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 1.91 by root, Mon Sep 22 07:26:26 2008 UTC vs.
Revision 1.105 by root, Tue Jan 19 01:07:27 2010 UTC

12#if defined(__BORLANDC__) || defined(_MSC_VER) 12#if defined(__BORLANDC__) || defined(_MSC_VER)
13# define snprintf _snprintf // C compilers have this in stdio.h 13# define snprintf _snprintf // C compilers have this in stdio.h
14#endif 14#endif
15 15
16// 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
17// guarentees, 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
175 *s++ = 0x80 | ( ch & 0x3f); 178 *s++ = 0x80 | ( ch & 0x3f);
176 179
177 return s; 180 return s;
178} 181}
179 182
183// convert offset pointer to character index, sv must be string
184static STRLEN
185ptr_to_index (SV *sv, char *offset)
186{
187 return SvUTF8 (sv)
188 ? utf8_distance (offset, SvPVX (sv))
189 : offset - SvPVX (sv);
190}
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)
198{
199 UV uaccum = 0;
200 int eaccum = 0;
201
202 for (;;)
203 {
204 U8 dig = (U8)*s - '0';
205
206 if (expect_false (dig >= 10))
207 {
208 if (dig == (U8)((U8)'.' - (U8)'0'))
209 {
210 ++s;
211 json_atof_scan1 (s, accum, expo, 1);
212 }
213 else if ((dig | ' ') == 'e' - '0')
214 {
215 int exp2 = 0;
216 int neg = 0;
217
218 ++s;
219
220 if (*s == '-')
221 {
222 ++s;
223 neg = 1;
224 }
225 else if (*s == '+')
226 ++s;
227
228 while ((dig = (U8)*s - '0') < 10)
229 exp2 = exp2 * 10 + *s++ - '0';
230
231 *expo += neg ? -exp2 : exp2;
232 }
233
234 break;
235 }
236
237 ++s;
238
239 uaccum = uaccum * 10 + dig;
240 ++eaccum;
241
242 // if we have too many digits, then recurse for more
243 // we actually do this for rather few digits
244 if (uaccum >= (UV_MAX - 9) / 10)
245 {
246 if (postdp) *expo -= eaccum;
247 json_atof_scan1 (s, accum, expo, postdp);
248 if (postdp) *expo += eaccum;
249
250 break;
251 }
252 }
253
254 // this relies greatly on the quality of the pow ()
255 // implementation of the platform, but a good
256 // implementation is hard to beat.
257 if (postdp) *expo -= eaccum;
258 *accum += uaccum * Perl_pow (10., *expo);
259 *expo += eaccum;
260}
261
262static NV
263json_atof (const char *s)
264{
265 NV accum = 0.;
266 int expo = 0;
267 int neg = 0;
268
269 if (*s == '-')
270 {
271 ++s;
272 neg = 1;
273 }
274
275 json_atof_scan1 (s, &accum, &expo, 0);
276
277 return neg ? -accum : accum;
278}
180///////////////////////////////////////////////////////////////////////////// 279/////////////////////////////////////////////////////////////////////////////
181// encoder 280// encoder
182 281
183// structure used for encoding JSON 282// structure used for encoding JSON
184typedef struct 283typedef struct
194INLINE void 293INLINE void
195need (enc_t *enc, STRLEN len) 294need (enc_t *enc, STRLEN len)
196{ 295{
197 if (expect_false (enc->cur + len >= enc->end)) 296 if (expect_false (enc->cur + len >= enc->end))
198 { 297 {
199 STRLEN cur = enc->cur - SvPVX (enc->sv); 298 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
200 SvGROW (enc->sv, cur + len + 1); 299 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
201 enc->cur = SvPVX (enc->sv) + cur; 300 enc->cur = SvPVX (enc->sv) + cur;
202 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 301 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
203 } 302 }
204} 303}
205 304
460 559
461 encode_ch (enc, '{'); 560 encode_ch (enc, '{');
462 561
463 // for canonical output we have to sort by keys first 562 // for canonical output we have to sort by keys first
464 // actually, this is mostly due to the stupid so-called 563 // actually, this is mostly due to the stupid so-called
465 // security workaround added somewhere in 5.8.x. 564 // security workaround added somewhere in 5.8.x
466 // that randomises hash orderings 565 // that randomises hash orderings
467 if (enc->json.flags & F_CANONICAL) 566 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
468 { 567 {
469 int count = hv_iterinit (hv); 568 int count = hv_iterinit (hv);
470 569
471 if (SvMAGICAL (hv)) 570 if (SvMAGICAL (hv))
472 { 571 {
750 : enc.json.flags & F_LATIN1 ? 0x000100UL 849 : enc.json.flags & F_LATIN1 ? 0x000100UL
751 : 0x110000UL; 850 : 0x110000UL;
752 851
753 SvPOK_only (enc.sv); 852 SvPOK_only (enc.sv);
754 encode_sv (&enc, scalar); 853 encode_sv (&enc, scalar);
854 encode_nl (&enc);
755 855
756 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 856 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
757 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 857 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
758 858
759 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 859 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
939 else if (expect_true (ch >= 0x20 && ch < 0x80)) 1039 else if (expect_true (ch >= 0x20 && ch < 0x80))
940 *cur++ = ch; 1040 *cur++ = ch;
941 else if (ch >= 0x80) 1041 else if (ch >= 0x80)
942 { 1042 {
943 STRLEN clen; 1043 STRLEN clen;
944 UV uch;
945 1044
946 --dec_cur; 1045 --dec_cur;
947 1046
948 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1047 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
949 if (clen == (STRLEN)-1) 1048 if (clen == (STRLEN)-1)
950 ERR ("malformed UTF-8 character in JSON string"); 1049 ERR ("malformed UTF-8 character in JSON string");
951 1050
952 do 1051 do
953 *cur++ = *dec_cur++; 1052 *cur++ = *dec_cur++;
970 { 1069 {
971 STRLEN len = cur - buf; 1070 STRLEN len = cur - buf;
972 1071
973 if (sv) 1072 if (sv)
974 { 1073 {
975 SvGROW (sv, SvCUR (sv) + len + 1); 1074 STRLEN cur = SvCUR (sv);
1075
1076 if (SvLEN (sv) <= cur + len)
1077 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1078
976 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1079 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
977 SvCUR_set (sv, SvCUR (sv) + len); 1080 SvCUR_set (sv, SvCUR (sv) + len);
978 } 1081 }
979 else 1082 else
980 sv = newSVpvn (buf, len); 1083 sv = newSVpvn (buf, len);
1071 1174
1072 // special case the rather common 1..5-digit-int case 1175 // special case the rather common 1..5-digit-int case
1073 if (*start == '-') 1176 if (*start == '-')
1074 switch (len) 1177 switch (len)
1075 { 1178 {
1076 case 2: return newSViv (-( start [1] - '0' * 1)); 1179 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1077 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1180 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1078 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1181 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1079 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1182 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1080 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); 1183 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1081 } 1184 }
1082 else 1185 else
1083 switch (len) 1186 switch (len)
1084 { 1187 {
1085 case 1: return newSViv ( start [0] - '0' * 1); 1188 case 1: return newSViv ( start [0] - '0' * 1);
1086 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1189 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1087 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1190 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1088 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1191 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1089 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); 1192 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1090 } 1193 }
1091 1194
1092 { 1195 {
1093 UV uv; 1196 UV uv;
1094 int numtype = grok_number (start, len, &uv); 1197 int numtype = grok_number (start, len, &uv);
1103 } 1206 }
1104 1207
1105 len -= *start == '-' ? 1 : 0; 1208 len -= *start == '-' ? 1 : 0;
1106 1209
1107 // does not fit into IV or UV, try NV 1210 // does not fit into IV or UV, try NV
1108 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1211 if (len <= NV_DIG)
1109 #if defined (LDBL_DIG)
1110 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1111 #endif
1112 )
1113 // fits into NV without loss of precision 1212 // fits into NV without loss of precision
1114 return newSVnv (Atof (start)); 1213 return newSVnv (json_atof (start));
1115 1214
1116 // everything else fails, convert it to a string 1215 // everything else fails, convert it to a string
1117 return newSVpvn (start, dec->cur - start); 1216 return newSVpvn (start, dec->cur - start);
1118 } 1217 }
1119 1218
1120 // loss of precision here 1219 // loss of precision here
1121 return newSVnv (Atof (start)); 1220 return newSVnv (json_atof (start));
1122 1221
1123fail: 1222fail:
1124 return 0; 1223 return 0;
1125} 1224}
1126 1225
1404fail: 1503fail:
1405 return 0; 1504 return 0;
1406} 1505}
1407 1506
1408static SV * 1507static SV *
1409decode_json (SV *string, JSON *json, STRLEN *offset_return) 1508decode_json (SV *string, JSON *json, char **offset_return)
1410{ 1509{
1411 dec_t dec; 1510 dec_t dec;
1412 STRLEN offset;
1413 SV *sv; 1511 SV *sv;
1414 1512
1513 /* work around bugs in 5.10 where manipulating magic values
1514 * will perl ignore the magic in subsequent accesses
1515 */
1415 SvGETMAGIC (string); 1516 /*SvGETMAGIC (string);*/
1517 if (SvMAGICAL (string))
1518 string = sv_2mortal (newSVsv (string));
1519
1416 SvUPGRADE (string, SVt_PV); 1520 SvUPGRADE (string, SVt_PV);
1417 1521
1418 /* work around a bug in perl 5.10, which causes SvCUR to fail an 1522 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1419 * assertion with -DDEBUGGING, although SvCUR is documented to 1523 * assertion with -DDEBUGGING, although SvCUR is documented to
1420 * return the xpv_cur field which certainly exists after upgrading. 1524 * return the xpv_cur field which certainly exists after upgrading.
1423 * and hope for the best. 1527 * and hope for the best.
1424 * Damnit, SvPV_nolen still trips over yet another assertion. This 1528 * Damnit, SvPV_nolen still trips over yet another assertion. This
1425 * assertion business is seriously broken, try yet another workaround 1529 * assertion business is seriously broken, try yet another workaround
1426 * for the broken -DDEBUGGING. 1530 * for the broken -DDEBUGGING.
1427 */ 1531 */
1532 {
1428#ifdef DEBUGGING 1533#ifdef DEBUGGING
1429 offset = SvOK (string) ? sv_len (string) : 0; 1534 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1430#else 1535#else
1431 offset = SvCUR (string); 1536 STRLEN offset = SvCUR (string);
1432#endif 1537#endif
1433 1538
1434 if (offset > json->max_size && json->max_size) 1539 if (offset > json->max_size && json->max_size)
1435 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1540 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1436 (unsigned long)SvCUR (string), (unsigned long)json->max_size); 1541 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1542 }
1437 1543
1438 if (json->flags & F_UTF8) 1544 if (json->flags & F_UTF8)
1439 sv_utf8_downgrade (string, 0); 1545 sv_utf8_downgrade (string, 0);
1440 else 1546 else
1441 sv_utf8_upgrade (string); 1547 sv_utf8_upgrade (string);
1453 1559
1454 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1560 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1455 1561
1456 decode_ws (&dec); 1562 decode_ws (&dec);
1457 sv = decode_sv (&dec); 1563 sv = decode_sv (&dec);
1564
1565 if (offset_return)
1566 *offset_return = dec.cur;
1458 1567
1459 if (!(offset_return || !sv)) 1568 if (!(offset_return || !sv))
1460 { 1569 {
1461 // check for trailing garbage 1570 // check for trailing garbage
1462 decode_ws (&dec); 1571 decode_ws (&dec);
1465 { 1574 {
1466 dec.err = "garbage after JSON object"; 1575 dec.err = "garbage after JSON object";
1467 SvREFCNT_dec (sv); 1576 SvREFCNT_dec (sv);
1468 sv = 0; 1577 sv = 0;
1469 } 1578 }
1470 }
1471
1472 if (offset_return || !sv)
1473 {
1474 offset = dec.json.flags & F_UTF8
1475 ? dec.cur - SvPVX (string)
1476 : utf8_distance (dec.cur, SvPVX (string));
1477
1478 if (offset_return)
1479 *offset_return = offset;
1480 } 1579 }
1481 1580
1482 if (!sv) 1581 if (!sv)
1483 { 1582 {
1484 SV *uni = sv_newmortal (); 1583 SV *uni = sv_newmortal ();
1490 SAVEVPTR (PL_curcop); 1589 SAVEVPTR (PL_curcop);
1491 PL_curcop = &cop; 1590 PL_curcop = &cop;
1492 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1591 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1493 LEAVE; 1592 LEAVE;
1494 1593
1495 croak ("%s, at character offset %d [\"%s\"]", 1594 croak ("%s, at character offset %d (before \"%s\")",
1496 dec.err, 1595 dec.err,
1497 (int)offset, 1596 ptr_to_index (string, dec.cur),
1498 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1597 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1499 } 1598 }
1500 1599
1501 sv = sv_2mortal (sv); 1600 sv = sv_2mortal (sv);
1502 1601
1511 1610
1512static void 1611static void
1513incr_parse (JSON *self) 1612incr_parse (JSON *self)
1514{ 1613{
1515 const char *p = SvPVX (self->incr_text) + self->incr_pos; 1614 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1615
1616 // the state machine here is a bit convoluted and could be simplified a lot
1617 // but this would make it slower, so...
1516 1618
1517 for (;;) 1619 for (;;)
1518 { 1620 {
1519 //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 1621 //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
1520 switch (self->incr_mode) 1622 switch (self->incr_mode)
1521 { 1623 {
1522 // only used for intiial whitespace skipping 1624 // only used for initial whitespace skipping
1523 case INCR_M_WS: 1625 case INCR_M_WS:
1524 for (;;) 1626 for (;;)
1525 { 1627 {
1526 if (*p > 0x20) 1628 if (*p > 0x20)
1527 { 1629 {
1630 if (*p == '#')
1631 {
1632 self->incr_mode = INCR_M_C0;
1633 goto incr_m_c;
1634 }
1635 else
1636 {
1528 self->incr_mode = INCR_M_JSON; 1637 self->incr_mode = INCR_M_JSON;
1529 goto incr_m_json; 1638 goto incr_m_json;
1639 }
1530 } 1640 }
1531 else if (!*p) 1641 else if (!*p)
1532 goto interrupt; 1642 goto interrupt;
1533 1643
1534 ++p; 1644 ++p;
1540 goto interrupt; 1650 goto interrupt;
1541 1651
1542 ++p; 1652 ++p;
1543 self->incr_mode = INCR_M_STR; 1653 self->incr_mode = INCR_M_STR;
1544 goto incr_m_str; 1654 goto incr_m_str;
1655
1656 // inside #-style comments
1657 case INCR_M_C0:
1658 case INCR_M_C1:
1659 incr_m_c:
1660 for (;;)
1661 {
1662 if (*p == '\n')
1663 {
1664 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1665 break;
1666 }
1667 else if (!*p)
1668 goto interrupt;
1669
1670 ++p;
1671 }
1672
1673 break;
1545 1674
1546 // inside a string 1675 // inside a string
1547 case INCR_M_STR: 1676 case INCR_M_STR:
1548 incr_m_str: 1677 incr_m_str:
1549 for (;;) 1678 for (;;)
1608 1737
1609 case ']': 1738 case ']':
1610 case '}': 1739 case '}':
1611 if (--self->incr_nest <= 0) 1740 if (--self->incr_nest <= 0)
1612 goto interrupt; 1741 goto interrupt;
1742 break;
1743
1744 case '#':
1745 self->incr_mode = INCR_M_C1;
1746 goto incr_m_c;
1613 } 1747 }
1614 } 1748 }
1615 } 1749 }
1616 1750
1617 modechange: 1751 modechange:
1618 ; 1752 ;
1619 } 1753 }
1620 1754
1621interrupt: 1755interrupt:
1622 self->incr_pos = p - SvPVX (self->incr_text); 1756 self->incr_pos = p - SvPVX (self->incr_text);
1757 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1623 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D 1758 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1624} 1759}
1625 1760
1626///////////////////////////////////////////////////////////////////////////// 1761/////////////////////////////////////////////////////////////////////////////
1627// XS interface functions 1762// XS interface functions
1642 json_stash = gv_stashpv ("JSON::XS" , 1); 1777 json_stash = gv_stashpv ("JSON::XS" , 1);
1643 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1778 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1644 1779
1645 json_true = get_bool ("JSON::XS::true"); 1780 json_true = get_bool ("JSON::XS::true");
1646 json_false = get_bool ("JSON::XS::false"); 1781 json_false = get_bool ("JSON::XS::false");
1782
1783 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1647} 1784}
1648 1785
1649PROTOTYPES: DISABLE 1786PROTOTYPES: DISABLE
1650 1787
1651void CLONE (...) 1788void CLONE (...)
1771 XPUSHs (decode_json (jsonstr, self, 0)); 1908 XPUSHs (decode_json (jsonstr, self, 0));
1772 1909
1773void decode_prefix (JSON *self, SV *jsonstr) 1910void decode_prefix (JSON *self, SV *jsonstr)
1774 PPCODE: 1911 PPCODE:
1775{ 1912{
1776 STRLEN offset; 1913 char *offset;
1777 EXTEND (SP, 2); 1914 EXTEND (SP, 2);
1778 PUSHs (decode_json (jsonstr, self, &offset)); 1915 PUSHs (decode_json (jsonstr, self, &offset));
1779 PUSHs (sv_2mortal (newSVuv (offset))); 1916 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1780} 1917}
1781 1918
1782void incr_parse (JSON *self, SV *jsonstr = 0) 1919void incr_parse (JSON *self, SV *jsonstr = 0)
1783 PPCODE: 1920 PPCODE:
1784{ 1921{
1786 self->incr_text = newSVpvn ("", 0); 1923 self->incr_text = newSVpvn ("", 0);
1787 1924
1788 // append data, if any 1925 // append data, if any
1789 if (jsonstr) 1926 if (jsonstr)
1790 { 1927 {
1791 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) 1928 if (SvUTF8 (jsonstr))
1792 { 1929 {
1930 if (!SvUTF8 (self->incr_text))
1931 {
1793 /* utf-8-ness differs, need to upgrade */ 1932 /* utf-8-ness differs, need to upgrade */
1794 sv_utf8_upgrade (self->incr_text); 1933 sv_utf8_upgrade (self->incr_text);
1795 1934
1796 if (self->incr_pos) 1935 if (self->incr_pos)
1797 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) 1936 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1798 - (U8 *)SvPVX (self->incr_text); 1937 - (U8 *)SvPVX (self->incr_text);
1938 }
1799 } 1939 }
1940 else if (SvUTF8 (self->incr_text))
1941 sv_utf8_upgrade (jsonstr);
1800 1942
1801 { 1943 {
1802 STRLEN len; 1944 STRLEN len;
1803 const char *str = SvPV (jsonstr, len); 1945 const char *str = SvPV (jsonstr, len);
1804 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); 1946 STRLEN cur = SvCUR (self->incr_text);
1947
1948 if (SvLEN (self->incr_text) <= cur + len)
1949 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1950
1805 Move (str, SvEND (self->incr_text), len, char); 1951 Move (str, SvEND (self->incr_text), len, char);
1806 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 1952 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1807 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there 1953 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1808 } 1954 }
1809 } 1955 }
1810 1956
1811 if (GIMME_V != G_VOID) 1957 if (GIMME_V != G_VOID)
1812 do 1958 do
1813 { 1959 {
1814 STRLEN offset; 1960 char *offset;
1815 1961
1816 if (!INCR_DONE (self)) 1962 if (!INCR_DONE (self))
1817 { 1963 {
1818 incr_parse (self); 1964 incr_parse (self);
1819 1965
1825 break; 1971 break;
1826 } 1972 }
1827 1973
1828 XPUSHs (decode_json (self->incr_text, self, &offset)); 1974 XPUSHs (decode_json (self->incr_text, self, &offset));
1829 1975
1830 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1831 self->incr_pos -= offset; 1976 self->incr_pos -= offset - SvPVX (self->incr_text);
1832 self->incr_nest = 0; 1977 self->incr_nest = 0;
1833 self->incr_mode = 0; 1978 self->incr_mode = 0;
1979
1980 sv_chop (self->incr_text, offset);
1834 } 1981 }
1835 while (GIMME_V == G_ARRAY); 1982 while (GIMME_V == G_ARRAY);
1836} 1983}
1837 1984
1838SV *incr_text (JSON *self) 1985SV *incr_text (JSON *self)
1899 json_init (&json); 2046 json_init (&json);
1900 json.flags |= ix; 2047 json.flags |= ix;
1901 XPUSHs (decode_json (jsonstr, &json, 0)); 2048 XPUSHs (decode_json (jsonstr, &json, 0));
1902} 2049}
1903 2050
1904

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines