ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/JSON-XS/XS.xs
(Generate patch)

Comparing JSON-XS/XS.xs (file contents):
Revision 1.101 by root, Tue Sep 8 18:00:03 2009 UTC vs.
Revision 1.106 by root, Tue Jan 19 01:36:34 2010 UTC

… …
17// guarantees, though. if it breaks, you get to keep the pieces. 17// guarantees, though. if it breaks, you get to keep the pieces.
18#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
19# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
20#endif 20#endif
21 21
22// three extra for rounding, sign, and end of string
22#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2) 23#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 3)
23 24
24#define F_ASCII 0x00000001UL 25#define F_ASCII 0x00000001UL
25#define F_LATIN1 0x00000002UL 26#define F_LATIN1 0x00000002UL
26#define F_UTF8 0x00000004UL 27#define F_UTF8 0x00000004UL
27#define F_INDENT 0x00000008UL 28#define F_INDENT 0x00000008UL
… …
76 77
77enum { 78enum {
78 INCR_M_WS = 0, // initial whitespace skipping, must be 0 79 INCR_M_WS = 0, // initial whitespace skipping, must be 0
79 INCR_M_STR, // inside string 80 INCR_M_STR, // inside string
80 INCR_M_BS, // inside backslash 81 INCR_M_BS, // inside backslash
82 INCR_M_C0, // inside comment in initial whitespace sequence
83 INCR_M_C1, // inside comment in other places
81 INCR_M_JSON // outside anything, count nesting 84 INCR_M_JSON // outside anything, count nesting
82}; 85};
83 86
84#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) 87#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
85 88
… …
184 return SvUTF8 (sv) 187 return SvUTF8 (sv)
185 ? utf8_distance (offset, SvPVX (sv)) 188 ? utf8_distance (offset, SvPVX (sv))
186 : offset - SvPVX (sv); 189 : offset - SvPVX (sv);
187} 190}
188 191
192/////////////////////////////////////////////////////////////////////////////
193// fp hell
194
195// scan a group of digits, and a trailing exponent
196static void
197json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
198{
199 UV uaccum = 0;
200 int eaccum = 0;
201
202 // if we recurse too deep, skip all remaining digits
203 // to avoid a stack overflow attack
204 if (expect_false (--maxdepth <= 0))
205 while (((U8)*s - '0') < 10)
206 ++s;
207
208 for (;;)
209 {
210 U8 dig = (U8)*s - '0';
211
212 if (expect_false (dig >= 10))
213 {
214 if (dig == (U8)((U8)'.' - (U8)'0'))
215 {
216 ++s;
217 json_atof_scan1 (s, accum, expo, 1, maxdepth);
218 }
219 else if ((dig | ' ') == 'e' - '0')
220 {
221 int exp2 = 0;
222 int neg = 0;
223
224 ++s;
225
226 if (*s == '-')
227 {
228 ++s;
229 neg = 1;
230 }
231 else if (*s == '+')
232 ++s;
233
234 while ((dig = (U8)*s - '0') < 10)
235 exp2 = exp2 * 10 + *s++ - '0';
236
237 *expo += neg ? -exp2 : exp2;
238 }
239
240 break;
241 }
242
243 ++s;
244
245 uaccum = uaccum * 10 + dig;
246 ++eaccum;
247
248 // if we have too many digits, then recurse for more
249 // we actually do this for rather few digits
250 if (uaccum >= (UV_MAX - 9) / 10)
251 {
252 if (postdp) *expo -= eaccum;
253 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
254 if (postdp) *expo += eaccum;
255
256 break;
257 }
258 }
259
260 // this relies greatly on the quality of the pow ()
261 // implementation of the platform, but a good
262 // implementation is hard to beat.
263 if (postdp) *expo -= eaccum;
264 *accum += uaccum * Perl_pow (10., *expo);
265 *expo += eaccum;
266}
267
268static NV
269json_atof (const char *s)
270{
271 NV accum = 0.;
272 int expo = 0;
273 int neg = 0;
274
275 if (*s == '-')
276 {
277 ++s;
278 neg = 1;
279 }
280
281 // a recursion depth of ten gives us >>500 bits
282 json_atof_scan1 (s, &accum, &expo, 0, 10);
283
284 return neg ? -accum : accum;
285}
189///////////////////////////////////////////////////////////////////////////// 286/////////////////////////////////////////////////////////////////////////////
190// encoder 287// encoder
191 288
192// structure used for encoding JSON 289// structure used for encoding JSON
193typedef struct 290typedef struct
… …
1116 } 1213 }
1117 1214
