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 1.74 by root, Wed Mar 19 15:17:53 2008 UTC vs.
Revision 1.83 by root, Wed Mar 26 22:54:38 2008 UTC

… …
4 4
5#include <assert.h> 5#include <assert.h>
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 <float.h> 10#include <float.h>
10 11
11#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
12# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
13#endif 14#endif
… …
15// some old perls do not have this, try to make it work, no 16// some old perls do not have this, try to make it work, no
16// guarentees, though. if it breaks, you get to keep the pieces. 17// guarentees, though. if it breaks, you get to keep the pieces.
17#ifndef UTF8_MAXBYTES 18#ifndef UTF8_MAXBYTES
18# define UTF8_MAXBYTES 13 19# define UTF8_MAXBYTES 13
19#endif 20#endif
21
22#define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2)
20 23
21#define F_ASCII 0x00000001UL 24#define F_ASCII 0x00000001UL
22#define F_LATIN1 0x00000002UL 25#define F_LATIN1 0x00000002UL
23#define F_UTF8 0x00000004UL 26#define F_UTF8 0x00000004UL
24#define F_INDENT 0x00000008UL 27#define F_INDENT 0x00000008UL
… …
75#endif 78#endif
76 79
77static HV *json_stash, *json_boolean_stash; // JSON::XS:: 80static HV *json_stash, *json_boolean_stash; // JSON::XS::
78static SV *json_true, *json_false; 81static SV *json_true, *json_false;
79 82
83enum {
84 INCR_M_WS = 0, // initial whitespace skipping, must be 0
85 INCR_M_STR, // inside string
86 INCR_M_BS, // inside backslash
87 INCR_M_JSON // outside anything, count nesting
88};
89
90#define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON)
91
80typedef struct { 92typedef struct {
81 U32 flags; 93 U32 flags;
82 SV *cb_object; 94 SV *cb_object;
83 HV *cb_sk_object; 95 HV *cb_sk_object;
96
97 // for the incremental parser
98 SV *incr_text; // the source text so far
99 STRLEN incr_pos; // the current offset into the text
100 int incr_nest; // {[]}-nesting level
101 int incr_mode;
84} JSON; 102} JSON;
85 103
86///////////////////////////////////////////////////////////////////////////// 104/////////////////////////////////////////////////////////////////////////////
87// utility functions 105// utility functions
106
107INLINE SV *
108get_bool (const char *name)
109{
110 SV *sv = get_sv (name, 1);
111
112 SvREADONLY_on (sv);
113 SvREADONLY_on (SvRV (sv));
114
115 return sv;
116}
88 117
89INLINE void 118INLINE void
90shrink (SV *sv) 119shrink (SV *sv)
91{ 120{
92 sv_utf8_downgrade (sv, 1); 121 sv_utf8_downgrade (sv, 1);
… …
420 449
421static void 450static void
422encode_hv (enc_t *enc, HV *hv) 451encode_hv (enc_t *enc, HV *hv)
423{ 452{
424 HE *he; 453 HE *he;
425 int count;
426 454
427 if (enc->indent >= enc->maxdepth) 455 if (enc->indent >= enc->maxdepth)
428 croak ("data structure too deep (hit recursion limit)"); 456 croak ("data structure too deep (hit recursion limit)");
429 457
430 encode_ch (enc, '{'); 458 encode_ch (enc, '{');
… …
648 else if (SvIOKp (sv)) 676 else if (SvIOKp (sv))
649 { 677 {
650 // we assume we can always read an IV as a UV and vice versa 678 // we assume we can always read an IV as a UV and vice versa
651 // we assume two's complement 679 // we assume two's complement
652 // we assume no aliasing issues in the union 680 // we assume no aliasing issues in the union
653 if (SvIsUV (sv) ? SvUVX (sv) > 59000 681 if (SvIsUV (sv) ? SvUVX (sv) <= 59000
654 : SvIVX (sv) > 59000 || SvIVX (sv) < -59000) 682 : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000)
655 {
656 // large integer, use the (rather slow) snprintf way.
657 need (enc, sizeof (UV) * 5 / 2 + 1); // CHAR_BIT is at least 8
658 enc->cur +=
659 SvIsUV(sv)
660 ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv))
661 : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv));
662 }
663 else
664 { 683 {
665 // optimise the "small number case" 684 // optimise the "small number case"
666 // code will likely be branchless and use only a single multiplication 685 // code will likely be branchless and use only a single multiplication
667 // works for numbers up to 59074 686 // works for numbers up to 59074
668 I32 i = SvIVX (sv); 687 I32 i = SvIVX (sv);
… …
679 698
680 // now output digit by digit, each time masking out the integer part 699 // now output digit by digit, each time masking out the integer part
681 // and multiplying by 5 while moving the decimal point one to the right, 700 // and multiplying by 5 while moving the decimal point one to the right,
682 // resulting in a net multiplication by 10. 701 // resulting in a net multiplication by 10.
683 // we always write the digit to memory but conditionally increment 702 // we always write the digit to memory but conditionally increment
684 // the pointer, to ease the usage of conditional move instructions. 703 // the pointer, to enable the use of conditional move instructions.
685 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; 704 digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5;
686 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; 705 digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5;
687 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; 706 digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5;
688 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; 707 digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5;
689 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' 708 digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0'
709 }
710 else
711 {
712 // large integer, use the (rather slow) snprintf way.
713 need (enc, IVUV_MAXCHARS);
714 enc->cur +=
715 SvIsUV(sv)
716 ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv))
717 : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv));
690 } 718 }
691 } 719 }
692 else if (SvROK (sv)) 720 else if (SvROK (sv))
693 encode_rv (enc, SvRV (sv)); 721 encode_rv (enc, SvRV (sv));
694 else if (!SvOK (sv)) 722 else if (!SvOK (sv))
… …
1168 char *p = dec->cur; 1196 char *p = dec->cur;
1169 char *e = p + 24; // only try up to 24 bytes 1197 char *e = p + 24; // only try up to 24 bytes
1170 1198
1171 for (;;) 1199 for (;;)
1172 { 1200 {
1173 // the >= 0x80 is true on most architectures 1201 // the >= 0x80 is false on most architectures
1174 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') 1202 if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\')
1175 { 1203 {
1176 // slow path, back up and use decode_str 1204 // slow path, back up and use decode_str
1177 SV *key = decode_str (dec); 1205 SV *key = decode_str (dec);
1178 if (!key) 1206 if (!key)
… …
1326 case 't': 1354 case 't':
1327 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) 1355 if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4))
1328 { 1356 {
1329 dec->cur += 4; 1357 dec->cur += 4;
1330#if JSON_SLOW 1358#if JSON_SLOW
1331 json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); 1359 json_true = get_bool ("JSON::XS::true");
1332#endif 1360#endif
1333 return SvREFCNT_inc (json_true); 1361 return newSVsv (json_true);
1334 } 1362 }
1335 else 1363 else
1336 ERR ("'true' expected"); 1364 ERR ("'true' expected");
1337 1365
1338 break; 1366 break;
… …
1340 case 'f': 1368 case 'f':
1341 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) 1369 if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5))
1342 { 1370 {
1343 dec->cur += 5; 1371 dec->cur += 5;
1344#if JSON_SLOW 1372#if JSON_SLOW
1345 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1373 json_false = get_bool ("JSON::XS::false");
1346#endif 1374#endif
1347 return SvREFCNT_inc (json_false); 1375 return newSVsv (json_false);
1348 } 1376 }
1349 else 1377 else
1350 ERR ("'false' expected"); 1378 ERR ("'false' expected");
1351 1379
1352 break; 1380 break;
… …
1370fail: 1398fail:
1371 return 0; 1399 return 0;
1372} 1400}
1373 1401
1374static SV * 1402static SV *
1375decode_json (SV *string, JSON *json, UV *offset_return) 1403decode_json (SV *string, JSON *json, STRLEN *offset_return)
1376{ 1404{
1377 dec_t dec; 1405 dec_t dec;
1378 UV offset; 1406 STRLEN offset;
1379 SV *sv; 1407 SV *sv;
1380 1408
1381 SvGETMAGIC (string); 1409 SvGETMAGIC (string);
1382 SvUPGRADE (string, SVt_PV); 1410 SvUPGRADE (string, SVt_PV);
1383 1411
… …
1456 1484
1457 return sv; 1485 return sv;
1458} 1486}
1459 1487
1460///////////////////////////////////////////////////////////////////////////// 1488/////////////////////////////////////////////////////////////////////////////
1489// incremental parser
1490
1491static void
1492incr_parse (JSON *self)
1493{
1494 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1495
1496 for (;;)
1497 {
1498 //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
1499 switch (self->incr_mode)
1500 {
1501 // only used for intiial whitespace skipping
1502 case INCR_M_WS:
1503 for (;;)
1504 {
1505 if (*p > 0x20)
1506 {
1507 self->incr_mode = INCR_M_JSON;
1508 goto incr_m_json;
1509 }
1510 else if (!*p)
1511 goto interrupt;
1512
1513 ++p;
1514 }
1515
1516 // skip a single char inside a string (for \\-processing)
1517 case INCR_M_BS:
1518 if (!*p)
1519 goto interrupt;
1520
1521 ++p;
1522 self->incr_mode = INCR_M_STR;
1523 goto incr_m_str;
1524
1525 // inside a string
1526 case INCR_M_STR:
1527 incr_m_str:
1528 for (;;)
1529 {
1530 if (*p == '"')
1531 {
1532 ++p;
1533 self->incr_mode = INCR_M_JSON;
1534
1535 if (!self->incr_nest)
1536 goto interrupt;
1537
1538 goto incr_m_json;
1539 }
1540 else if (*p == '\\')
1541 {
1542 ++p; // "virtually" consumes character after \
1543
1544 if (!*p) // if at end of string we have to switch modes
1545 {
1546 self->incr_mode = INCR_M_BS;
1547 goto interrupt;
1548 }
1549 }
1550 else if (!*p)
1551 goto interrupt;
1552
1553 ++p;
1554 }
1555
1556 // after initial ws, outside string
1557 case INCR_M_JSON:
1558 incr_m_json:
1559 for (;;)
1560 {
1561 switch (*p++)
1562 {
1563 case 0:
1564 --p;
1565 goto interrupt;
1566
1567 case 0x09:
1568 case 0x0a:
1569 case 0x0d:
1570 case 0x20:
1571 if (!self->incr_nest)
1572 {
1573 --p; // do not eat the whitespace, let the next round do it
1574 goto interrupt;
1575 }
1576 break;
1577
1578 case '"':
1579 self->incr_mode = INCR_M_STR;
1580 goto incr_m_str;
1581
1582 case '[':
1583 case '{':
1584 ++self->incr_nest;
1585 break;
1586
1587 case ']':
1588 case '}':
1589 if (!--self->incr_nest)
1590 goto interrupt;
1591 }
1592 }
1593 }
1594
1595 modechange:
1596 ;
1597 }
1598
1599interrupt:
1600 self->incr_pos = p - SvPVX (self->incr_text);
1601 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1602}
1603
1604/////////////////////////////////////////////////////////////////////////////
1461// XS interface functions 1605// XS interface functions
1462 1606
1463MODULE = JSON::XS PACKAGE = JSON::XS 1607MODULE = JSON::XS PACKAGE = JSON::XS
1464 1608
1465BOOT: 1609BOOT:
… …
1474 : -1; 1618 : -1;
1475 1619
1476 json_stash = gv_stashpv ("JSON::XS" , 1); 1620 json_stash = gv_stashpv ("JSON::XS" , 1);
1477 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1621 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1478 1622
1479 json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); 1623 json_true = get_bool ("JSON::XS::true");
1480 json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); 1624 json_false = get_bool ("JSON::XS::false");
1481} 1625}
1482 1626
1483PROTOTYPES: DISABLE 1627PROTOTYPES: DISABLE
1484 1628
1485void CLONE (...) 1629void CLONE (...)
… …
1625 XPUSHs (decode_json (jsonstr, self, 0)); 1769 XPUSHs (decode_json (jsonstr, self, 0));
1626 1770
1627void decode_prefix (JSON *self, SV *jsonstr) 1771void decode_prefix (JSON *self, SV *jsonstr)
1628 PPCODE: 1772 PPCODE:
1629{ 1773{
1630 UV offset; 1774 STRLEN offset;
1631 EXTEND (SP, 2); 1775 EXTEND (SP, 2);
1632 PUSHs (decode_json (jsonstr, self, &offset)); 1776 PUSHs (decode_json (jsonstr, self, &offset));
1633 PUSHs (sv_2mortal (newSVuv (offset))); 1777 PUSHs (sv_2mortal (newSVuv (offset)));
1778}
1779
1780void incr_parse (JSON *self, SV *jsonstr = 0)
1781 PPCODE:
1782{
1783 if (!self->incr_text)
1784 self->incr_text = newSVpvn ("", 0);
1785
1786 // append data, if any
1787 if (jsonstr)
1788 {
1789 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text))
1790 {
1791 /* utf-8-ness differs, need to upgrade */
1792 sv_utf8_upgrade (self->incr_text);
1793
1794 if (self->incr_pos)
1795 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1796 - (U8 *)SvPVX (self->incr_text);
1797 }
1798
1799 {
1800 STRLEN len;
1801 const char *str = SvPV (jsonstr, len);
1802 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1);
1803 Move (str, SvEND (self->incr_text), len, char);
1804 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1805 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1806 }
1807 }
1808
1809 if (GIMME_V != G_VOID)
1810 do
1811 {
1812 STRLEN offset;
1813
1814 if (!INCR_DONE (self))
1815 {
1816 incr_parse (self);
1817 if (!INCR_DONE (self))
1818 break;
1819 }
1820
1821 XPUSHs (decode_json (self->incr_text, self, &offset));
1822
1823 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1824 self->incr_pos -= offset;
1825 self->incr_nest = 0;
1826 self->incr_mode = 0;
1827 }
1828 while (GIMME_V == G_ARRAY);
1829}
1830
1831SV *incr_text (JSON *self)
1832 ATTRS: lvalue
1833 CODE:
1834{
1835 if (self->incr_pos)
1836 croak ("incr_text can not be called when the incremental parser already started parsing");
1837
1838 RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef;
1839}
1840 OUTPUT:
1841 RETVAL
1842
1843void incr_skip (JSON *self)
1844 CODE:
1845{
1846 if (self->incr_pos)
1847 {
1848 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos);
1849 self->incr_pos = 0;
1850 self->incr_nest = 0;
1851 self->incr_mode = 0;
1852 }
1634} 1853}
1635 1854
1636void DESTROY (JSON *self) 1855void DESTROY (JSON *self)
1637 CODE: 1856 CODE:
1638 SvREFCNT_dec (self->cb_sk_object); 1857 SvREFCNT_dec (self->cb_sk_object);
1639 SvREFCNT_dec (self->cb_object); 1858 SvREFCNT_dec (self->cb_object);
1859 SvREFCNT_dec (self->incr_text);
1640 1860
1641PROTOTYPES: ENABLE 1861PROTOTYPES: ENABLE
1642 1862
1643void encode_json (SV *scalar) 1863void encode_json (SV *scalar)
1864 ALIAS:
1865 to_json_ = 0
1866 encode_json = F_UTF8
1644 PPCODE: 1867 PPCODE:
1645{ 1868{
1646 JSON json = { F_DEFAULT | F_UTF8 }; 1869 JSON json = { F_DEFAULT | ix };
1647 XPUSHs (encode_json (scalar, &json)); 1870 XPUSHs (encode_json (scalar, &json));
1648} 1871}
1649 1872
1650void decode_json (SV *jsonstr) 1873void decode_json (SV *jsonstr)
1874 ALIAS:
1875 from_json_ = 0
1876 decode_json = F_UTF8
1651 PPCODE: 1877 PPCODE:
1652{ 1878{
1653 JSON json = { F_DEFAULT | F_UTF8 }; 1879 JSON json = { F_DEFAULT | ix };
1654 XPUSHs (decode_json (jsonstr, &json, 0)); 1880 XPUSHs (decode_json (jsonstr, &json, 0));
1655} 1881}
1656 1882
1883

Diff Legend

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