… | |
… | |
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 | |
… | |
… | |
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_SLOW 1 |
83 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
84 | # define JSON_STASH (json_stash ? json_stash : gv_stashpv ("JSON::XS", 1)) |
|
|
85 | # define BOOL_STASH (bool_stash ? bool_stash : gv_stashpv ("Types::Serialiser::Boolean", 1)) |
84 | #else |
86 | #else |
85 | # define JSON_SLOW 0 |
87 | # define JSON_SLOW 0 |
86 | # define JSON_STASH json_stash |
88 | # define JSON_STASH json_stash |
|
|
89 | # define BOOL_STASH bool_stash |
87 | #endif |
90 | #endif |
88 | |
91 | |
89 | // 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 |
90 | #define STACK_HES 64 |
93 | #define STACK_HES 64 |
91 | |
94 | |
92 | static HV *json_stash, *types_boolean_stash; // JSON::XS:: |
95 | static HV *json_stash, *bool_stash; // JSON::XS::, Types::Serialiser::Boolean:: |
93 | static SV *types_true, *types_false, *sv_json; |
96 | static SV *bool_true, *bool_false, *sv_json; |
94 | |
97 | |
95 | enum { |
98 | enum { |
96 | INCR_M_WS = 0, // initial whitespace skipping, must be 0 |
99 | INCR_M_WS = 0, // initial whitespace skipping, must be 0 |
97 | INCR_M_STR, // inside string |
100 | INCR_M_STR, // inside string |
98 | INCR_M_BS, // inside backslash |
101 | INCR_M_BS, // inside backslash |
… | |
… | |
299 | // a recursion depth of ten gives us >>500 bits |
302 | // a recursion depth of ten gives us >>500 bits |
300 | json_atof_scan1 (s, &accum, &expo, 0, 10); |
303 | json_atof_scan1 (s, &accum, &expo, 0, 10); |
301 | |
304 | |
302 | return neg ? -accum : accum; |
305 | return neg ? -accum : accum; |
303 | } |
306 | } |
|
|
307 | |
|
|
308 | // target of scalar reference is bool? -1 == nope, 0 == false, 1 == true |
|
|
309 | static int |
|
|
310 | ref_bool_type (SV *sv) |
|
|
311 | { |
|
|
312 | svtype svt = SvTYPE (sv); |
|
|
313 | |
|
|
314 | if (svt < SVt_PVAV) |
|
|
315 | { |
|
|
316 | STRLEN len = 0; |
|
|
317 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
318 | |
|
|
319 | if (len == 1) |
|
|
320 | if (*pv == '1') |
|
|
321 | return 1; |
|
|
322 | else if (*pv == '0') |
|
|
323 | return 0; |
|
|
324 | } |
|
|
325 | |
|
|
326 | return -1; |
|
|
327 | } |
|
|
328 | |
|
|
329 | // returns whether scalar is not a reference in the sense of allow_nonref |
|
|
330 | static int |
|
|
331 | json_nonref (SV *scalar) |
|
|
332 | { |
|
|
333 | if (!SvROK (scalar)) |
|
|
334 | return 1; |
|
|
335 | |
|
|
336 | scalar = SvRV (scalar); |
|
|
337 | |
|
|
338 | if (SvTYPE (scalar) >= SVt_PVMG) |
|
|
339 | { |
|
|
340 | if (SvSTASH (scalar) == bool_stash) |
|
|
341 | return 1; |
|
|
342 | |
|
|
343 | if (!SvOBJECT (scalar) && ref_bool_type (scalar) >= 0) |
|
|
344 | return 1; |
|
|
345 | } |
|
|
346 | |
|
|
347 | return 0; |
|
|
348 | } |
|
|
349 | |
304 | ///////////////////////////////////////////////////////////////////////////// |
350 | ///////////////////////////////////////////////////////////////////////////// |
305 | // encoder |
351 | // encoder |
306 | |
352 | |
307 | // structure used for encoding JSON |
353 | // structure used for encoding JSON |
308 | typedef struct |
354 | typedef struct |
… | |
… | |
316 | } enc_t; |
362 | } enc_t; |
317 | |
363 | |
318 | INLINE void |
364 | INLINE void |
319 | need (enc_t *enc, STRLEN len) |
365 | need (enc_t *enc, STRLEN len) |
320 | { |
366 | { |
321 | if (expect_false (enc->cur + len >= enc->end)) |
367 | if (expect_false ((uintptr_t)(enc->end - enc->cur) < len)) |
322 | { |
368 | { |
323 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
369 | STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); |
324 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
370 | SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); |
325 | enc->cur = SvPVX (enc->sv) + cur; |
371 | enc->cur = SvPVX (enc->sv) + cur; |
326 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
372 | enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; |
… | |
… | |
701 | SvGETMAGIC (sv); |
747 | SvGETMAGIC (sv); |
702 | svt = SvTYPE (sv); |
748 | svt = SvTYPE (sv); |
703 | |
749 | |
704 | if (expect_false (SvOBJECT (sv))) |
750 | if (expect_false (SvOBJECT (sv))) |
705 | { |
751 | { |
706 | HV *boolean_stash = !JSON_SLOW || types_boolean_stash |
|
|
707 | ? types_boolean_stash |
|
|
708 | : gv_stashpv ("Types::Serialiser::Boolean", 1); |
|
|
709 | HV *stash = SvSTASH (sv); |
752 | HV *stash = SvSTASH (sv); |
710 | |
753 | |
711 | if (stash == boolean_stash) |
754 | if (stash == bool_stash) |
712 | { |
755 | { |
713 | if (SvIV (sv)) |
756 | if (SvIV (sv)) |
714 | encode_str (enc, "true", 4, 0); |
757 | encode_str (enc, "true", 4, 0); |
715 | else |
758 | else |
716 | encode_str (enc, "false", 5, 0); |
759 | encode_str (enc, "false", 5, 0); |
… | |
… | |
790 | encode_hv (enc, (HV *)sv); |
833 | encode_hv (enc, (HV *)sv); |
791 | else if (svt == SVt_PVAV) |
834 | else if (svt == SVt_PVAV) |
792 | encode_av (enc, (AV *)sv); |
835 | encode_av (enc, (AV *)sv); |
793 | else if (svt < SVt_PVAV) |
836 | else if (svt < SVt_PVAV) |
794 | { |
837 | { |
795 | STRLEN len = 0; |
838 | int bool_type = ref_bool_type (sv); |
796 | char *pv = svt ? SvPV (sv, len) : 0; |
|
|
797 | |
839 | |
798 | if (len == 1 && *pv == '1') |
840 | if (bool_type == 1) |
799 | encode_str (enc, "true", 4, 0); |
841 | encode_str (enc, "true", 4, 0); |
800 | else if (len == 1 && *pv == '0') |
842 | else if (bool_type == 0) |
801 | encode_str (enc, "false", 5, 0); |
843 | encode_str (enc, "false", 5, 0); |
802 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
844 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
803 | encode_str (enc, "null", 4, 0); |
845 | encode_str (enc, "null", 4, 0); |
804 | else |
846 | else |
805 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
847 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
… | |
… | |
883 | else |
925 | else |
884 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data", |
926 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, check your input data", |
885 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
927 | SvPV_nolen (sv), (unsigned int)SvFLAGS (sv)); |
886 | } |
928 | } |
887 | |
929 | |
888 | static int |
|
|
889 | json_scalar (SV *scalar) |
|
|
890 | { |
|
|
891 | return 0;//D |
|
|
892 | if (!SvROK (scalar)) |
|
|
893 | return 1; |
|
|
894 | } |
|
|
895 | |
|
|
896 | static SV * |
930 | static SV * |
897 | encode_json (SV *scalar, JSON *json) |
931 | encode_json (SV *scalar, JSON *json) |
898 | { |
932 | { |
899 | enc_t enc; |
933 | enc_t enc; |
900 | |
934 | |
901 | if (!(json->flags & F_ALLOW_NONREF) && json_scalar (scalar)) |
935 | if (!(json->flags & F_ALLOW_NONREF) && json_nonref (scalar)) |
902 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
936 | croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); |
903 | |
937 | |
904 | enc.json = *json; |
938 | enc.json = *json; |
905 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
939 | enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); |
906 | enc.cur = SvPVX (enc.sv); |
940 | enc.cur = SvPVX (enc.sv); |
… | |
… | |
1612 | case 't': |
1646 | case 't': |
1613 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1647 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1614 | { |
1648 | { |
1615 | dec->cur += 4; |
1649 | dec->cur += 4; |
1616 | #if JSON_SLOW |
1650 | #if JSON_SLOW |
1617 | types_true = get_bool ("Types::Serialiser::true"); |
1651 | bool_true = get_bool ("Types::Serialiser::true"); |
1618 | #endif |
1652 | #endif |
1619 | return newSVsv (types_true); |
1653 | return newSVsv (bool_true); |
1620 | } |
1654 | } |
1621 | else |
1655 | else |
1622 | ERR ("'true' expected"); |
1656 | ERR ("'true' expected"); |
1623 | |
1657 | |
1624 | break; |
1658 | break; |
… | |
… | |
1626 | case 'f': |
1660 | case 'f': |
1627 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1661 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1628 | { |
1662 | { |
1629 | dec->cur += 5; |
1663 | dec->cur += 5; |
1630 | #if JSON_SLOW |
1664 | #if JSON_SLOW |
1631 | types_false = get_bool ("Types::Serialiser::false"); |
1665 | bool_false = get_bool ("Types::Serialiser::false"); |
1632 | #endif |
1666 | #endif |
1633 | return newSVsv (types_false); |
1667 | return newSVsv (bool_false); |
1634 | } |
1668 | } |
1635 | else |
1669 | else |
1636 | ERR ("'false' expected"); |
1670 | ERR ("'false' expected"); |
1637 | |
1671 | |
1638 | break; |
1672 | break; |
… | |
… | |
1656 | fail: |
1690 | fail: |
1657 | return 0; |
1691 | return 0; |
1658 | } |
1692 | } |
1659 | |
1693 | |
1660 | static SV * |
1694 | static SV * |
1661 | decode_json (SV *string, JSON *json, char **offset_return) |
1695 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1662 | { |
1696 | { |
1663 | dec_t dec; |
1697 | dec_t dec; |
1664 | SV *sv; |
1698 | SV *sv; |
1665 | |
1699 | |
1666 | /* work around bugs in 5.10 where manipulating magic values |
1700 | /* work around bugs in 5.10 where manipulating magic values |
1667 | * makes perl ignore the magic in subsequent accesses. |
1701 | * makes perl ignore the magic in subsequent accesses. |
1668 | * also make a copy of non-PV values, to get them into a clean |
1702 | * also make a copy of non-PV values, to get them into a clean |
1669 | * state (SvPV should do that, but it's buggy, see below). |
1703 | * state (SvPV should do that, but it's buggy, see below). |
|
|
1704 | * |
|
|
1705 | * SvIsCOW_shared_hash works around a bug in perl (possibly 5.16), |
|
|
1706 | * as reported by Reini Urban. |
1670 | */ |
1707 | */ |
1671 | /*SvGETMAGIC (string);*/ |
1708 | /*SvGETMAGIC (string);*/ |
1672 | if (SvMAGICAL (string) || !SvPOK (string)) |
1709 | if (SvMAGICAL (string) || !SvPOK (string) || SvIsCOW_shared_hash (string)) |
1673 | string = sv_2mortal (newSVsv (string)); |
1710 | string = sv_2mortal (newSVsv (string)); |
1674 | |
1711 | |
1675 | SvUPGRADE (string, SVt_PV); |
1712 | SvUPGRADE (string, SVt_PV); |
1676 | |
1713 | |
1677 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
1714 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
… | |
… | |
1716 | |
1753 | |
1717 | decode_ws (&dec); |
1754 | decode_ws (&dec); |
1718 | sv = decode_sv (&dec); |
1755 | sv = decode_sv (&dec); |
1719 | |
1756 | |
1720 | if (offset_return) |
1757 | if (offset_return) |
1721 | *offset_return = dec.cur; |
1758 | *offset_return = dec.cur - SvPVX (string); |
1722 | |
1759 | else if (sv) |
1723 | if (!(offset_return || !sv)) |
|
|
1724 | { |
1760 | { |
1725 | // check for trailing garbage |
1761 | // check for trailing garbage |
1726 | decode_ws (&dec); |
1762 | decode_ws (&dec); |
1727 | |
1763 | |
1728 | if (*dec.cur) |
1764 | if (*dec.cur) |
… | |
… | |
1752 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1788 | dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); |
1753 | } |
1789 | } |
1754 | |
1790 | |
1755 | sv = sv_2mortal (sv); |
1791 | sv = sv_2mortal (sv); |
1756 | |
1792 | |
1757 | if (!(dec.json.flags & F_ALLOW_NONREF) && !SvROK (sv)) |
1793 | if (!(dec.json.flags & F_ALLOW_NONREF) && json_nonref (sv)) |
1758 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1794 | croak ("JSON text must be an object or array (but found number, string, true, false or null, use allow_nonref to allow this)"); |
1759 | |
1795 | |
1760 | return sv; |
1796 | return sv; |
1761 | } |
1797 | } |
1762 | |
1798 | |
… | |
… | |
1932 | i >= '0' && i <= '9' ? i - '0' |
1968 | i >= '0' && i <= '9' ? i - '0' |
1933 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1969 | : i >= 'a' && i <= 'f' ? i - 'a' + 10 |
1934 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1970 | : i >= 'A' && i <= 'F' ? i - 'A' + 10 |
1935 | : -1; |
1971 | : -1; |
1936 | |
1972 | |
1937 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1973 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1938 | types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); |
1974 | bool_stash = gv_stashpv ("Types::Serialiser::Boolean", 1); |
1939 | |
|
|
1940 | types_true = get_bool ("Types::Serialiser::true"); |
1975 | bool_true = get_bool ("Types::Serialiser::true"); |
1941 | types_false = get_bool ("Types::Serialiser::false"); |
1976 | bool_false = get_bool ("Types::Serialiser::false"); |
1942 | |
1977 | |
1943 | sv_json = newSVpv ("JSON", 0); |
1978 | sv_json = newSVpv ("JSON", 0); |
1944 | SvREADONLY_on (sv_json); |
1979 | SvREADONLY_on (sv_json); |
1945 | |
1980 | |
1946 | CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ |
1981 | CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */ |
… | |
… | |
1948 | |
1983 | |
1949 | PROTOTYPES: DISABLE |
1984 | PROTOTYPES: DISABLE |
1950 | |
1985 | |
1951 | void CLONE (...) |
1986 | void CLONE (...) |
1952 | CODE: |
1987 | CODE: |
1953 | json_stash = 0; |
1988 | json_stash = 0; |
1954 | types_boolean_stash = 0; |
1989 | bool_stash = 0; |
1955 | |
1990 | |
1956 | void new (char *klass) |
1991 | void new (char *klass) |
1957 | PPCODE: |
1992 | PPCODE: |
1958 | { |
1993 | { |
1959 | SV *pv = NEWSV (0, sizeof (JSON)); |
1994 | SV *pv = NEWSV (0, sizeof (JSON)); |
… | |
… | |
2076 | |
2111 | |
2077 | void decode_prefix (JSON *self, SV *jsonstr) |
2112 | void decode_prefix (JSON *self, SV *jsonstr) |
2078 | PPCODE: |
2113 | PPCODE: |
2079 | { |
2114 | { |
2080 | SV *sv; |
2115 | SV *sv; |
2081 | char *offset; |
2116 | STRLEN offset; |
2082 | PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; |
2117 | PUTBACK; sv = decode_json (jsonstr, self, &offset); SPAGAIN; |
2083 | EXTEND (SP, 2); |
2118 | EXTEND (SP, 2); |
2084 | PUSHs (sv); |
2119 | PUSHs (sv); |
2085 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset)))); |
2120 | PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, SvPV_nolen (jsonstr) + offset)))); |
2086 | } |
2121 | } |
2087 | |
2122 | |
2088 | void incr_parse (JSON *self, SV *jsonstr = 0) |
2123 | void incr_parse (JSON *self, SV *jsonstr = 0) |
2089 | PPCODE: |
2124 | PPCODE: |
2090 | { |
2125 | { |
… | |
… | |
2137 | |
2172 | |
2138 | if (GIMME_V != G_VOID) |
2173 | if (GIMME_V != G_VOID) |
2139 | do |
2174 | do |
2140 | { |
2175 | { |
2141 | SV *sv; |
2176 | SV *sv; |
2142 | char *offset; |
2177 | STRLEN offset; |
2143 | |
2178 | |
2144 | if (!INCR_DONE (self)) |
2179 | if (!INCR_DONE (self)) |
2145 | { |
2180 | { |
2146 | incr_parse (self); |
2181 | incr_parse (self); |
2147 | |
2182 | |
… | |
… | |
2163 | } |
2198 | } |
2164 | |
2199 | |
2165 | PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; |
2200 | PUTBACK; sv = decode_json (self->incr_text, self, &offset); SPAGAIN; |
2166 | XPUSHs (sv); |
2201 | XPUSHs (sv); |
2167 | |
2202 | |
2168 | self->incr_pos -= offset - SvPVX (self->incr_text); |
2203 | self->incr_pos -= offset; |
2169 | self->incr_nest = 0; |
2204 | self->incr_nest = 0; |
2170 | self->incr_mode = 0; |
2205 | self->incr_mode = 0; |
2171 | |
2206 | |
2172 | sv_chop (self->incr_text, offset); |
2207 | sv_chop (self->incr_text, SvPVX (self->incr_text) + offset); |
2173 | } |
2208 | } |
2174 | while (GIMME_V == G_ARRAY); |
2209 | while (GIMME_V == G_ARRAY); |
2175 | } |
2210 | } |
2176 | |
2211 | |
2177 | SV *incr_text (JSON *self) |
2212 | SV *incr_text (JSON *self) |