1118 len -= *start == '-' ? 1 : 0; 1215 len -= *start == '-' ? 1 : 0;
1119 1216
1120 // does not fit into IV or UV, try NV 1217 // does not fit into IV or UV, try NV
1121 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1218 if (len <= NV_DIG)
1122 #if defined (LDBL_DIG)
1123 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1124 #endif
1125 )
1126 // fits into NV without loss of precision 1219 // fits into NV without loss of precision
1127 return newSVnv (Atof (start)); 1220 return newSVnv (json_atof (start));
1128 1221
1129 // everything else fails, convert it to a string 1222 // everything else fails, convert it to a string
1130 return newSVpvn (start, dec->cur - start); 1223 return newSVpvn (start, dec->cur - start);
1131 } 1224 }
1132 1225
1133 // loss of precision here 1226 // loss of precision here
1134 return newSVnv (Atof (start)); 1227 return newSVnv (json_atof (start));
1135 1228
1136fail: 1229fail:
1137 return 0; 1230 return 0;
1138} 1231}
1139 1232
… …
1525static void 1618static void
1526incr_parse (JSON *self) 1619incr_parse (JSON *self)
1527{ 1620{
1528 const char *p = SvPVX (self->incr_text) + self->incr_pos; 1621 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1529 1622
1623 // the state machine here is a bit convoluted and could be simplified a lot
1624 // but this would make it slower, so...
1625
1530 for (;;) 1626 for (;;)
1531 { 1627 {
1532 //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 1628 //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
1533 switch (self->incr_mode) 1629 switch (self->incr_mode)
1534 { 1630 {
1535 // only used for intiial whitespace skipping 1631 // only used for initial whitespace skipping
1536 case INCR_M_WS: 1632 case INCR_M_WS:
1537 for (;;) 1633 for (;;)
1538 { 1634 {
1539 if (*p > 0x20) 1635 if (*p > 0x20)
1540 { 1636 {
1637 if (*p == '#')
1638 {
1639 self->incr_mode = INCR_M_C0;
1640 goto incr_m_c;
1641 }
1642 else
1643 {
1541 self->incr_mode = INCR_M_JSON; 1644 self->incr_mode = INCR_M_JSON;
1542 goto incr_m_json; 1645 goto incr_m_json;
1646 }
1543 } 1647 }
1544 else if (!*p) 1648 else if (!*p)
1545 goto interrupt; 1649 goto interrupt;
1546 1650
1547 ++p; 1651 ++p;
… …
1553 goto interrupt; 1657 goto interrupt;
1554 1658
1555 ++p; 1659 ++p;
1556 self->incr_mode = INCR_M_STR; 1660 self->incr_mode = INCR_M_STR;
1557 goto incr_m_str; 1661 goto incr_m_str;
1662
1663 // inside #-style comments
1664 case INCR_M_C0:
1665 case INCR_M_C1:
1666 incr_m_c:
1667 for (;;)
1668 {
1669 if (*p == '\n')
1670 {
1671 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1672 break;
1673 }
1674 else if (!*p)
1675 goto interrupt;
1676
1677 ++p;
1678 }
1679
1680 break;
1558 1681
1559 // inside a string 1682 // inside a string
1560 case INCR_M_STR: 1683 case INCR_M_STR:
1561 incr_m_str: 1684 incr_m_str:
1562 for (;;) 1685 for (;;)
… …
1621 1744
1622 case ']': 1745 case ']':
1623 case '}': 1746 case '}':
1624 if (--self->incr_nest <= 0) 1747 if (--self->incr_nest <= 0)
1625 goto interrupt; 1748 goto interrupt;
1749 break;
1750
1751 case '#':
1752 self->incr_mode = INCR_M_C1;
1753 goto incr_m_c;
1626 } 1754 }
1627 } 1755 }
1628 } 1756 }
1629 1757
1630 modechange: 1758 modechange:
1631 ; 1759 ;
1632 } 1760 }
1633 1761
1634interrupt: 1762interrupt:
1635 self->incr_pos = p - SvPVX (self->incr_text); 1763 self->incr_pos = p - SvPVX (self->incr_text);
1764 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1636 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D 1765 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1637} 1766}
1638 1767
1639///////////////////////////////////////////////////////////////////////////// 1768/////////////////////////////////////////////////////////////////////////////
1640// XS interface functions 1769// XS interface functions
… …
1924 json_init (&json); 2053 json_init (&json);
1925 json.flags |= ix; 2054 json.flags |= ix;
1926 XPUSHs (decode_json (jsonstr, &json, 0)); 2055 XPUSHs (decode_json (jsonstr, &json, 0));
1927} 2056}
1928 2057
1929

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines