… | |
… | |
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 | 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 |
… | |
… | |
361 | } enc_t; |
393 | } enc_t; |
362 | |
394 | |
363 | INLINE void |
395 | INLINE void |
364 | need (enc_t *enc, STRLEN len) |
396 | need (enc_t *enc, STRLEN len) |
365 | { |
397 | { |
366 | if (expect_false (enc->cur + len >= enc->end)) |
398 | if (expect_false ((uintptr_t)(enc->end - enc->cur) < len)) |
367 | { |
399 | { |
368 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
400 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
369 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
401 | char *buf = json_sv_grow (enc->sv, cur, len); |
370 | enc->cur = SvPVX (enc->sv) + cur; |
402 | enc->cur = buf + cur; |
371 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
403 | enc->end = buf + SvLEN (enc->sv) - 1; |
372 | } |
404 | } |
373 | } |
405 | } |
374 | |
406 | |
375 | INLINE void |
407 | INLINE void |
376 | encode_ch (enc_t *enc, char ch) |
408 | encode_ch (enc_t *enc, char ch) |
… | |
… | |
392 | |
424 | |
393 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
425 | if (expect_true (ch >= 0x20 && ch < 0x80)) // most common case |
394 | { |
426 | { |
395 | if (expect_false (ch == '"')) // but with slow exceptions |
427 | if (expect_false (ch == '"')) // but with slow exceptions |
396 | { |
428 | { |
397 | need (enc, len += 1); |
429 | need (enc, len + 1); |
398 | *enc->cur++ = '\\'; |
430 | *enc->cur++ = '\\'; |
399 | *enc->cur++ = '"'; |
431 | *enc->cur++ = '"'; |
400 | } |
432 | } |
401 | else if (expect_false (ch == '\\')) |
433 | else if (expect_false (ch == '\\')) |
402 | { |
434 | { |
403 | need (enc, len += 1); |
435 | need (enc, len + 1); |
404 | *enc->cur++ = '\\'; |
436 | *enc->cur++ = '\\'; |
405 | *enc->cur++ = '\\'; |
437 | *enc->cur++ = '\\'; |
406 | } |
438 | } |
407 | else |
439 | else |
408 | *enc->cur++ = ch; |
440 | *enc->cur++ = ch; |
… | |
… | |
411 | } |
443 | } |
412 | else |
444 | else |
413 | { |
445 | { |
414 | switch (ch) |
446 | switch (ch) |
415 | { |
447 | { |
416 | 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; |
417 | 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; |
418 | 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; |
419 | 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; |
420 | 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; |
421 | |
453 | |
422 | default: |
454 | default: |
423 | { |
455 | { |
424 | STRLEN clen; |
456 | STRLEN clen; |
425 | UV uch; |
457 | UV uch; |
… | |
… | |
441 | if (uch >= 0x10000UL) |
473 | if (uch >= 0x10000UL) |
442 | { |
474 | { |
443 | if (uch >= 0x110000UL) |
475 | if (uch >= 0x110000UL) |
444 | 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); |
445 | |
477 | |
446 | need (enc, len += 11); |
478 | need (enc, len + 11); |
447 | sprintf (enc->cur, "\\u%04x\\u%04x", |
479 | sprintf (enc->cur, "\\u%04x\\u%04x", |
448 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
480 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
449 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
481 | (int)((uch - 0x10000) % 0x400 + 0xDC00)); |
450 | enc->cur += 12; |
482 | enc->cur += 12; |
451 | } |
483 | } |
452 | else |
484 | else |
453 | { |
485 | { |
454 | need (enc, len += 5); |
486 | need (enc, len + 5); |
455 | *enc->cur++ = '\\'; |
487 | *enc->cur++ = '\\'; |
456 | *enc->cur++ = 'u'; |
488 | *enc->cur++ = 'u'; |
457 | *enc->cur++ = PL_hexdigit [ uch >> 12 ]; |
489 | *enc->cur++ = PL_hexdigit [ uch >> 12 ]; |
458 | *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; |
490 | *enc->cur++ = PL_hexdigit [(uch >> 8) & 15]; |
459 | *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; |
491 | *enc->cur++ = PL_hexdigit [(uch >> 4) & 15]; |
… | |
… | |
467 | *enc->cur++ = uch; |
499 | *enc->cur++ = uch; |
468 | str += clen; |
500 | str += clen; |
469 | } |
501 | } |
470 | else if (is_utf8) |
502 | else if (is_utf8) |
471 | { |
503 | { |
472 | need (enc, len += clen); |
504 | need (enc, len + clen); |
473 | do |
505 | do |
474 | { |
506 | { |
475 | *enc->cur++ = *str++; |
507 | *enc->cur++ = *str++; |
476 | } |
508 | } |
477 | while (--clen); |
509 | while (--clen); |
478 | } |
510 | } |
479 | else |
511 | else |
480 | { |
512 | { |
481 | 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 |
482 | enc->cur = encode_utf8 (enc->cur, uch); |
514 | enc->cur = encode_utf8 (enc->cur, uch); |
483 | ++str; |
515 | ++str; |
484 | } |
516 | } |
485 | } |
517 | } |
486 | } |
518 | } |
… | |
… | |
750 | { |
782 | { |
751 | HV *stash = SvSTASH (sv); |
783 | HV *stash = SvSTASH (sv); |
752 | |
784 | |
753 | if (stash == bool_stash) |
785 | if (stash == bool_stash) |
754 | { |
786 | { |
755 | if (SvIV (sv)) |
|
|
756 | encode_str (enc, "true", 4, 0); |
787 | if (SvIV (sv)) encode_str (enc, "true" , 4, 0); |
757 | else |
|
|
758 | encode_str (enc, "false", 5, 0); |
788 | else encode_str (enc, "false", 5, 0); |
759 | } |
789 | } |
760 | 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))) |
761 | { |
791 | { |
762 | int count; |
792 | int count; |
763 | dSP; |
793 | dSP; |
764 | |
794 | |
765 | ENTER; SAVETMPS; |
795 | ENTER; SAVETMPS; |
766 | SAVESTACK_POS (); |
|
|
767 | PUSHMARK (SP); |
796 | PUSHMARK (SP); |
768 | EXTEND (SP, 2); |
797 | EXTEND (SP, 2); |
769 | // 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 |
770 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
799 | PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); |
771 | PUSHs (sv_json); |
800 | PUSHs (sv_json); |
… | |
… | |
783 | encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash)); |
812 | encode_str (enc, HvNAME (stash), HvNAMELEN (stash), HvNAMEUTF8 (stash)); |
784 | encode_ch (enc, '"'); |
813 | encode_ch (enc, '"'); |
785 | encode_ch (enc, ')'); |
814 | encode_ch (enc, ')'); |
786 | encode_ch (enc, '['); |
815 | encode_ch (enc, '['); |
787 | |
816 | |
788 | while (count) |
817 | if (count) |
789 | { |
818 | { |
|
|
819 | int i; |
|
|
820 | |
|
|
821 | for (i = 0; i < count - 1; ++i) |
|
|
822 | { |
790 | encode_sv (enc, SP[1 - count--]); |
823 | encode_sv (enc, SP[i + 1 - count]); |
791 | |
|
|
792 | if (count) |
|
|
793 | encode_ch (enc, ','); |
824 | encode_ch (enc, ','); |
|
|
825 | } |
|
|
826 | |
|
|
827 | encode_sv (enc, TOPs); |
|
|
828 | SP -= count; |
794 | } |
829 | } |
795 | |
830 | |
796 | encode_ch (enc, ']'); |
831 | encode_ch (enc, ']'); |
797 | |
832 | |
798 | FREETMPS; LEAVE; |
833 | FREETMPS; LEAVE; |
… | |
… | |
1167 | |
1202 | |
1168 | if (sv) |
1203 | if (sv) |
1169 | { |
1204 | { |
1170 | STRLEN cur = SvCUR (sv); |
1205 | STRLEN cur = SvCUR (sv); |
1171 | |
1206 | |
1172 | if (SvLEN (sv) <= cur + len) |
1207 | if (SvLEN (sv) - cur <= len) |
1173 | SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
1208 | json_sv_grow (sv, cur, len); |
1174 | |
1209 | |
1175 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
1210 | memcpy (SvPVX (sv) + SvCUR (sv), buf, len); |
1176 | SvCUR_set (sv, SvCUR (sv) + len); |
1211 | SvCUR_set (sv, SvCUR (sv) + len); |
1177 | } |
1212 | } |
1178 | else |
1213 | else |
… | |
… | |
1489 | { |
1524 | { |
1490 | dSP; |
1525 | dSP; |
1491 | int count; |
1526 | int count; |
1492 | |
1527 | |
1493 | ENTER; SAVETMPS; |
1528 | ENTER; SAVETMPS; |
1494 | SAVESTACK_POS (); |
|
|
1495 | PUSHMARK (SP); |
1529 | PUSHMARK (SP); |
1496 | XPUSHs (HeVAL (he)); |
1530 | XPUSHs (HeVAL (he)); |
1497 | sv_2mortal (sv); |
1531 | sv_2mortal (sv); |
1498 | |
1532 | |
1499 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1533 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
… | |
… | |
1502 | { |
1536 | { |
1503 | sv = newSVsv (POPs); |
1537 | sv = newSVsv (POPs); |
1504 | FREETMPS; LEAVE; |
1538 | FREETMPS; LEAVE; |
1505 | return sv; |
1539 | return sv; |
1506 | } |
1540 | } |
|
|
1541 | else if (count) |
|
|
1542 | croak ("filter_json_single_key_object callbacks must not return more than one scalar"); |
1507 | |
1543 | |
1508 | SvREFCNT_inc (sv); |
1544 | SvREFCNT_inc (sv); |
1509 | FREETMPS; LEAVE; |
1545 | FREETMPS; LEAVE; |
1510 | } |
1546 | } |
1511 | } |
1547 | } |
… | |
… | |
1514 | { |
1550 | { |
1515 | dSP; |
1551 | dSP; |
1516 | int count; |
1552 | int count; |
1517 | |
1553 | |
1518 | ENTER; SAVETMPS; |
1554 | ENTER; SAVETMPS; |
1519 | SAVESTACK_POS (); |
|
|
1520 | PUSHMARK (SP); |
1555 | PUSHMARK (SP); |
1521 | XPUSHs (sv_2mortal (sv)); |
1556 | XPUSHs (sv_2mortal (sv)); |
1522 | |
1557 | |
1523 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
1558 | PUTBACK; count = call_sv (dec->json.cb_object, G_ARRAY); SPAGAIN; |
1524 | |
1559 | |
… | |
… | |
1526 | { |
1561 | { |
1527 | sv = newSVsv (POPs); |
1562 | sv = newSVsv (POPs); |
1528 | FREETMPS; LEAVE; |
1563 | FREETMPS; LEAVE; |
1529 | return sv; |
1564 | return sv; |
1530 | } |
1565 | } |
|
|
1566 | else if (count) |
|
|
1567 | croak ("filter_json_object callbacks must not return more than one scalar"); |
1531 | |
1568 | |
1532 | SvREFCNT_inc (sv); |
1569 | SvREFCNT_inc (sv); |
1533 | FREETMPS; LEAVE; |
1570 | FREETMPS; LEAVE; |
1534 | } |
1571 | } |
1535 | } |
1572 | } |
… | |
… | |
1640 | case '-': |
1677 | case '-': |
1641 | case '0': case '1': case '2': case '3': case '4': |
1678 | case '0': case '1': case '2': case '3': case '4': |
1642 | case '5': case '6': case '7': case '8': case '9': |
1679 | case '5': case '6': case '7': case '8': case '9': |
1643 | return decode_num (dec); |
1680 | return decode_num (dec); |
1644 | |
1681 | |
|
|
1682 | case 'f': |
|
|
1683 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
|
|
1684 | { |
|
|
1685 | dec->cur += 5; |
|
|
1686 | |
|
|
1687 | if (expect_false (!dec->json.v_false)) |
|
|
1688 | dec->json.v_false = GET_BOOL (false); |
|
|
1689 | |
|
|
1690 | return newSVsv (dec->json.v_false); |
|
|
1691 | } |
|
|
1692 | else |
|
|
1693 | ERR ("'false' expected"); |
|
|
1694 | |
|
|
1695 | break; |
|
|
1696 | |
1645 | case 't': |
1697 | case 't': |
1646 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1698 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1647 | { |
1699 | { |
1648 | dec->cur += 4; |
1700 | dec->cur += 4; |
1649 | #if JSON_SLOW |
1701 | |
1650 | bool_true = get_bool ("Types::Serialiser::true"); |
1702 | if (expect_false (!dec->json.v_true)) |
1651 | #endif |
1703 | dec->json.v_true = GET_BOOL (true); |
|
|
1704 | |
1652 | return newSVsv (bool_true); |
1705 | return newSVsv (dec->json.v_true); |
1653 | } |
1706 | } |
1654 | else |
1707 | else |
1655 | ERR ("'true' expected"); |
1708 | ERR ("'true' expected"); |
1656 | |
|
|
1657 | break; |
|
|
1658 | |
|
|
1659 | case 'f': |
|
|
1660 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
|
|
1661 | { |
|
|
1662 | dec->cur += 5; |
|
|
1663 | #if JSON_SLOW |
|
|
1664 | bool_false = get_bool ("Types::Serialiser::false"); |
|
|
1665 | #endif |
|
|
1666 | return newSVsv (bool_false); |
|
|
1667 | } |
|
|
1668 | else |
|
|
1669 | ERR ("'false' expected"); |
|
|
1670 | |
1709 | |
1671 | break; |
1710 | break; |
1672 | |
1711 | |
1673 | case 'n': |
1712 | case 'n': |
1674 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
1713 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "null", 4)) |
… | |
… | |
1689 | fail: |
1728 | fail: |
1690 | return 0; |
1729 | return 0; |
1691 | } |
1730 | } |
1692 | |
1731 | |
1693 | static SV * |
1732 | static SV * |
1694 | decode_json (SV *string, JSON *json, char **offset_return) |
1733 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1695 | { |
1734 | { |
1696 | dec_t dec; |
1735 | dec_t dec; |
1697 | SV *sv; |
1736 | SV *sv; |
1698 | |
1737 | |
1699 | /* work around bugs in 5.10 where manipulating magic values |
1738 | /* work around bugs in 5.10 where manipulating magic values |
… | |
… | |
1752 | |
1791 | |
1753 | decode_ws (&dec); |
1792 | decode_ws (&dec); |
1754 | sv = decode_sv (&dec); |
1793 | sv = decode_sv (&dec); |
1755 | |
1794 | |
1756 | if (offset_return) |
1795 | if (offset_return) |
1757 | *offset_return = dec.cur; |
1796 | *offset_return = dec.cur - SvPVX (string); |
1758 | |
1797 | else if (sv) |
1759 | if (!(offset_return || !sv)) |
|
|
1760 | { |
1798 | { |
1761 | // check for trailing garbage |
1799 | // check for trailing garbage |
1762 | decode_ws (&dec); |
1800 | decode_ws (&dec); |
1763 | |
1801 | |
1764 | if (*dec.cur) |
1802 | if (dec.cur != dec.end) |
1765 | { |
1803 | { |
1766 | dec.err = "garbage after JSON object"; |
1804 | dec.err = "garbage after JSON object"; |
1767 | SvREFCNT_dec (sv); |
1805 | SvREFCNT_dec (sv); |
1768 | sv = 0; |
1806 | sv = 0; |
1769 | } |
1807 | } |
… | |
… | |
1807 | // the state machine here is a bit convoluted and could be simplified a lot |
1845 | // the state machine here is a bit convoluted and could be simplified a lot |
1808 | // but this would make it slower, so... |
1846 | // but this would make it slower, so... |
1809 | |
1847 | |
1810 | for (;;) |
1848 | for (;;) |
1811 | { |
1849 | { |
1812 | //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 |
|
|
1813 | switch (self->incr_mode) |
1850 | switch (self->incr_mode) |
1814 | { |
1851 | { |
|
|
1852 | // reached end of a scalar, see if we are inside a nested structure or not |
|
|
1853 | end_of_scalar: |
|
|
1854 | self->incr_mode = INCR_M_JSON; |
|
|
1855 | |
|
|
1856 | if (self->incr_nest) // end of a scalar inside array, object or tag |
|
|
1857 | goto incr_m_json; |
|
|
1858 | else // end of scalar outside structure, json text ends here |
|
|
1859 | goto interrupt; |
|
|
1860 | |
1815 | // only used for initial whitespace skipping |
1861 | // only used for initial whitespace skipping |
1816 | case INCR_M_WS: |
1862 | case INCR_M_WS: |
1817 | for (;;) |
1863 | for (;;) |
1818 | { |
1864 | { |
1819 | if (*p > 0x20) |
1865 | if (*p > 0x20) |
… | |
… | |
1861 | ++p; |
1907 | ++p; |
1862 | } |
1908 | } |
1863 | |
1909 | |
1864 | break; |
1910 | break; |
1865 | |
1911 | |
|
|
1912 | // inside true/false/null |
|
|
1913 | case INCR_M_TFN: |
|
|
1914 | incr_m_tfn: |
|
|
1915 | for (;;) |
|
|
1916 | switch (*p++) |
|
|
1917 | { |
|
|
1918 | case 'r': case 'u': case 'e': // tRUE, falsE, nUll |
|
|
1919 | case 'a': case 'l': case 's': // fALSe, nuLL |
|
|
1920 | // allowed |
|
|
1921 | break; |
|
|
1922 | |
|
|
1923 | default: |
|
|
1924 | --p; |
|
|
1925 | goto end_of_scalar; |
|
|
1926 | } |
|
|
1927 | |
|
|
1928 | // inside a number |
|
|
1929 | case INCR_M_NUM: |
|
|
1930 | incr_m_num: |
|
|
1931 | for (;;) |
|
|
1932 | switch (*p++) |
|
|
1933 | { |
|
|
1934 | case 'e': case 'E': case '.': case '+': |
|
|
1935 | case '-': |
|
|
1936 | case '0': case '1': case '2': case '3': case '4': |
|
|
1937 | case '5': case '6': case '7': case '8': case '9': |
|
|
1938 | // allowed |
|
|
1939 | break; |
|
|
1940 | |
|
|
1941 | default: |
|
|
1942 | --p; |
|
|
1943 | goto end_of_scalar; |
|
|
1944 | } |
|
|
1945 | |
1866 | // inside a string |
1946 | // inside a string |
1867 | case INCR_M_STR: |
1947 | case INCR_M_STR: |
1868 | incr_m_str: |
1948 | incr_m_str: |
1869 | for (;;) |
1949 | for (;;) |
1870 | { |
1950 | { |
1871 | if (*p == '"') |
1951 | if (*p == '"') |
1872 | { |
1952 | { |
1873 | ++p; |
1953 | ++p; |
1874 | self->incr_mode = INCR_M_JSON; |
|
|
1875 | |
|
|
1876 | if (!self->incr_nest) |
|
|
1877 | goto interrupt; |
|
|
1878 | |
|
|
1879 | goto incr_m_json; |
1954 | goto end_of_scalar; |
1880 | } |
1955 | } |
1881 | else if (*p == '\\') |
1956 | else if (*p == '\\') |
1882 | { |
1957 | { |
1883 | ++p; // "virtually" consumes character after \ |
1958 | ++p; // "virtually" consumes character after \ |
1884 | |
1959 | |
… | |
… | |
1914 | --p; // do not eat the whitespace, let the next round do it |
1989 | --p; // do not eat the whitespace, let the next round do it |
1915 | goto interrupt; |
1990 | goto interrupt; |
1916 | } |
1991 | } |
1917 | break; |
1992 | break; |
1918 | |
1993 | |
|
|
1994 | // the following three blocks handle scalars. this makes the parser |
|
|
1995 | // more strict than required inside arrays or objects, and could |
|
|
1996 | // be moved to a special case on the toplevel (except strings) |
|
|
1997 | case 't': |
|
|
1998 | case 'f': |
|
|
1999 | case 'n': |
|
|
2000 | self->incr_mode = INCR_M_TFN; |
|
|
2001 | goto incr_m_tfn; |
|
|
2002 | |
|
|
2003 | case '-': |
|
|
2004 | case '0': case '1': case '2': case '3': case '4': |
|
|
2005 | case '5': case '6': case '7': case '8': case '9': |
|
|
2006 | self->incr_mode = INCR_M_NUM; |
|
|
2007 | goto incr_m_num; |
|
|
2008 | |
1919 | case '"': |
2009 | case '"': |
