with slow exceptions |
427 | if (expect_false (ch == '"')) // but with slow exceptions |
394 | { |
428 | { |
395 | need (enc, len += 1); |
429 | need (enc, len + 1); |
396 | *enc->cur++ = '\\'; |
430 | *enc->cur++ = '\\'; |
397 | *enc->cur++ = '"'; |
431 | *enc->cur++ = '"'; |
398 | } |
432 | } |
399 | else if (expect_false (ch == '\\')) |
433 | else if (expect_false (ch == '\\')) |
400 | { |
434 | { |
401 | need (enc, len += 1); |
435 | need (enc, len + 1); |
402 | *enc->cur++ = '\\'; |
436 | *enc->cur++ = '\\'; |
403 | *enc->cur++ = '\\'; |
437 | *enc->cur++ = '\\'; |
404 | } |
438 | } |
405 | else |
439 | else |
406 | *enc->cur++ = ch; |
440 | *enc->cur++ = ch; |
… | |
… | |
409 | } |
443 | } |
410 | else |
444 | else |
411 | { |
445 | { |
412 | switch (ch) |
446 | switch (ch) |
413 | { |
447 | { |
414 | 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; |
415 | 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; |
416 | 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; |
417 | 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; |
418 | 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; |
419 | |
453 | |
420 | default: |
454 | default: |
421 | { |
455 | { |
422 | STRLEN clen; |
456 | STRLEN clen; |
423 | UV uch; |
457 | UV uch; |
… | |
… | |
439 | if (uch >= 0x10000UL) |
473 | if (uch >= 0x10000UL) |
440 | { |
474 | { |
441 | if (uch >= 0x110000UL) |
475 | if (uch >= 0x110000UL) |
442 | 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); |
443 | |
477 | |
444 | need (enc, len += 11); |
478 | need (enc, len + 11); |
445 | sprintf (enc->cur, "\\u%04x\\u%04x", |
479 | sprintf (enc->cur, "\\u%04x\\u%04x", |
446 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
480 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
447 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
481 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
448 | enc->cur += 12; |
482 | enc->cur += 12; |
449 | } |
483 | } |
450 | else |
484 | else |
451 | { |
485 | { |
452 | need (enc, len += 5); |
486 | need (enc, len + 5); |
453 | *enc->cur++ = '\\'; |
487 | *enc->cur++ = '\\'; |
454 | *enc->cur++ = 'u'; |
488 | *enc->cur++ = 'u'; |
455 | *enc->cur++ = PL_hexdigit [ uch >> 12 ]; |
489 | *enc->cur++ = PL_hexdigit [ uch >> 12 ]; |
456 | *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; |
490 | *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; |
457 | *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; |
491 | *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; |
… | |
… | |
465 | *enc->cur++ = uch; |
499 | *enc->cur++ = uch; |
466 | str += clen; |
500 | str += clen; |
467 | } |
501 | } |
468 | else if (is_utf8) |
502 | else if (is_utf8) |
469 | { |
503 | { |
470 | need (enc, len += clen); |
504 | need (enc, len + clen); |
471 | do |
505 | do |
472 | { |
506 | { |
473 | *enc->cur++ = *str++; |
507 | *enc->cur++ = *str++; |
474 | } |
508 | } |
475 | while (--clen); |
509 | while (--clen); |
476 | } |
510 | } |
477 | else |
511 | else |
478 | { |
512 | { |
479 | 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 |
480 | enc->cur = encode_utf8 (enc->cur, uch); |
514 | enc->cur = encode_utf8 (enc->cur, uch); |
