--- JSON-XS/XS.xs 2007/03/31 14:20:06 1.22 +++ JSON-XS/XS.xs 2007/05/23 22:07:13 1.34 @@ -5,15 +5,21 @@ #include "assert.h" #include "string.h" #include "stdlib.h" +#include "stdio.h" + +#if defined(__BORLANDC__) || defined(_MSC_VER) +# define snprintf _snprintf // C compilers have this in stdio.h +#endif #define F_ASCII 0x00000001UL -#define F_UTF8 0x00000002UL -#define F_INDENT 0x00000004UL -#define F_CANONICAL 0x00000008UL -#define F_SPACE_BEFORE 0x00000010UL -#define F_SPACE_AFTER 0x00000020UL -#define F_ALLOW_NONREF 0x00000080UL -#define F_SHRINK 0x00000100UL +#define F_LATIN1 0x00000002UL +#define F_UTF8 0x00000004UL +#define F_INDENT 0x00000008UL +#define F_CANONICAL 0x00000010UL +#define F_SPACE_BEFORE 0x00000020UL +#define F_SPACE_AFTER 0x00000040UL +#define F_ALLOW_NONREF 0x00000100UL +#define F_SHRINK 0x00000200UL #define F_MAXDEPTH 0xf8000000UL #define S_MAXDEPTH 27 @@ -23,12 +29,11 @@ // F_BLESSED? <=> { $__class__$ => } #define F_PRETTY F_INDENT | F_SPACE_BEFORE | F_SPACE_AFTER -#define F_DEFAULT (12UL << S_MAXDEPTH) +#define F_DEFAULT (9UL << S_MAXDEPTH) #define INIT_SIZE 32 // initial scalar size to be allocated #define INDENT_STEP 3 // spaces per indentation level -#define UTF8_MAX_LEN 11 // for perls UTF-X: max. number of octets per character #define SHORT_STRING_LEN 512 // special-case strings of up to this size #define SB do { @@ -78,7 +83,10 @@ return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); } else - return (UV)-1; + { + *clen = (STRLEN)-1; + return (UV)-1; + } } ///////////////////////////////////////////////////////////////////////////// @@ -103,7 +111,7 @@ STRLEN cur = enc->cur - SvPVX (enc->sv); SvGROW (enc->sv, cur + len + 1); enc->cur = SvPVX (enc->sv) + cur; - enc->end = SvPVX (enc->sv) + SvLEN (enc->sv); + enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; } } @@ -175,7 +183,7 @@ if (uch > 0x10FFFFUL) croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); - if (uch < 0x80 || enc->flags & F_ASCII) + if (uch < 0x80 || enc->flags & F_ASCII || (enc->flags & F_LATIN1 && uch > 0xFF)) { if (uch > 0xFFFFUL) { @@ -199,6 +207,11 @@ str += clen; } + else if (enc->flags & F_LATIN1) + { + *enc->cur++ = uch; + str += clen; + } else if (is_utf8) { need (enc, len += clen); @@ -210,7 +223,7 @@ } else { - need (enc, len += UTF8_MAX_LEN - 1); // never more than 11 bytes needed + need (enc, len += UTF8_MAXBYTES - 1); // never more than 11 bytes needed enc->cur = uvuni_to_utf8_flags (enc->cur, uch, 0); ++str; } @@ -361,8 +374,13 @@ // that randomises hash orderings if (enc->flags & F_CANONICAL) { - HE *he, *hes [count]; // if your compiler dies here, you need to enable C99 mode int fast = 1; + HE *he; +#if defined(__BORLANDC__) || defined(_MSC_VER) + HE **hes = _alloca (count * sizeof (HE)); +#else + HE *hes [count]; // if your compiler dies here, you need to enable C99 mode +#endif i = 0; while ((he = hv_iternext (hv))) @@ -407,7 +425,6 @@ } else { - SV *sv; HE *he = hv_iternext (hv); for (;;) @@ -432,9 +449,10 @@ static void encode_rv (enc_t *enc, SV *sv) { - SvGETMAGIC (sv); + svtype svt; - svtype svt = SvTYPE (sv); + SvGETMAGIC (sv); + svt = SvTYPE (sv); if (svt == SVt_PVHV) encode_hv (enc, (HV *)sv); @@ -494,10 +512,11 @@ static SV * encode_json (SV *scalar, U32 flags) { + enc_t enc; + if (!(flags & F_ALLOW_NONREF) && !SvROK (scalar)) croak ("hash- or arrayref expected (not a simple scalar, use allow_nonref to allow this)"); - enc_t enc; enc.flags = flags; enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); enc.cur = SvPVX (enc.sv); @@ -508,10 +527,11 @@ SvPOK_only (enc.sv); encode_sv (&enc, scalar); - if (!(flags & (F_ASCII | F_UTF8))) - SvUTF8_on (enc.sv); - SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); + *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings + + if (!