… | |
… | |
7 | #include "stdlib.h" |
7 | #include "stdlib.h" |
8 | #include "stdio.h" |
8 | #include "stdio.h" |
9 | |
9 | |
10 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
10 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
11 | # define snprintf _snprintf // C compilers have this in stdio.h |
11 | # define snprintf _snprintf // C compilers have this in stdio.h |
|
|
12 | #endif |
|
|
13 | |
|
|
14 | // some old perls do not have this, try to make it work, no |
|
|
15 | // guarentees, though. |
|
|
16 | #ifndef UTF8_MAXBYTES |
|
|
17 | # define UTF8_MAXBYTES 13 |
12 | #endif |
18 | #endif |
13 | |
19 | |
14 | #define F_ASCII 0x00000001UL |
20 | #define F_ASCII 0x00000001UL |
15 | #define F_LATIN1 0x00000002UL |
21 | #define F_LATIN1 0x00000002UL |
16 | #define F_UTF8 0x00000004UL |
22 | #define F_UTF8 0x00000004UL |
… | |
… | |
32 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
38 | #define F_DEFAULT (9UL << S_MAXDEPTH) |
33 | |
39 | |
34 | #define INIT_SIZE 32 // initial scalar size to be allocated |
40 | #define INIT_SIZE 32 // initial scalar size to be allocated |
35 | #define INDENT_STEP 3 // spaces per indentation level |
41 | #define INDENT_STEP 3 // spaces per indentation level |
36 | |
42 | |
37 | #define SHORT_STRING_LEN 512 // special-case strings of up to this size |
43 | #define SHORT_STRING_LEN 16384 // special-case strings of up to this size |
38 | |
44 | |
39 | #define SB do { |
45 | #define SB do { |
40 | #define SE } while (0) |
46 | #define SE } while (0) |
41 | |
47 | |
42 | #if __GNUC__ >= 3 |
48 | #if __GNUC__ >= 3 |
… | |
… | |
178 | STRLEN clen; |
184 | STRLEN clen; |
179 | UV uch; |
185 | UV uch; |
180 | |
186 | |
181 | if (is_utf8) |
187 | if (is_utf8) |
182 | { |
188 | { |
183 | //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY); |
|
|
184 | uch = decode_utf8 (str, end - str, &clen); |
189 | uch = decode_utf8 (str, end - str, &clen); |
185 | if (clen == (STRLEN)-1) |
190 | if (clen == (STRLEN)-1) |
186 | croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
191 | croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str); |
187 | } |
192 | } |
188 | else |
193 | else |
… | |
… | |
497 | encode_str (enc, str, len, SvUTF8 (sv)); |
502 | encode_str (enc, str, len, SvUTF8 (sv)); |
498 | encode_ch (enc, '"'); |
503 | encode_ch (enc, '"'); |
499 | } |
504 | } |
500 | else if (SvNOKp (sv)) |
505 | else if (SvNOKp (sv)) |
501 | { |
506 | { |
|
|
507 | // trust that perl will do the right thing w.r.t. JSON syntax. |
502 | need (enc, NV_DIG + 32); |
508 | need (enc, NV_DIG + 32); |
503 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
509 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
504 | enc->cur += strlen (enc->cur); |
510 | enc->cur += strlen (enc->cur); |
505 | } |
511 | } |
506 | else if (SvIOKp (sv)) |
512 | else if (SvIOKp (sv)) |
507 | { |
513 | { |
508 | // we assume we can always read an IV as a UV |
514 | // we assume we can always read an IV as a UV |
509 | if (SvUV (sv) & ~(UV)0x7fff) |
515 | if (SvUV (sv) & ~(UV)0x7fff) |
510 | { |
516 | { |
|
|
517 | // large integer, use the (rather slow) snprintf way. |
511 | need (enc, 32); |
518 | need (enc, sizeof (UV) * 3); |
512 | enc->cur += |
519 | enc->cur += |
513 | SvIsUV(sv) |
520 | SvIsUV(sv) |