481 | ++str; |
515 | ++str; |
482 | } |
516 | } |
483 | } |
517 | } |
484 | } |
518 | } |
… | |
… | |
748 | { |
782 | { |
749 | HV *stash = SvSTASH (sv); |
783 | HV *stash = SvSTASH (sv); |
750 | |
784 | |
751 | if (stash == bool_stash) |
785 | if (stash == bool_stash) |
752 | { |
786 | { |
753 | if (SvIV (sv)) |
|
|
754 | encode_str (enc, "true", 4, 0); |
787 | if (SvIV (sv)) encode_str (enc, "true" , 4, 0); |
755 | else |
|
|
756 | encode_str (enc, "false", 5, 0); |
788 | else encode_str (enc, "false", 5, 0); |
757 | } |
789 | } |
758 | else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0))) |
790 | else if ((enc->json.flags & F_ALLOW_TAGS) && (method = gv_fetchmethod_autoload (stash, "FREEZE", 0))) |
759 | { |
791 | { |
760 | int count; |
792 | int count; |
761 | dSP; |
793 | dSP; |
762 | |
794 | |
763 | ENTER; SAVETMPS; |
795 | ENTER; SAVETMPS; |
764 | SAVESTACK_POS (); |
|
|
765 | PUSHMARK (SP); |
796 | PUSHMARK (SP); |
766 | EXTEND (SP, 2); |
797 | EXTEND (SP, 2); |
767 | // we re-bless the reference to get overload and other niceties right |
798 | // we re-bless the reference to get overload and other niceties right |
768 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
799 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
769 | PUSHs (sv_json); |
800 | PUSHs (sv_json); |
… | |
… | |
781 | encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash)); |
812 | encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash)); |
782 | encode_ch (enc, '"'); |
813 | encode_ch (enc, '"'); |
783 | encode_ch (enc, ')'); |
814 | encode_ch (enc, ')'); |
784 | encode_ch (enc, '['); |
815 | encode_ch (enc, '['); |
785 | |
816 | |
786 | while (count) |
817 | if (count) |
787 | { |
818 | { |
|
|
819 | int i; |
|
|
820 | |
|
|
821 | for (i = 0; i < count - 1; ++i) |
|
|
822 | { |
788 | encode_sv (enc, SP[1 - count--]); |
823 | encode_sv (enc, SP[i + 1 - count]); |
789 | |
|
|
790 | if (count) |
|
|
791 | encode_ch (enc, ','); |
824 | encode_ch (enc, ','); |
|
|
825 | } |
|
|
826 | |
|
|
827 | encode_sv (enc, TOPs); |
|
|
828 | SP -= count; |
792 | } |
829 | } |
|
|
830 | |
|
|
831 | PUTBACK; |
793 | |
832 | |
794 | encode_ch (enc, ']'); |
833 | encode_ch (enc, ']'); |
795 | |
834 | |
796 | FREETMPS; LEAVE; |
835 | FREETMPS; LEAVE; |
797 | } |
836 | } |
… | |
… | |
879 | if (SvIsUV (sv) ? SvUVX (sv) <= 59000 |
918 | if (SvIsUV (sv) ? SvUVX (sv) <= 59000 |
880 | : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000) |
919 | : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000) |
881 | { |
920 | { |
882 | // optimise the "small number case" |
921 | // optimise the "small number case" |
883 | // code will likely be branchless and use only a single multiplication |
922 | // code will likely be branchless and use only a single multiplication |
884 | // works for numbers up to 59074 |
923 | // 4.28 works for numbers up to 59074, 4.29 works up to 59076 |
885 | I32 i = SvIVX (sv); |
924 | I32 i = SvIVX (sv); |
886 | U32 u; |
925 | U32 u; |
887 | char digit, nz = 0; |
926 | char digit, nz = 0; |
888 | |
927 | |
889 | need (enc, 6); |
928 | need (enc, 6); |
… | |
… | |
999 | else |
1038 | else |
1000 | break; |
1039 | break; |
1001 | } |
1040 | } |
1002 | else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09) |
1041 | else if (ch != 0x20 && ch != 0x0a && ch != 0x0d && ch != 0x09) |
1003 | break; // parse error, but let higher level handle it, gives better error messages |
1042 | break; // parse error, but let higher level handle it, gives better error messages |
1004 | |
1043 | else |
1005 | ++dec->cur; |
1044 | ++dec->cur; |
1006 | } |
1045 | } |
1007 | } |
1046 | } |
1008 | |
1047 | |
1009 | #define ERR(reason) SB dec->err = reason; goto fail; SE |
1048 | #define ERR(reason) SB dec->err = reason; goto fail; SE |
1010 | |
1049 | |
… | |
… | |
1165 | |
1204 | |
1166 | if (sv) |
1205 | if (sv) |
1167 | { |
1206 | { |
1168 | STRLEN cur = SvCUR (sv); |
1207 | STRLEN cur = SvCUR (sv); |
1169 | |
1208 | |
1170 | if (SvLEN (sv) <= cur + len) |
1209 | if (SvLEN (sv) - cur <= len) |
1171 | SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
1210 | json_sv_grow (sv, cur, len); |
1172 | |
1211 | |
1173 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
1212 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
1174 | SvCUR_set (sv, SvCUR (sv) + len); |
1213 | SvCUR_set (sv, SvCUR (sv) + len); |
1175 | } |
1214 | } |
1176 | else |
1215 | else |
… | |
… | |
1467 | |
1506 | |
1468 | DEC_DEC_DEPTH; |
1507 | DEC_DEC_DEPTH; |
1469 | sv = newRV_noinc ((SV *)hv); |
1508 | sv = newRV_noinc ((SV *)hv); |
1470 | |
1509 | |
1471 | // check filter callbacks |
1510 | // check filter callbacks |
1472 | if (dec->json.flags & F_HOOK) |
1511 | if (expect_false (dec->json.flags & F_HOOK)) |
1473 | { |
1512 | { |
1474 | if (dec->json.cb_sk_object && HvKEYS (hv) == 1) |
1513 | if (dec->json.cb_sk_object && HvKEYS (hv) == 1) |
1475 | { |
1514 | { |
1476 | HE *cb, *he; |
1515 | HE *cb, *he; |
1477 | |
1516 | |
… | |
… | |
1487 | { |
1526 | { |
1488 | dSP; |
1527 | dSP; |
1489 | int count; |
1528 | int count; |
1490 | |
1529 | |
1491 | ENTER; SAVETMPS; |
1530 | ENTER; SAVETMPS; |
1492 | SAVESTACK_POS (); |
|
|
1493 | PUSHMARK (SP); |
1531 | PUSHMARK (SP); |
1494 | XPUSHs (HeVAL (he)); |
1532 | XPUSHs (HeVAL (he)); |
1495 | sv_2mortal (sv); |
1533 | sv_2mortal (sv); |
1496 | |
1534 | |
1497 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1535 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1498 | |
1536 | |
1499 | if (count == 1) |
1537 | if (count == 1) |
1500 | { |
1538 | { |
1501 | sv = newSVsv (POPs); |
1539 | sv = newSVsv (POPs); |
|
|
1540 | PUTBACK; |
1502 | FREETMPS; LEAVE; |
1541 | FREETMPS; LEAVE; |
1503 | return sv; |
1542 | return sv; |
1504 | } |
1543 | } |
|
|
1544 | else if (count) |
|
|
1545 | croak ("filter_json_single_key_object callbacks must not return more than one scalar"); |
|
|
1546 | |
|
|
1547 | PUTBACK; |
1505 | |
1548 | |
1506 | SvREFCNT_inc (sv); |
1549 | SvREFCNT_inc (sv); |
|
|
1550 | |
1507 | FREETMPS; LEAVE; |
1551 | FREETMPS; LEAVE; |
1508 | } |
1552 | } |
1509 | } |
1553 | } |
1510 | |
1554 | |
1511 | if (dec->json.cb_object) |
1555 | if (dec->json.cb_object) |
1512 | { |
1556 | { |
1513 | dSP; |
1557 | dSP; |
1514 | int count; |
1558 | int count; |
1515 | |
1559 | |
1516 | ENTER; SAVETMPS; |
1560 | ENTER; SAVETMPS; |
1517 | SAVESTACK_POS (); |
|
|
1518 | PUSHMARK (SP); |
1561 | PUSHMARK (SP); |
1519 | XPUSHs (sv_2mortal (sv)); |
1562 | XPUSHs (sv_2mortal (sv)); |
1520 | |
1563 | |
1521 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
1564 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
1522 | |
1565 | |
1523 | if (count == 1) |
1566 | if (count == 1) |
1524 | { |
|
|
1525 | sv = newSVsv (POPs); |
1567 | sv = newSVsv (POPs); |
1526 | FREETMPS; LEAVE; |
1568 | else if (count == 0) |
1527 | return sv; |
|
|
1528 | } |
|
|
1529 | |
|
|
1530 | SvREFCNT_inc (sv); |
1569 | SvREFCNT_inc (sv); |
|
|
1570 | else |
|
|
1571 | croak ("filter_json_object callbacks must not return more than one scalar"); |
|
|
1572 | |
|
|
1573 | PUTBACK; |
|
|
1574 | |
1531 | FREETMPS; LEAVE; |
1575 | FREETMPS; LEAVE; |
1532 | } |
1576 | } |
1533 | } |
1577 | } |
1534 | |
1578 | |
1535 | return sv; |
1579 | return sv; |
… | |
… | |
1638 | case '-': |
1682 | case '-': |
1639 | case '0': case '1': case '2': case '3': case '4': |
1683 | case '0': case '1': case '2': case '3': case '4': |
1640 | case '5': case '6': case '7': case '8': case '9': |
1684 | case '5': case '6': case '7': case '8': case '9': |
1641 | return decode_num (dec); |
1685 | return decode_num (dec); |
1642 | |
1686 | |
|
|
1687 | case 'f': |
|
|
1688 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
|
|
1689 | { |
|
|
1690 | dec->cur += 5; |
|
|
1691 | |
|
|
1692 | if (expect_false (!dec->json.v_false)) |
|
|
1693 | dec->json.v_false = GET_BOOL (false); |
|
|
1694 | |
|
|
1695 | return newSVsv (dec->json.v_false); |
|
|
1696 | } |
|
|
1697 | else |
|
|
1698 | ERR ("'false' expected"); |
|
|
1699 | |
|
|
1700 | break; |
|
|
1701 | |
1643 | case 't': |
1702 | case 't': |
1644 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1703 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1645 | { |
1704 | { |
1646 | dec->cur += 4; |
1705 | dec->cur += 4; |
1647 | #if JSON_SLOW |
1706 | |
1648 | bool_true = get_bool ("Types::Serialiser::true"); |
1707 | if (expect_false (!dec->json.v_true)) |
1649 | #endif |
1708 | dec->json.v_true = GET_BOOL (true); |
|
|
1709 | |
1650 | return newSVsv (bool_true); |
1710 | return newSVsv (dec->json.v_true); |
1651 | } |
1711 | } |
1652 | else |
1712 | else |
1653 | ERR ("'true' expected"); |
1713 | ERR ("'true' expected"); |
1654 | |
|
|
1655 | break; |
|
|
1656 | |
|
|
1657 | case 'f': |
|
|
1658 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
|
|
1659 | { |
|
|
1660 | dec->cur += 5; |
|
|
1661 | #if JSON_SLOW |
|
|
1662 | bool_false = get_bool ("Types::Serialiser::false"); |
|
|
1663 | #endif |
|
|
1664 | return newSVsv (bool_false); |
|
|
1665 | } |
|
|
1666 | else |
|
|
1667 | ERR ("'false' expected"); |
|
|
1668 | |
1714 | |
1669 | break; |
1715 | break; |
1670 | |
1716 | |
1671 | case 'n': |
1717 | case 'n': |
1672 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
1718 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
… | |
… | |
1687 | fail: |
1733 | fail: |
1688 | return 0; |
1734 | return 0; |
1689 | } |
1735 | } |
1690 | |
1736 | |
1691 | static SV * |
1737 | static SV * |
1692 | decode_json (SV *string, JSON *json, char **offset_return) |
1738 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1693 | { |
1739 | { |
1694 | dec_t dec; |
1740 | dec_t dec; |
1695 | SV *sv; |
1741 | SV *sv; |
1696 | |
1742 | |
1697 | /* work around bugs in 5.10 where manipulating magic values |
1743 | /* work around bugs in 5.10 where manipulating magic values |
1698 | * makes perl ignore the magic in subsequent accesses. |
1744 | * makes perl ignore the magic in subsequent accesses. |
1699 | * also make a copy of non-PV values, to get them into a clean |
1745 | * also make a copy of non-PV values, to get them into a clean |
1700 | * state (SvPV should do that, but it's buggy, see below). |
1746 | * state (SvPV should do that, but it's buggy, see below). |
|
|
1747 | * |
|
|
1748 | * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16), |
|
|
1749 | * as reported by Reini Urban. |
1701 | */ |
1750 | */ |
1702 | /*SvGETMAGIC (string);*/ |
1751 | /*SvGETMAGIC (string);*/ |
1703 | if (SvMAGICAL (string) || !SvPOK (string)) |
1752 | if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string)) |
1704 | string = sv_2mortal (newSVsv (string)); |
1753 | string = sv_2mortal (newSVsv (string)); |
1705 | |
1754 | |
1706 | SvUPGRADE (string, SVt_PV); |
1755 | SvUPGRADE (string, SVt_PV); |
1707 | |
1756 | |
1708 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
1757 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
… | |
… | |
1747 | |
1796 | |
1748 | decode_ws (&dec); |
1797 | decode_ws (&dec); |
1749 | sv = decode_sv (&dec); |
1798 | sv = decode_sv (&dec); |
1750 | |
1799 | |
1751 | if (offset_return) |
1800 | if (offset_return) |
1752 | *offset_return = dec.cur; |
1801 | *offset_return = dec.cur - SvPVX (string); |
1753 | |
1802 | else if (sv) |
1754 | if (!(offset_return || !sv)) |
|
|
1755 | { |
1803 | { |
1756 | // check for trailing garbage |
1804 | // check for trailing garbage |
1757 | decode_ws (&dec); |
1805 | decode_ws (&dec); |
1758 | |
1806 | |
1759 | if (*dec.cur) |
1807 | if (dec.cur != dec.end) |
1760 | { |
1808 | { |
1761 | dec.err = "garbage after JSON object"; |
1809 | dec.err = "garbage after JSON object"; |
1762 | SvREFCNT_dec (sv); |
1810 | SvREFCNT_dec (sv); |
1763 | sv = 0; |
1811 | sv = 0; |
1764 | } |
1812 | } |
… | |
… | |
1802 | // the state machine here is a bit convoluted and could be simplified a lot |
1850 | // the state machine here is a bit convoluted and could be simplified a lot |
1803 | // but this would make it slower, so... |
1851 | // but this would make it slower, so... |
1804 | |
1852 | |
1805 | for (;;) |
1853 | for (;;) |
1806 | { |
1854 | { |
1807 | //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 |
|
|
1808 | switch (self->incr_mode) |
1855 | switch (self->incr_mode) |
1809 | { |
1856 | { |
|
|
1857 | // reached end of a scalar, see if we are inside a nested structure or not |
|
|
1858 | end_of_scalar: |
|
|
1859 | self->incr_mode = INCR_M_JSON; |
|
|
1860 | |
|
|
1861 | if (self->incr_nest) // end of a scalar inside array, object or tag |
|
|
1862 | goto incr_m_json; |
|
|
1863 | else // end of scalar outside structure, json text ends here |
|
|
1864 | goto interrupt; |
|
|
1865 | |
1810 | // only used for initial whitespace skipping |
1866 | // only used for initial whitespace skipping |
1811 | case INCR_M_WS: |
1867 | case INCR_M_WS: |
1812 | for (;;) |
1868 | for (;;) |
1813 | { |
1869 | { |
1814 | if (*p > 0x20) |
1870 | if (*p > 0x20) |
… | |
… | |
1856 | ++p; |
1912 | ++p; |
1857 | } |
1913 | } |
1858 | |
1914 | |
1859 | break; |
1915 | break; |
1860 | |
1916 | |
|
|
1917 | // inside true/false/null |
|
|
1918 | case INCR_M_TFN: |
|
|
1919 | incr_m_tfn: |
|
|
1920 | for (;;) |
|
|
1921 | switch (*p++) |
|
|
1922 | { |
|
|
1923 | case 'r': case 'u': case 'e': // tRUE, falsE, nUll |
|
|
1924 | case 'a': case 'l': case 's': // fALSe, nuLL |
|
|
1925 | // allowed |
|
|
1926 | break; |
|
|
1927 | |
|
|
1928 | default: |
|
|
1929 | --p; |
|
|
1930 | goto end_of_scalar; |
|
|
1931 | } |
|
|
1932 | |
|
|
1933 | // inside a number |
|
|
1934 | case INCR_M_NUM: |
|
|
1935 | incr_m_num: |
|
|
1936 | for (;;) |
|
|
1937 | switch (*p++) |
|
|
1938 | { |
|
|
1939 | case 'e': case 'E': case '.': case '+': |
|
|
1940 | case '-': |
|
|
1941 | case '0': case '1': case '2': case '3': case '4': |
|
|
1942 | case '5': case '6': case '7': case '8': case '9': |
|
|
1943 | // allowed |
|
|
1944 | break; |
|
|
1945 | |
|
|
1946 | default: |
|
|
1947 | --p; |
|
|
1948 | goto end_of_scalar; |
|
|
1949 | } |
|
|
1950 | |
1861 | // inside a string |
1951 | // inside a string |
1862 | case INCR_M_STR: |
1952 | case INCR_M_STR: |
1863 | incr_m_str: |
1953 | incr_m_str: |
1864 | for (;;) |
1954 | for (;;) |
1865 | { |
1955 | { |
1866 | if (*p == '"') |
1956 | if (*p == '"') |
1867 | { |
1957 | { |
1868 | ++p; |
1958 | ++p; |
1869 | self->incr_mode = INCR_M_JSON; |
|
|
1870 | |
|
|
1871 | if (!self->incr_nest) |
|
|
1872 | goto interrupt; |
|
|
1873 | |
|
|
1874 | goto incr_m_json; |
1959 | goto end_of_scalar; |
1875 | } |
1960 | } |
1876 | else if (*p == '\\') |
1961 | else if (*p == '\\') |
1877 | { |
1962 | { |
1878 | ++p; // "virtually" consumes character after \ |
1963 | ++p; // "virtually" consumes character after \ |
1879 | |
1964 | |
… | |
… | |
1909 | --p; // do not eat the whitespace, let the next round do it |
1994 | --p; // do not eat the whitespace, let the next round do it |
1910 | goto interrupt; |
1995 | goto interrupt; |
1911 | } |
1996 | } |
1912 | break; |
1997 | break; |
1913 | |
1998 | |
|
|
1999 | // the following three blocks handle scalars. this makes the parser |
|
|
2000 | // more strict than required inside arrays or objects, and could |
|
|
2001 | // be moved to a special case on the toplevel (except strings) |
|
|
2002 | case 't': |
|
|
2003 | case 'f': |
|
|
2004 | case 'n': |
|
|
2005 | self->incr_mode = INCR_M_TFN; |
|
|
2006 | goto incr_m_tfn; |
|
|
2007 | |
|
|
2008 | case '-': |
|
|
2009 | case '0': case '1': case '2': case '3': case '4': |
|
|
2010 | case '5': case '6': case '7': case '8': case '9': |
|
|
2011 | self->incr_mode = INCR_M_NUM; |
|
|
2012 | goto incr_m_num; |
|
|
2013 | |
1914 | case '"': |
2014 | case '"': |
1915 | self->incr_mode = INCR_M_STR; |
2015 | self->incr_mode = INCR_M_STR; |
1916 | goto incr_m_str; |
2016 | goto incr_m_str; |
1917 | |
2017 | |
1918 | case '[': |
2018 | case '[': |
… | |
… | |