1920 | self->incr_mode = INCR_M_STR; |
2010 | self->incr_mode = INCR_M_STR; |
1921 | goto incr_m_str; |
2011 | goto incr_m_str; |
1922 | |
2012 | |
1923 | case '[': |
2013 | case '[': |
… | |
… | |
1970 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
2060 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1971 | : -1; |
2061 | : -1; |
1972 | |
2062 | |
1973 | json_stash = gv_stashpv ("JSON::XS" , 1); |
2063 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1974 | bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); |
2064 | bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); |
|
|
2065 | bool_false = get_bool ("Types::Serialiser::false"); |
1975 | bool_true = get_bool ("Types::Serialiser::true"); |
2066 | bool_true = get_bool ("Types::Serialiser::true"); |
1976 | bool_false = get_bool ("Types::Serialiser::false"); |
|
|
1977 | |
2067 | |
1978 | sv_json = newSVpv ("JSON", 0); |
2068 | sv_json = newSVpv ("JSON", 0); |
1979 | SvREADONLY_on (sv_json); |
2069 | SvREADONLY_on (sv_json); |
1980 | |
2070 | |
1981 | CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ |
2071 | CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ |
… | |
… | |
1983 | |
2073 | |
1984 | PROTOTYPES: DISABLE |
2074 | PROTOTYPES: DISABLE |
1985 | |
2075 | |
1986 | void CLONE (...) |
2076 | void CLONE (...) |
1987 | CODE: |
2077 | CODE: |
|
|
2078 | // as long as these writes are atomic, the race should not matter |
|
|
2079 | // as existing threads either already use 0, or use the old value, |
|
|
2080 | // which is sitll correct for the initial thread. |
1988 | json_stash = 0; |
2081 | json_stash = 0; |
1989 | bool_stash = 0; |
2082 | bool_stash = 0; |
|
|
2083 | bool_false = 0; |
|
|
2084 | bool_true = 0; |
1990 | |
2085 | |
1991 | void new (char *klass) |
2086 | void new (char *klass) |
1992 | PPCODE: |
2087 | PPCODE: |
1993 | { |
2088 | { |
1994 | SV *pv = NEWSV (0, sizeof (JSON)); |
2089 | SV *pv = NEWSV (0, sizeof (JSON)); |
… | |
… | |
1997 | XPUSHs (sv_2mortal (sv_bless ( |
2092 | XPUSHs (sv_2mortal (sv_bless ( |
1998 | newRV_noinc (pv), |
2093 | newRV_noinc (pv), |
1999 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
2094 | strEQ (klass, "JSON::XS") ? JSON_STASH : gv_stashpv (klass, 1) |
2000 | ))); |
2095 | ))); |
2001 | } |
2096 | } |
|
|
2097 | |
|
|
2098 | void boolean_values (JSON *self, SV *v_false = 0, SV *v_true = 0) |
|
|
2099 | PPCODE: |
|
|
2100 | self->v_false = newSVsv (v_false); |
|
|
2101 | self->v_true = newSVsv (v_true); |
|
|
2102 | XPUSHs (ST (0)); |
|
|
2103 | |
|
|
2104 | void get_boolean_values (JSON *self) |
|
|
2105 | PPCODE: |
|
|
2106 | if (self->v_false && self->v_true) |
|
|
2107 | { |
|
|
2108 | EXTEND (SP, 2); |
|
|
2109 | PUSHs (self->v_false); |
|
|
2110 | PUSHs (self->v_true); |
|
|
2111 | } |
2002 | |
2112 | |
2003 | void ascii (JSON *self, int enable = 1) |
2113 | void ascii (JSON *self, int enable = 1) |
2004 | ALIAS: |
2114 | ALIAS: |
2005 | ascii = F_ASCII |
2115 | ascii = F_ASCII |
2006 | latin1 = F_LATIN1 |
2116 | latin1 = F_LATIN1 |
… | |
… | |
2111 | |
2221 | |
2112 | void decode_prefix (JSON *self, SV *jsonstr) |
2222 | void decode_prefix (JSON *self, SV *jsonstr) |
2113 | PPCODE: |
2223 | PPCODE: |
2114 | { |
2224 | { |
2115 | SV *sv; |
2225 | SV *sv; |
2116 | char *offset; |
2226 | STRLEN offset; |
2117 | PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; |
2227 | PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; |
2118 | EXTEND (SP, 2); |
2228 | EXTEND (SP, 2); |
2119 | PUSHs (sv); |
2229 | PUSHs (sv); |
2120 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); |
2230 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset)))); |
2121 | } |
2231 | } |
2122 | |
2232 | |
2123 | void incr_parse (JSON *self, SV *jsonstr = 0) |
2233 | void incr_parse (JSON *self, SV *jsonstr = 0) |
2124 | PPCODE: |
2234 | PPCODE: |
2125 | { |
2235 | { |
… | |
… | |
2159 | { |
2269 | { |
2160 | STRLEN len; |
2270 | STRLEN len; |
2161 | const char *str = SvPV (jsonstr, len); |
2271 | const char *str = SvPV (jsonstr, len); |
2162 | STRLEN cur = SvCUR (self->incr_text); |
2272 | STRLEN cur = SvCUR (self->incr_text); |
2163 | |
2273 | |
2164 | if (SvLEN (self->incr_text) <= cur + len) |
2274 | if (SvLEN (self->incr_text) - cur <= len) |
2165 | SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
2275 | json_sv_grow (self->incr_text, cur, len); |
2166 | |
2276 | |
2167 | Move (str, SvEND (self->incr_text), len, char); |
2277 | Move (str, SvEND (self->incr_text), len, char); |
2168 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
2278 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
2169 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
2279 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
2170 | } |
2280 | } |
… | |
… | |
2172 | |
2282 | |
2173 | if (GIMME_V != G_VOID) |
2283 | if (GIMME_V != G_VOID) |
2174 | do |
2284 | do |
2175 | { |
2285 | { |
2176 | SV *sv; |
2286 | SV *sv; |
2177 | char *offset; |
2287 | STRLEN offset; |
2178 | |
2288 | |
2179 | if (!INCR_DONE (self)) |
2289 | if (!INCR_DONE (self)) |
2180 | { |
2290 | { |
2181 | incr_parse (self); |
2291 | incr_parse (self); |
2182 | |
2292 | |
… | |
… | |
2198 | } |
2308 | } |
2199 | |
2309 | |
2200 | PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; |
2310 | PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; |
2201 | XPUSHs (sv); |
2311 | XPUSHs (sv); |
2202 | |
2312 | |
2203 | self->incr_pos -= offset - SvPVX (self->incr_text); |
2313 | self->incr_pos -= offset; |
2204 | self->incr_nest = 0; |
2314 | self->incr_nest = 0; |
2205 | self->incr_mode = 0; |
2315 | self->incr_mode = 0; |
2206 | |
2316 | |
2207 | sv_chop (self->incr_text, offset); |
2317 | sv_chop (self->incr_text, SvPVX (self->incr_text) + offset); |
2208 | } |
2318 | } |
2209 | while (GIMME_V == G_ARRAY); |
2319 | while (GIMME_V == G_ARRAY); |
2210 | } |
2320 | } |
2211 | |
2321 | |
2212 | SV *incr_text (JSON *self) |
2322 | SV *incr_text (JSON *self) |
… | |
… | |
2243 | self->incr_mode = 0; |
2353 | self->incr_mode = 0; |
2244 | } |
2354 | } |
2245 | |
2355 | |
2246 | void DESTROY (JSON *self) |
2356 | void DESTROY (JSON *self) |
2247 | CODE: |
2357 | CODE: |
|
|
2358 | SvREFCNT_dec (self->v_false); |
|
|
2359 | SvREFCNT_dec (self->v_true); |
2248 | SvREFCNT_dec (self->cb_sk_object); |
2360 | SvREFCNT_dec (self->cb_sk_object); |
2249 | SvREFCNT_dec (self->cb_object); |
2361 | SvREFCNT_dec (self->cb_object); |
2250 | SvREFCNT_dec (self->incr_text); |
2362 | SvREFCNT_dec (self->incr_text); |
2251 | |
2363 | |
2252 | PROTOTYPES: ENABLE |
2364 | PROTOTYPES: ENABLE |