514 | ? snprintf (enc->cur, 32, "%"UVuf, (UV)SvUVX (sv)) |
521 | ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv)) |
515 | : snprintf (enc->cur, 32, "%"IVdf, (IV)SvIVX (sv)); |
522 | : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv)); |
516 | } |
523 | } |
517 | else |
524 | else |
518 | { |
525 | { |
519 | // optimise the "small number case" |
526 | // optimise the "small number case" |
520 | // code will likely be branchless and use only a single multiplication |
527 | // code will likely be branchless and use only a single multiplication |
521 | I32 i = SvIV (sv); |
528 | I32 i = SvIV (sv); |
522 | U32 u; |
529 | U32 u; |
|
|
530 | char digit, nz = 0; |
523 | |
531 | |
524 | need (enc, 6); |
532 | need (enc, 6); |
525 | |
533 | |
526 | *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0; |
534 | *enc->cur = '-'; enc->cur += i < 0 ? 1 : 0; |
527 | u = i < 0 ? -i : i; |
535 | u = i < 0 ? -i : i; |
528 | |
536 | |
529 | // convert to 4.28 fixed-point representation |
537 | // convert to 4.28 fixed-point representation |
530 | u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits |
538 | u = u * ((0xfffffff + 10000) / 10000); // 10**5, 5 fractional digits |
531 | |
539 | |
532 | char digit, nz = 0; |
540 | // now output digit by digit, each time masking out the integer part |
533 | |
541 | // and multiplying by 5 while moving the decimal point one to the right, |
|
|
542 | // resulting in a net multiplication by 10. |
|
|
543 | // we always write the digit to memory but conditionally increment |
|
|
544 | // the pointer, to ease the usage of conditional move instructions. |
534 | digit = u >> 28; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0xfffffff) * 5; |
545 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; |
535 | digit = u >> 27; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0x7ffffff) * 5; |
546 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; |
536 | digit = u >> 26; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0x3ffffff) * 5; |
547 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; |
537 | digit = u >> 25; *enc->cur = digit + '0'; nz |= digit; enc->cur += nz ? 1 : 0; u = (u & 0x1ffffff) * 5; |
548 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; |
538 | digit = u >> 24; *enc->cur = digit + '0'; nz |= digit; enc->cur += 1; |
549 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
539 | } |
550 | } |
540 | } |
551 | } |
541 | else if (SvROK (sv)) |
552 | else if (SvROK (sv)) |
542 | encode_rv (enc, SvRV (sv)); |
553 | encode_rv (enc, SvRV (sv)); |
543 | else if (!SvOK (sv)) |
554 | else if (!SvOK (sv)) |
… | |
… | |
625 | decode_4hex (dec_t *dec) |
636 | decode_4hex (dec_t *dec) |
626 | { |
637 | { |
627 | signed char d1, d2, d3, d4; |
638 | signed char d1, d2, d3, d4; |
628 | unsigned char *cur = (unsigned char *)dec->cur; |
639 | unsigned char *cur = (unsigned char *)dec->cur; |
629 | |
640 | |
630 | d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("four hexadecimal digits expected"); |
641 | d1 = decode_hexdigit [cur [0]]; if (expect_false (d1 < 0)) ERR ("exactly four hexadecimal digits expected"); |
631 | d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("four hexadecimal digits expected"); |
642 | d2 = decode_hexdigit [cur [1]]; if (expect_false (d2 < 0)) ERR ("exactly four hexadecimal digits expected"); |