1965 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
2065 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1966 | : -1; |
2066 | : -1; |
1967 | |
2067 | |
1968 | json_stash = gv_stashpv ("JSON::XS" , 1); |
2068 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1969 | bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); |
2069 | bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); |
|
|
2070 | bool_false = get_bool ("Types::Serialiser::false"); |
1970 | bool_true = get_bool ("Types::Serialiser::true"); |
2071 | bool_true = get_bool ("Types::Serialiser::true"); |
1971 | bool_false = get_bool ("Types::Serialiser::false"); |
|
|
1972 | |
2072 | |
1973 | sv_json = newSVpv ("JSON", 0); |
2073 | sv_json = newSVpv ("JSON", 0); |
1974 | SvREADONLY_on (sv_json); |
2074 | SvREADONLY_on (sv_json); |
1975 | |
2075 | |
1976 | CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ |
2076 | CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ |
… | |
… | |
1978 | |
2078 | |
1979 | PROTOTYPES: DISABLE |
2079 | PROTOTYPES: DISABLE |
1980 | |
2080 | |
1981 | void CLONE (...) |
2081 | void CLONE (...) |
1982 | CODE: |
2082 | CODE: |
|
|
2083 | // as long as these writes are atomic, the race should not matter |
|
|
2084 | // as existing threads either already use 0, or use the old value, |
|
|
2085 | // which is sitll correct for the initial thread. |
1983 | json_stash = 0; |
2086 | json_stash = 0; |
1984 | bool_stash = 0; |
2087 | bool_stash = 0; |
|
|
2088 | bool_false = 0; |
|
|
2089 | bool_true = 0; |
1985 | |
2090 | |
1986 | void new (char *klass) |
2091 | void new (char *klass) |
1987 | PPCODE: |
2092 | PPCODE: |
1988 | { |
2093 | { |
1989 | SV *pv = NEWSV (0, sizeof (JSON)); |
2094 | SV *pv = NEWSV (0, sizeof (JSON)); |
… | |
… | |
1992 | XPUSHs (sv_2mortal (sv_bless ( |
2097 | XPUSHs (sv_2mortal (sv_bless ( |
1993 | newRV_noinc (pv), |
2098 | newRV_noinc (pv), |
1994 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
2099 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
1995 | ))); |
2100 | ))); |
1996 | } |
2101 | } |
|
|
2102 | |
|
|
2103 | void boolean_values (JSON *self, SV *v_false = 0, SV *v_true = 0) |
|
|
2104 | PPCODE: |
|
|
2105 | self->v_false = newSVsv (v_false); |
|
|
2106 | self->v_true = newSVsv (v_true); |
|
|
2107 | XPUSHs (ST (0)); |
|
|
2108 | |
|
|
2109 | void get_boolean_values (JSON *self) |
|
|
2110 | PPCODE: |
|
|
2111 | if (self->v_false && self->v_true) |
|
|
2112 | { |
|
|
2113 | EXTEND (SP, 2); |
|
|
2114 | PUSHs (self->v_false); |
|
|
2115 | PUSHs (self->v_true); |
|
|
2116 | } |
1997 | |
2117 | |
1998 | void ascii (JSON *self, int enable = 1) |
2118 | void ascii (JSON *self, int enable = 1) |
1999 | ALIAS: |
2119 | ALIAS: |
2000 | ascii = F_ASCII |
2120 | ascii = F_ASCII |
2001 | latin1 = F_LATIN1 |
2121 | latin1 = F_LATIN1 |
… | |
… | |
2106 | |
2226 | |
2107 | void decode_prefix (JSON *self, SV *jsonstr) |
2227 | void decode_prefix (JSON *self, SV *jsonstr) |
2108 | PPCODE: |
2228 | PPCODE: |
2109 | { |
2229 | { |
2110 | SV *sv; |
2230 | SV *sv; |
2111 | char *offset; |
2231 | STRLEN offset; |
2112 | PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; |
2232 | PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; |
2113 | EXTEND (SP, 2); |
2233 | EXTEND (SP, 2); |
2114 | PUSHs (sv); |
2234 | PUSHs (sv); |
2115 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); |
2235 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset)))); |
2116 | } |
2236 | } |
2117 | |
2237 | |
2118 | void incr_parse (JSON *self, SV *jsonstr = 0) |
2238 | void incr_parse (JSON *self, SV *jsonstr = 0) |
2119 | PPCODE: |
2239 | PPCODE: |
2120 | { |
2240 | { |
… | |
… | |
2154 | { |
2274 | { |
2155 | STRLEN len; |
2275 | STRLEN len; |
2156 | const char *str = SvPV (jsonstr, len); |
2276 | const char *str = SvPV (jsonstr, len); |
2157 | STRLEN cur = SvCUR (self->incr_text); |
2277 | STRLEN cur = SvCUR (self->incr_text); |
2158 | |
2278 | |
2159 | if (SvLEN (self->incr_text) <= cur + len) |
2279 | if (SvLEN (self->incr_text) - cur <= len) |
2160 | SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
2280 | json_sv_grow (self->incr_text, cur, len); |
2161 | |
2281 | |
2162 | Move (str, SvEND (self->incr_text), len, char); |
2282 | Move (str, SvEND (self->incr_text), len, char); |
2163 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
2283 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
2164 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
2284 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
2165 | } |
2285 | } |
… | |
… | |
2167 | |
2287 | |
2168 | if (GIMME_V != G_VOID) |
2288 | if (GIMME_V != G_VOID) |
2169 | do |
2289 | do |
2170 | { |
2290 | { |
2171 | SV *sv; |
2291 | SV *sv; |
2172 | char *offset; |
2292 | STRLEN offset; |
2173 | |
2293 | |
2174 | if (!INCR_DONE (self)) |
2294 | if (!INCR_DONE (self)) |
2175 | { |
2295 | { |
2176 | incr_parse (self); |
2296 | incr_parse (self); |
2177 | |
2297 | |
… | |
… | |
2193 | } |
2313 | } |
2194 | |
2314 | |
2195 | PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; |
2315 | PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; |
2196 | XPUSHs (sv); |
2316 | XPUSHs (sv); |
2197 | |
2317 | |
2198 | self->incr_pos -= offset - SvPVX (self->incr_text); |
2318 | self->incr_pos -= offset; |
2199 | self->incr_nest = 0; |
2319 | self->incr_nest = 0; |
2200 | self->incr_mode = 0; |
2320 | self->incr_mode = 0; |
2201 | |
2321 | |
2202 | sv_chop (self->incr_text, offset); |
2322 | sv_chop (self->incr_text, SvPVX (self->incr_text) + offset); |
2203 | } |
2323 | } |
2204 | while (GIMME_V == G_ARRAY); |
2324 | while (GIMME_V == G_ARRAY); |
2205 | } |
2325 | } |
2206 | |
2326 | |
2207 | SV *incr_text (JSON *self) |
2327 | SV *incr_text (JSON *self) |
… | |
… | |
2238 | self->incr_mode = 0; |
2358 | self->incr_mode = 0; |
2239 | } |
2359 | } |
2240 | |
2360 | |
2241 | void DESTROY (JSON *self) |
2361 | void DESTROY (JSON *self) |
2242 | CODE: |
2362 | CODE: |
|
|
2363 | SvREFCNT_dec (self->v_false); |
|
|
2364 | SvREFCNT_dec (self->v_true); |
2243 | SvREFCNT_dec (self->cb_sk_object); |
2365 | SvREFCNT_dec (self->cb_sk_object); |
2244 | SvREFCNT_dec (self->cb_object); |
2366 | SvREFCNT_dec (self->cb_object); |
2245 | SvREFCNT_dec (self->incr_text); |
2367 | SvREFCNT_dec (self->incr_text); |
2246 | |
2368 | |
2247 | PROTOTYPES: ENABLE |
2369 | PROTOTYPES: ENABLE |