(flags & (F_ASCII | F_LATIN1 | F_UTF8))) + SvUTF8_on (enc.sv); if (enc.flags & F_SHRINK) shrink (enc.sv); @@ -593,7 +613,7 @@ do { - char buf [SHORT_STRING_LEN + UTF8_MAX_LEN]; + char buf [SHORT_STRING_LEN + UTF8_MAXBYTES]; char *cur = buf; do @@ -669,39 +689,45 @@ *cur++ = ch; else if (ch >= 0x80) { + STRLEN clen; + UV uch; + --dec->cur; - STRLEN clen; - UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); + uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen); if (clen == (STRLEN)-1) ERR ("malformed UTF-8 character in JSON string"); do - { - *cur++ = *dec->cur++; - } + *cur++ = *dec->cur++; while (--clen); utf8 = 1; } - else if (!ch) - ERR ("unexpected end of string while parsing json string"); else - ERR ("invalid character encountered"); + { + --dec->cur; + if (!ch) + ERR ("unexpected end of string while parsing JSON string"); + else + ERR ("invalid character encountered while parsing JSON string"); + } } while (cur < buf + SHORT_STRING_LEN); - STRLEN len = cur - buf; + { + STRLEN len = cur - buf; - if (sv) - { - SvGROW (sv, SvCUR (sv) + len + 1); - memcpy (SvPVX (sv) + SvCUR (sv), buf, len); - SvCUR_set (sv, SvCUR (sv) + len); - } - else - sv = newSVpvn (buf, len); + if (sv) + { + SvGROW (sv, SvCUR (sv) + len + 1); + memcpy (SvPVX (sv) + SvCUR (sv), buf, len); + SvCUR_set (sv, SvCUR (sv) + len); + } + else + sv = newSVpvn (buf, len); + } } while (*dec->cur != '"'); @@ -955,7 +981,7 @@ break; default: - ERR ("malformed json string, neither array, object, number, string or atom"); + ERR ("malformed JSON string, neither array, object, number, string or atom"); break; } @@ -964,10 +990,13 @@ } static SV * -decode_json (SV *string, U32 flags) +decode_json (SV *string, U32 flags, UV *offset_return) { + dec_t dec; + UV offset; SV *sv; + SvGETMAGIC (string); SvUPGRADE (string, SVt_PV); if (flags & F_UTF8) @@ -977,7 +1006,6 @@ SvGROW (string, SvCUR (string) + 1); // should basically be a NOP - dec_t dec; dec.flags = flags; dec.cur = SvPVX (string); dec.end = SvEND (string); @@ -985,14 +1013,34 @@ dec.depth = 0; dec.maxdepth = DEC_DEPTH (dec.flags); - *dec.end = 0; // this should basically be a nop, too, but make sure its there + *dec.end = 0; // this should basically be a nop, too, but make sure it's there sv = decode_sv (&dec); + if (!(offset_return || !sv)) + { + // check for trailing garbage + decode_ws (&dec); + + if (*dec.cur) + { + dec.err = "garbage after JSON object"; + SvREFCNT_dec (sv); + sv = 0; + } + } + + if (offset_return || !sv) + { + offset = dec.flags & F_UTF8 + ? dec.cur - SvPVX (string) + : utf8_distance (dec.cur, SvPVX (string)); + + if (offset_return) + *offset_return = offset; + } + if (!sv) { - IV offset = dec.flags & F_UTF8 - ? dec.cur - SvPVX (string) - : utf8_distance (dec.cur, SvPVX (string)); SV *uni = sv_newmortal (); // horrible hack to silence warning inside pv_uni_display @@ -1004,7 +1052,7 @@ pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); LEAVE; - croak ("%s, at character offset %d (%s)", + croak ("%s, at character offset %d [\"%s\"]", dec.err, (int)offset, dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); @@ -1050,6 +1098,7 @@ SV *ascii (SV *self, int enable = 1) ALIAS: ascii = F_ASCII + latin1 = F_LATIN1 utf8 = F_UTF8 indent = F_INDENT canonical = F_CANONICAL @@ -1071,7 +1120,7 @@ OUTPUT: RETVAL -SV *max_depth (SV *self, int max_depth = 0x80000000UL) +SV *max_depth (SV *self, UV max_depth = 0x80000000UL) CODE: { UV *uv = SvJSON (self); @@ -1095,7 +1144,16 @@ void decode (SV *self, SV *jsonstr) PPCODE: - XPUSHs (decode_json (jsonstr, *SvJSON (self))); + XPUSHs (decode_json (jsonstr, *SvJSON (self), 0)); + +void decode_prefix (SV *self, SV *jsonstr) + PPCODE: +{ + UV offset; + EXTEND (SP, 2); + PUSHs (decode_json (jsonstr, *SvJSON (self), &offset)); + PUSHs (sv_2mortal (newSVuv (offset))); +} PROTOTYPES: ENABLE @@ -1109,5 +1167,5 @@ ALIAS: jsonToObj = 0 PPCODE: - XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8)); + XPUSHs (decode_json (jsonstr, F_DEFAULT | F_UTF8, 0));