632 | d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("four hexadecimal digits expected"); |
643 | d3 = decode_hexdigit [cur [2]]; if (expect_false (d3 < 0)) ERR ("exactly four hexadecimal digits expected"); |
633 | d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("four hexadecimal digits expected"); |
644 | d4 = decode_hexdigit [cur [3]]; if (expect_false (d4 < 0)) ERR ("exactly four hexadecimal digits expected"); |
634 | |
645 | |
635 | dec->cur += 4; |
646 | dec->cur += 4; |
636 | |
647 | |
637 | return ((UV)d1) << 12 |
648 | return ((UV)d1) << 12 |
638 | | ((UV)d2) << 8 |
649 | | ((UV)d2) << 8 |
… | |
… | |
646 | static SV * |
657 | static SV * |
647 | decode_str (dec_t *dec) |
658 | decode_str (dec_t *dec) |
648 | { |
659 | { |
649 | SV *sv = 0; |
660 | SV *sv = 0; |
650 | int utf8 = 0; |
661 | int utf8 = 0; |
|
|
662 | char *dec_cur = dec->cur; |
651 | |
663 | |
652 | do |
664 | do |
653 | { |
665 | { |
654 | char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; |
666 | char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; |
655 | char *cur = buf; |
667 | char *cur = buf; |
656 | |
668 | |
657 | do |
669 | do |
658 | { |
670 | { |
659 | unsigned char ch = *(unsigned char *)dec->cur++; |
671 | unsigned char ch = *(unsigned char *)dec_cur++; |
660 | |
672 | |
661 | if (expect_false (ch == '"')) |
673 | if (expect_false (ch == '"')) |
662 | { |
674 | { |
663 | --dec->cur; |
675 | --dec_cur; |
664 | break; |
676 | break; |
665 | } |
677 | } |
666 | else if (expect_false (ch == '\\')) |
678 | else if (expect_false (ch == '\\')) |
667 | { |
679 | { |
668 | switch (*dec->cur) |
680 | switch (*dec_cur) |
669 | { |
681 | { |
670 | case '\\': |
682 | case '\\': |
671 | case '/': |
683 | case '/': |
672 | case '"': *cur++ = *dec->cur++; break; |
684 | case '"': *cur++ = *dec_cur++; break; |
673 | |
685 | |
674 | case 'b': ++dec->cur; *cur++ = '\010'; break; |
686 | case 'b': ++dec_cur; *cur++ = '\010'; break; |
675 | case 't': ++dec->cur; *cur++ = '\011'; break; |
687 | case 't': ++dec_cur; *cur++ = '\011'; break; |
676 | case 'n': ++dec->cur; *cur++ = '\012'; break; |
688 | case 'n': ++dec_cur; *cur++ = '\012'; break; |
677 | case 'f': ++dec->cur; *cur++ = '\014'; break; |
689 | case 'f': ++dec_cur; *cur++ = '\014'; break; |
678 | case 'r': ++dec->cur; *cur++ = '\015'; break; |
690 | case 'r': ++dec_cur; *cur++ = '\015'; break; |
679 | |
691 | |
680 | case 'u': |
692 | case 'u': |
681 | { |
693 | { |
682 | UV lo, hi; |
694 | UV lo, hi; |
683 | ++dec->cur; |
695 | ++dec_cur; |
684 | |
696 | |
|
|
697 | dec->cur = dec_cur; |
685 | hi = decode_4hex (dec); |
698 | hi = decode_4hex (dec); |
|
|
699 | dec_cur = dec->cur; |
686 | if (hi == (UV)-1) |
700 | if (hi == (UV)-1) |
687 | goto fail; |
701 | goto fail; |
688 | |
702 | |
689 | // possibly a surrogate pair |
703 | // possibly a surrogate pair |
690 | if (hi >= 0xd800) |
704 | if (hi >= 0xd800) |
691 | if (hi < 0xdc00) |
705 | if (hi < 0xdc00) |
692 | { |
706 | { |
693 | if (dec->cur [0] != '\\' || dec->cur [1] != 'u') |
707 | if (dec_cur [0] != '\\' || dec_cur [1] != 'u') |
694 | ERR ("missing low surrogate character in surrogate pair"); |
708 | ERR ("missing low surrogate character in surrogate pair"); |
695 | |
709 | |
696 | dec->cur += 2; |
710 | dec_cur += 2; |
697 | |
711 | |
|
|
712 | dec->cur = dec_cur; |
698 | lo = decode_4hex (dec); |
713 | lo = decode_4hex (dec); |
|
|
714 | dec_cur = dec->cur; |
699 | if (lo == (UV)-1) |
715 | if (lo == (UV)-1) |
700 | goto fail; |
716 | goto fail; |
701 | |
717 | |
702 | if (lo < 0xdc00 || lo >= 0xe000) |
718 | if (lo < 0xdc00 || lo >= 0xe000) |
703 | ERR ("surrogate pair expected"); |
719 | ERR ("surrogate pair expected"); |
… | |
… | |
717 | *cur++ = hi; |
733 | *cur++ = hi; |
718 | } |
734 | } |
719 | break; |
735 | break; |
720 | |
736 | |
721 | default: |
737 | default: |
722 | --dec->cur; |
738 | --dec_cur; |
723 | ERR ("illegal backslash escape sequence in string"); |
739 | ERR ("illegal backslash escape sequence in string"); |
724 | } |
740 | } |
725 | } |
741 | } |
726 | else if (expect_true (ch >= 0x20 && ch <= 0x7f)) |
742 | else if (expect_true (ch >= 0x20 && ch <= 0x7f)) |
727 | *cur++ = ch; |
743 | *cur++ = ch; |
728 | else if (ch >= 0x80) |
744 | else if (ch >= 0x80) |
729 | { |
745 | { |
730 | STRLEN clen; |
746 | STRLEN clen; |
731 | UV uch; |
747 | UV uch; |
732 | |
748 | |
733 | --dec->cur; |
749 | --dec_cur; |
734 | |
750 | |
735 | uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); |
751 | uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
736 | if (clen == (STRLEN)-1) |
752 | if (clen == (STRLEN)-1) |
737 | ERR ("malformed UTF-8 character in JSON string"); |
753 | ERR ("malformed UTF-8 character in JSON string"); |
738 | |
754 | |
739 | do |
755 | do |
740 | *cur++ = *dec->cur++; |
756 | *cur++ = *dec_cur++; |
741 | while (--clen); |
757 | while (--clen); |
742 | |
758 | |
743 | utf8 = 1; |
759 | utf8 = 1; |
744 | } |
760 | } |
745 | else |
761 | else |
746 | { |
762 | { |
747 | --dec->cur; |
763 | --dec_cur; |
748 | |
764 | |
749 | if (!ch) |
765 | if (!ch) |
750 | ERR ("unexpected end of string while parsing JSON string"); |
766 | ERR ("unexpected end of string while parsing JSON string"); |
751 | else |
767 | else |
752 | ERR ("invalid character encountered while parsing JSON string"); |
768 | ERR ("invalid character encountered while parsing JSON string"); |
… | |
… | |
765 | } |
781 | } |
766 | else |
782 | else |
767 | sv = newSVpvn (buf, len); |
783 | sv = newSVpvn (buf, len); |
768 | } |
784 | } |
769 | } |
785 | } |
770 | while (*dec->cur != '"'); |
786 | while (*dec_cur != '"'); |
771 | |
787 | |
772 | ++dec->cur; |
788 | ++dec_cur; |
773 | |
789 | |
774 | if (sv) |
790 | if (sv) |
775 | { |
791 | { |
776 | SvPOK_only (sv); |
792 | SvPOK_only (sv); |
777 | *SvEND (sv) = 0; |
793 | *SvEND (sv) = 0; |
… | |
… | |
780 | SvUTF8_on (sv); |
796 | SvUTF8_on (sv); |
781 | } |
797 | } |
782 | else |
798 | else |
783 | sv = newSVpvn ("", 0); |
799 | sv = newSVpvn ("", 0); |
784 | |
800 | |
|
|
801 | dec->cur = dec_cur; |
785 | return sv; |
802 | return sv; |
786 | |
803 | |
787 | fail: |
804 | fail: |
|
|
805 | dec->cur = dec_cur; |
788 | return 0; |
806 | return 0; |
789 | } |
807 | } |
790 | |
808 | |
791 | static SV * |
809 | static SV * |
792 | decode_num (dec_t *dec) |
810 | decode_num (dec_t *dec) |
… | |
… | |
854 | { |
872 | { |
855 | // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so |
873 | // special case the rather common 1..4-digit-int case, assumes 32 bit ints or so |
856 | if (*start == '-') |
874 | if (*start == '-') |
857 | switch (dec->cur - start) |
875 | switch (dec->cur - start) |
858 | { |
876 | { |
859 | case 2: return newSViv (-( start [1] - '0' )); |
877 | case 2: return newSViv (-( start [1] - '0' * 1)); |
860 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
878 | case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); |
861 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
879 | case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); |
862 | case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
880 | case 5: return newSViv (-(start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); |
863 | } |
881 | } |
864 | else |
882 | else |
865 | switch (dec->cur - start) |
883 | switch (dec->cur - start) |
866 | { |
884 | { |
867 | case 1: return newSViv ( start [0] - '0' ); |
885 | case 1: return newSViv ( start [0] - '0' * 1); |
868 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
886 | case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); |
869 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
887 | case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); |
870 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
888 | case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); |
871 | } |
889 | } |
872 | |
890 | |
… | |
… | |
879 | if (uv < (UV)IV_MIN) |
897 | if (uv < (UV)IV_MIN) |
880 | return newSViv (-(IV)uv); |
898 | return newSViv (-(IV)uv); |
881 | } |
899 | } |
882 | else |
900 | else |
883 | return newSVuv (uv); |
901 | return newSVuv (uv); |
|
|
902 | |
|
|
903 | // here would likely be the place for bigint support |
884 | } |
904 | } |
885 | } |
905 | } |
886 | |
906 | |
|
|
907 | // if we ever support bigint or bigfloat, this is the place for bigfloat |
887 | return newSVnv (Atof (start)); |
908 | return newSVnv (Atof (start)); |
888 | |
909 | |
889 | fail: |
910 | fail: |
890 | return 0; |
911 | return 0; |
891 | } |
912 | } |
… | |
… | |
992 | |
1013 | |
993 | static SV * |
1014 | static SV * |
994 | decode_sv (dec_t *dec) |
1015 | decode_sv (dec_t *dec) |
995 | { |
1016 | { |
996 | decode_ws (dec); |
1017 | decode_ws (dec); |
|
|
1018 | |
|
|
1019 | // the beauty of JSON: you need exactly one character lookahead |
|
|
1020 | // to parse anything. |
997 | switch (*dec->cur) |
1021 | switch (*dec->cur) |
998 | { |
1022 | { |
999 | case '"': ++dec->cur; return decode_str (dec); |
1023 | case '"': ++dec->cur; return decode_str (dec); |
1000 | case '[': ++dec->cur; return decode_av (dec); |
1024 | case '[': ++dec->cur; return decode_av (dec); |
1001 | case '{': ++dec->cur; return decode_hv (dec); |
1025 | case '{': ++dec->cur; return decode_hv (dec); |
… | |
… | |
1130 | MODULE = JSON::XS PACKAGE = JSON::XS |
1154 | MODULE = JSON::XS PACKAGE = JSON::XS |
1131 | |
1155 | |
1132 | BOOT: |
1156 | BOOT: |
1133 | { |
1157 | { |
1134 | int i; |
1158 | int i; |
1135 | |
|
|
1136 | memset (decode_hexdigit, 0xff, 256); |
|
|
1137 | |
1159 | |
1138 | for (i = 0; i < 256; ++i) |
1160 | for (i = 0; i < 256; ++i) |
1139 | decode_hexdigit [i] = |
1161 | decode_hexdigit [i] = |
1140 | i >= '0' && i <= '9' ? i - '0' |
1162 | i >= '0' && i <= '9' ? i - '0' |
1141 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1163 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |