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 1.90 by root, Sun Jul 20 17:55:19 2008 UTC vs.
Revision 1.109 by root, Tue Mar 8 10:18:50 2011 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
… …
41#define INIT_SIZE 32 // initial scalar size to be allocated 42#define INIT_SIZE 32 // initial scalar size to be allocated
42#define INDENT_STEP 3 // spaces per indentation level 43#define INDENT_STEP 3 // spaces per indentation level
43 44
44#define SHORT_STRING_LEN 16384 // special-case strings of up to this size 45#define SHORT_STRING_LEN 16384 // special-case strings of up to this size
45 46
47#define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8)
48
46#define SB do { 49#define SB do {
47#define SE } while (0) 50#define SE } while (0)
48 51
49#if __GNUC__ >= 3 52#if __GNUC__ >= 3
50# define expect(expr,value) __builtin_expect ((expr), (value)) 53# define expect(expr,value) __builtin_expect ((expr), (value))
… …
76 79
77enum { 80enum {
78 INCR_M_WS = 0, // initial whitespace skipping, must be 0 81 INCR_M_WS = 0, // initial whitespace skipping, must be 0
79 INCR_M_STR, // inside string 82 INCR_M_STR, // inside string
80 INCR_M_BS, // inside backslash 83 INCR_M_BS, // inside backslash
84 INCR_M_C0, // inside comment in initial whitespace sequence
85 INCR_M_C1, // inside comment in other places
81 INCR_M_JSON // outside anything, count nesting 86 INCR_M_JSON // outside anything, count nesting
82}; 87};
83 88
84#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON) 89#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
85 90
… …
175 *s++ = 0x80 | ( ch & 0x3f); 180 *s++ = 0x80 | ( ch & 0x3f);
176 181
177 return s; 182 return s;
178} 183}
179 184
185// convert offset pointer to character index, sv must be string
186static STRLEN
187ptr_to_index (SV *sv, char *offset)
188{
189 return SvUTF8 (sv)
190 ? utf8_distance (offset, SvPVX (sv))
191 : offset - SvPVX (sv);
192}
193
194/////////////////////////////////////////////////////////////////////////////
195// fp hell
196
197// scan a group of digits, and a trailing exponent
198static void
199json_atof_scan1 (const char *s, NV *accum, int *expo, int postdp, int maxdepth)
200{
201 UV uaccum = 0;
202 int eaccum = 0;
203
204 // if we recurse too deep, skip all remaining digits
205 // to avoid a stack overflow attack
206 if (expect_false (--maxdepth <= 0))
207 while (((U8)*s - '0') < 10)
208 ++s;
209
210 for (;;)
211 {
212 U8 dig = (U8)*s - '0';
213
214 if (expect_false (dig >= 10))
215 {
216 if (dig == (U8)((U8)'.' - (U8)'0'))
217 {
218 ++s;
219 json_atof_scan1 (s, accum, expo, 1, maxdepth);
220 }
221 else if ((dig | ' ') == 'e' - '0')
222 {
223 int exp2 = 0;
224 int neg = 0;
225
226 ++s;
227
228 if (*s == '-')
229 {
230 ++s;
231 neg = 1;
232 }
233 else if (*s == '+')
234 ++s;
235
236 while ((dig = (U8)*s - '0') < 10)
237 exp2 = exp2 * 10 + *s++ - '0';
238
239 *expo += neg ? -exp2 : exp2;
240 }
241
242 break;
243 }
244
245 ++s;
246
247 uaccum = uaccum * 10 + dig;
248 ++eaccum;
249
250 // if we have too many digits, then recurse for more
251 // we actually do this for rather few digits
252 if (uaccum >= (UV_MAX - 9) / 10)
253 {
254 if (postdp) *expo -= eaccum;
255 json_atof_scan1 (s, accum, expo, postdp, maxdepth);
256 if (postdp) *expo += eaccum;
257
258 break;
259 }
260 }
261
262 // this relies greatly on the quality of the pow ()
263 // implementation of the platform, but a good
264 // implementation is hard to beat.
265 if (postdp) *expo -= eaccum;
266 *accum += uaccum * Perl_pow (10., *expo);
267 *expo += eaccum;
268}
269
270static NV
271json_atof (const char *s)
272{
273 NV accum = 0.;
274 int expo = 0;
275 int neg = 0;
276
277 if (*s == '-')
278 {
279 ++s;
280 neg = 1;
281 }
282
283 // a recursion depth of ten gives us >>500 bits
284 json_atof_scan1 (s, &accum, &expo, 0, 10);
285
286 return neg ? -accum : accum;
287}
180///////////////////////////////////////////////////////////////////////////// 288/////////////////////////////////////////////////////////////////////////////
181// encoder 289// encoder
182 290
183// structure used for encoding JSON 291// structure used for encoding JSON
184typedef struct 292typedef struct
… …
194INLINE void 302INLINE void
195need (enc_t *enc, STRLEN len) 303need (enc_t *enc, STRLEN len)
196{ 304{
197 if (expect_false (enc->cur + len >= enc->end)) 305 if (expect_false (enc->cur + len >= enc->end))
198 { 306 {
199 STRLEN cur = enc->cur - SvPVX (enc->sv); 307 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
200 SvGROW (enc->sv, cur + len + 1); 308 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
201 enc->cur = SvPVX (enc->sv) + cur; 309 enc->cur = SvPVX (enc->sv) + cur;
202 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 310 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
203 } 311 }
204} 312}
205 313
… …
280 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 388 (int)((uch - 0x10000) % 0x400 + 0xDC00));
281 enc->cur += 12; 389 enc->cur += 12;
282 } 390 }
283 else 391 else
284 { 392 {
285 static char hexdigit [16] = "0123456789abcdef";
286 need (enc, len += 5); 393 need (enc, len += 5);
287 *enc->cur++ = '\\'; 394 *enc->cur++ = '\\';
288 *enc->cur++ = 'u'; 395 *enc->cur++ = 'u';
289 *enc->cur++ = hexdigit [ uch >> 12 ]; 396 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
290 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 397 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
291 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 398 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
292 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 399 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
293 } 400 }
294 401
295 str += clen; 402 str += clen;
296 } 403 }
297 else if (enc->json.flags & F_LATIN1) 404 else if (enc->json.flags & F_LATIN1)
… …
461 568
462 encode_ch (enc, '{'); 569 encode_ch (enc, '{');
463 570
464 // for canonical output we have to sort by keys first 571 // for canonical output we have to sort by keys first
465 // actually, this is mostly due to the stupid so-called 572 // actually, this is mostly due to the stupid so-called
466 // security workaround added somewhere in 5.8.x. 573 // security workaround added somewhere in 5.8.x
467 // that randomises hash orderings 574 // that randomises hash orderings
468 if (enc->json.flags & F_CANONICAL) 575 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
469 { 576 {
470 int count = hv_iterinit (hv); 577 int count = hv_iterinit (hv);
471 578
472 if (SvMAGICAL (hv)) 579 if (SvMAGICAL (hv))
473 { 580 {
… …
751 : enc.json.flags & F_LATIN1 ? 0x000100UL 858 : enc.json.flags & F_LATIN1 ? 0x000100UL
752 : 0x110000UL; 859 : 0x110000UL;
753 860
754 SvPOK_only (enc.sv); 861 SvPOK_only (enc.sv);
755 encode_sv (&enc, scalar); 862 encode_sv (&enc, scalar);
863 encode_nl (&enc);
756 864
757 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 865 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
758 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 866 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
759 867
760 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 868 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
940 else if (expect_true (ch >= 0x20 && ch < 0x80)) 1048 else if (expect_true (ch >= 0x20 && ch < 0x80))
941 *cur++ = ch; 1049 *cur++ = ch;
942 else if (ch >= 0x80) 1050 else if (ch >= 0x80)
943 { 1051 {
944 STRLEN clen; 1052 STRLEN clen;
945 UV uch;
946 1053
947 --dec_cur; 1054 --dec_cur;
948 1055
949 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1056 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
950 if (clen == (STRLEN)-1) 1057 if (clen == (STRLEN)-1)
951 ERR ("malformed UTF-8 character in JSON string"); 1058 ERR ("malformed UTF-8 character in JSON string");
952 1059
953 do 1060 do
954 *cur++ = *dec_cur++; 1061 *cur++ = *dec_cur++;
… …
971 { 1078 {
972 STRLEN len = cur - buf; 1079 STRLEN len = cur - buf;
973 1080
974 if (sv) 1081 if (sv)
975 { 1082 {
976 SvGROW (sv, SvCUR (sv) + len + 1); 1083 STRLEN cur = SvCUR (sv);
1084
1085 if (SvLEN (sv) <= cur + len)
1086 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1087
977 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1088 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
978 SvCUR_set (sv, SvCUR (sv) + len); 1089 SvCUR_set (sv, SvCUR (sv) + len);
979 } 1090 }
980 else 1091 else
981 sv = newSVpvn (buf, len); 1092 sv = newSVpvn (buf, len);
… …
1072 1183
1073 // special case the rather common 1..5-digit-int case 1184 // special case the rather common 1..5-digit-int case
1074 if (*start == '-') 1185 if (*start == '-')
1075 switch (len) 1186 switch (len)
1076 { 1187 {
1077 case 2: return newSViv (-( start [1] - '0' * 1)); 1188 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1078 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1189 case 3: return newSViv (-(IV)( start [1] * 10 + start [2] - '0' * 11));
1079 case 4: return newSViv (-( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111)); 1190 case 4: return newSViv (-(IV)( start [1] * 100 + start [2] * 10 + start [3] - '0' * 111));
1080 case 5: return newSViv (-( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111)); 1191 case 5: return newSViv (-(IV)( start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 1111));
1081 case 6: return newSViv (-(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111)); 1192 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1082 } 1193 }
1083 else 1194 else
1084 switch (len) 1195 switch (len)
1085 { 1196 {
1086 case 1: return newSViv ( start [0] - '0' * 1); 1197 case 1: return newSViv ( start [0] - '0' * 1);
1087 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1198 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11);
1088 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111); 1199 case 3: return newSViv ( start [0] * 100 + start [1] * 10 + start [2] - '0' * 111);
1089 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111); 1200 case 4: return newSViv ( start [0] * 1000 + start [1] * 100 + start [2] * 10 + start [3] - '0' * 1111);
1090 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111); 1201 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1091 } 1202 }
1092 1203
1093 { 1204 {
1094 UV uv; 1205 UV uv;
1095 int numtype = grok_number (start, len, &uv); 1206 int numtype = grok_number (start, len, &uv);
… …
1104 } 1215 }
1105 1216
1106 len -= *start == '-' ? 1 : 0; 1217 len -= *start == '-' ? 1 : 0;
1107 1218
1108 // does not fit into IV or UV, try NV 1219 // does not fit into IV or UV, try NV
1109 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1220 if (len <= NV_DIG)
1110 #if defined (LDBL_DIG)
1111 || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len)
1112 #endif
1113 )
1114 // fits into NV without loss of precision 1221 // fits into NV without loss of precision
1115 return newSVnv (Atof (start)); 1222 return newSVnv (json_atof (start));
1116 1223
1117 // everything else fails, convert it to a string 1224 // everything else fails, convert it to a string
1118 return newSVpvn (start, dec->cur - start); 1225 return newSVpvn (start, dec->cur - start);
1119 } 1226 }
1120 1227
1121 // loss of precision here 1228 // loss of precision here
1122 return newSVnv (Atof (start)); 1229 return newSVnv (json_atof (start));
1123 1230
1124fail: 1231fail:
1125 return 0; 1232 return 0;
1126} 1233}
1127 1234
… …
1296 dSP; 1403 dSP;
1297 int count; 1404 int count;
1298 1405
1299 ENTER; SAVETMPS; PUSHMARK (SP); 1406 ENTER; SAVETMPS; PUSHMARK (SP);
1300 XPUSHs (HeVAL (he)); 1407 XPUSHs (HeVAL (he));
1408 sv_2mortal (sv);
1301 1409
1302 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1410 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1303 1411
1304 if (count == 1) 1412 if (count == 1)
1305 { 1413 {
1306 sv = newSVsv (POPs); 1414 sv = newSVsv (POPs);
1307 FREETMPS; LEAVE; 1415 FREETMPS; LEAVE;
1308 return sv; 1416 return sv;
1309 } 1417 }
1310 1418
1419 SvREFCNT_inc (sv);
1311 FREETMPS; LEAVE; 1420 FREETMPS; LEAVE;
1312 } 1421 }
1313 } 1422 }
1314 1423
1315 if (dec->json.cb_object) 1424 if (dec->json.cb_object)
… …
1405fail: 1514fail:
1406 return 0; 1515 return 0;
1407} 1516}
1408 1517
1409static SV * 1518static SV *
1410decode_json (SV *string, JSON *json, STRLEN *offset_return) 1519decode_json (SV *string, JSON *json, char **offset_return)
1411{ 1520{
1412 dec_t dec; 1521 dec_t dec;
1413 STRLEN offset;
1414 SV *sv; 1522 SV *sv;
1415 1523
1524 /* work around bugs in 5.10 where manipulating magic values
1525 * will perl ignore the magic in subsequent accesses.
1526 * also make a copy of non-PV values, to get them into a clean
1527 * state (SvPV should do that, but it's buggy, see below).
1528 */
1416 SvGETMAGIC (string); 1529 /*SvGETMAGIC (string);*/
1530 if (SvMAGICAL (string) || !SvPOK (string))
1531 string = sv_2mortal (newSVsv (string));
1532
1417 SvUPGRADE (string, SVt_PV); 1533 SvUPGRADE (string, SVt_PV);
1418 1534
1419 /* work around a bug in perl 5.10, which causes SvCUR to fail an 1535 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1420 * assertion with -DDEBUGGING, although SvCUR is documented to 1536 * assertion with -DDEBUGGING, although SvCUR is documented to
1421 * return the xpv_cur field which certainly exists after upgrading. 1537 * return the xpv_cur field which certainly exists after upgrading.
… …
1424 * and hope for the best. 1540 * and hope for the best.
1425 * Damnit, SvPV_nolen still trips over yet another assertion. This 1541 * Damnit, SvPV_nolen still trips over yet another assertion. This
1426 * assertion business is seriously broken, try yet another workaround 1542 * assertion business is seriously broken, try yet another workaround
1427 * for the broken -DDEBUGGING. 1543 * for the broken -DDEBUGGING.
1428 */ 1544 */
1545 {
1429#ifdef DEBUGGING 1546#ifdef DEBUGGING
1430 offset = SvOK (string) ? sv_len (string) : 0; 1547 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1431#else 1548#else
1432 offset = SvCUR (string); 1549 STRLEN offset = SvCUR (string);
1433#endif 1550#endif
1434 1551
1435 if (offset > json->max_size && json->max_size) 1552 if (offset > json->max_size && json->max_size)
1436 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1553 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1437 (unsigned long)SvCUR (string), (unsigned long)json->max_size); 1554 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1555 }
1438 1556
1439 if (json->flags & F_UTF8) 1557 if (DECODE_WANTS_OCTETS (json))
1440 sv_utf8_downgrade (string, 0); 1558 sv_utf8_downgrade (string, 0);
1441 else 1559 else
1442 sv_utf8_upgrade (string); 1560 sv_utf8_upgrade (string);
1443 1561
1444 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1562 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
… …
1454 1572
1455 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1573 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1456 1574
1457 decode_ws (&dec); 1575 decode_ws (&dec);
1458 sv = decode_sv (&dec); 1576 sv = decode_sv (&dec);
1577
1578 if (offset_return)
1579 *offset_return = dec.cur;
1459 1580
1460 if (!(offset_return || !sv)) 1581 if (!(offset_return || !sv))
1461 { 1582 {
1462 // check for trailing garbage 1583 // check for trailing garbage
1463 decode_ws (&dec); 1584 decode_ws (&dec);
… …
1466 { 1587 {
1467 dec.err = "garbage after JSON object"; 1588 dec.err = "garbage after JSON object";
1468 SvREFCNT_dec (sv); 1589 SvREFCNT_dec (sv);
1469 sv = 0; 1590 sv = 0;
1470 } 1591 }
1471 }
1472
1473 if (offset_return || !sv)
1474 {
1475 offset = dec.json.flags & F_UTF8
1476 ? dec.cur - SvPVX (string)
1477 : utf8_distance (dec.cur, SvPVX (string));
1478
1479 if (offset_return)
1480 *offset_return = offset;
1481 } 1592 }
1482 1593
1483 if (!sv) 1594 if (!sv)
1484 { 1595 {
1485 SV *uni = sv_newmortal (); 1596 SV *uni = sv_newmortal ();
… …
1491 SAVEVPTR (PL_curcop); 1602 SAVEVPTR (PL_curcop);
1492 PL_curcop = &cop; 1603 PL_curcop = &cop;
1493 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1604 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1494 LEAVE; 1605 LEAVE;
1495 1606
1496 croak ("%s, at character offset %d [\"%s\"]", 1607 croak ("%s, at character offset %d (before \"%s\")",
1497 dec.err, 1608 dec.err,
1498 (int)offset, 1609 ptr_to_index (string, dec.cur),
1499 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1610 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1500 } 1611 }
1501 1612
1502 sv = sv_2mortal (sv); 1613 sv = sv_2mortal (sv);
1503 1614
… …
1512 1623
1513static void 1624static void
1514incr_parse (JSON *self) 1625incr_parse (JSON *self)
1515{ 1626{
1516 const char *p = SvPVX (self->incr_text) + self->incr_pos; 1627 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1628
1629 // the state machine here is a bit convoluted and could be simplified a lot
1630 // but this would make it slower, so...
1517 1631
1518 for (;;) 1632 for (;;)
1519 { 1633 {
1520 //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 1634 //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
1521 switch (self->incr_mode) 1635 switch (self->incr_mode)
1522 { 1636 {
1523 // only used for intiial whitespace skipping 1637 // only used for initial whitespace skipping
1524 case INCR_M_WS: 1638 case INCR_M_WS:
1525 for (;;) 1639 for (;;)
1526 { 1640 {
1527 if (*p > 0x20) 1641 if (*p > 0x20)
1528 { 1642 {
1643 if (*p == '#')
1644 {
1645 self->incr_mode = INCR_M_C0;
1646 goto incr_m_c;
1647 }
1648 else
1649 {
1529 self->incr_mode = INCR_M_JSON; 1650 self->incr_mode = INCR_M_JSON;
1530 goto incr_m_json; 1651 goto incr_m_json;
1652 }
1531 } 1653 }
1532 else if (!*p) 1654 else if (!*p)
1533 goto interrupt; 1655 goto interrupt;
1534 1656
1535 ++p; 1657 ++p;
… …
1541 goto interrupt; 1663 goto interrupt;
1542 1664
1543 ++p; 1665 ++p;
1544 self->incr_mode = INCR_M_STR; 1666 self->incr_mode = INCR_M_STR;
1545 goto incr_m_str; 1667 goto incr_m_str;
1668
1669 // inside #-style comments
1670 case INCR_M_C0:
1671 case INCR_M_C1:
1672 incr_m_c:
1673 for (;;)
1674 {
1675 if (*p == '\n')
1676 {
1677 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1678 break;
1679 }
1680 else if (!*p)
1681 goto interrupt;
1682
1683 ++p;
1684 }
1685
1686 break;
1546 1687
1547 // inside a string 1688 // inside a string
1548 case INCR_M_STR: 1689 case INCR_M_STR:
1549 incr_m_str: 1690 incr_m_str:
1550 for (;;) 1691 for (;;)
… …
1609 1750
1610 case ']': 1751 case ']':
1611 case '}': 1752 case '}':
1612 if (--self->incr_nest <= 0) 1753 if (--self->incr_nest <= 0)
1613 goto interrupt; 1754 goto interrupt;
1755 break;
1756
1757 case '#':
1758 self->incr_mode = INCR_M_C1;
1759 goto incr_m_c;
1614 } 1760 }
1615 } 1761 }
1616 } 1762 }
1617 1763
1618 modechange: 1764 modechange:
1619 ; 1765 ;
1620 } 1766 }
1621 1767
1622interrupt: 1768interrupt:
1623 self->incr_pos = p - SvPVX (self->incr_text); 1769 self->incr_pos = p - SvPVX (self->incr_text);
1770 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1624 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D 1771 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1625} 1772}
1626 1773
1627///////////////////////////////////////////////////////////////////////////// 1774/////////////////////////////////////////////////////////////////////////////
1628// XS interface functions 1775// XS interface functions
… …
1643 json_stash = gv_stashpv ("JSON::XS" , 1); 1790 json_stash = gv_stashpv ("JSON::XS" , 1);
1644 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1791 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1645 1792
1646 json_true = get_bool ("JSON::XS::true"); 1793 json_true = get_bool ("JSON::XS::true");
1647 json_false = get_bool ("JSON::XS::false"); 1794 json_false = get_bool ("JSON::XS::false");
1795
1796 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1648} 1797}
1649 1798
1650PROTOTYPES: DISABLE 1799PROTOTYPES: DISABLE
1651 1800
1652void CLONE (...) 1801void CLONE (...)
… …
1772 XPUSHs (decode_json (jsonstr, self, 0)); 1921 XPUSHs (decode_json (jsonstr, self, 0));
1773 1922
1774void decode_prefix (JSON *self, SV *jsonstr) 1923void decode_prefix (JSON *self, SV *jsonstr)
1775 PPCODE: 1924 PPCODE:
1776{ 1925{
1777 STRLEN offset; 1926 char *offset;
1778 EXTEND (SP, 2); 1927 EXTEND (SP, 2);
1779 PUSHs (decode_json (jsonstr, self, &offset)); 1928 PUSHs (decode_json (jsonstr, self, &offset));
1780 PUSHs (sv_2mortal (newSVuv (offset))); 1929 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1781} 1930}
1782 1931
1783void incr_parse (JSON *self, SV *jsonstr = 0) 1932void incr_parse (JSON *self, SV *jsonstr = 0)
1784 PPCODE: 1933 PPCODE:
1785{ 1934{
1786 if (!self->incr_text) 1935 if (!self->incr_text)
1787 self->incr_text = newSVpvn ("", 0); 1936 self->incr_text = newSVpvn ("", 0);
1937
1938 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
1939 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
1940 if (DECODE_WANTS_OCTETS (self))
1941 {
1942 if (self->incr_pos)
1943 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
1944 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
1945
1946 sv_utf8_downgrade (self->incr_text, 0);
1947 }
1948 else
1949 {
1950 sv_utf8_upgrade (self->incr_text);
1951
1952 if (self->incr_pos)
1953 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1954 - (U8 *)SvPVX (self->incr_text);
1955 }
1788 1956
1789 // append data, if any 1957 // append data, if any
1790 if (jsonstr) 1958 if (jsonstr)
1791 { 1959 {
1960 /* make sure both strings have same encoding */
1792 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) 1961 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
1962 if (SvUTF8 (jsonstr))
1963 sv_utf8_downgrade (jsonstr, 0);
1793 { 1964 else
1794 /* utf-8-ness differs, need to upgrade */
1795 sv_utf8_upgrade (self->incr_text); 1965 sv_utf8_upgrade (jsonstr);
1796 1966
1797 if (self->incr_pos) 1967 /* and then just blindly append */
1798 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1799 - (U8 *)SvPVX (self->incr_text);
1800 }
1801
1802 { 1968 {
1803 STRLEN len; 1969 STRLEN len;
1804 const char *str = SvPV (jsonstr, len); 1970 const char *str = SvPV (jsonstr, len);
1805 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); 1971 STRLEN cur = SvCUR (self->incr_text);
1972
1973 if (SvLEN (self->incr_text) <= cur + len)
1974 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1975
1806 Move (str, SvEND (self->incr_text), len, char); 1976 Move (str, SvEND (self->incr_text), len, char);
1807 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 1977 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1808 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there 1978 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1809 } 1979 }
1810 } 1980 }
1811 1981
1812 if (GIMME_V != G_VOID) 1982 if (GIMME_V != G_VOID)
1813 do 1983 do
1814 { 1984 {
1815 STRLEN offset; 1985 char *offset;
1816 1986
1817 if (!INCR_DONE (self)) 1987 if (!INCR_DONE (self))
1818 { 1988 {
1819 incr_parse (self); 1989 incr_parse (self);
1820 1990
… …
1826 break; 1996 break;
1827 } 1997 }
1828 1998
1829 XPUSHs (decode_json (self->incr_text, self, &offset)); 1999 XPUSHs (decode_json (self->incr_text, self, &offset));
1830 2000
1831 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1832 self->incr_pos -= offset; 2001 self->incr_pos -= offset - SvPVX (self->incr_text);
1833 self->incr_nest = 0; 2002 self->incr_nest = 0;
1834 self->incr_mode = 0; 2003 self->incr_mode = 0;
2004
2005 sv_chop (self->incr_text, offset);
1835 } 2006 }
1836 while (GIMME_V == G_ARRAY); 2007 while (GIMME_V == G_ARRAY);
1837} 2008}
1838 2009
1839SV *incr_text (JSON *self) 2010SV *incr_text (JSON *self)
… …
1900 json_init (&json); 2071 json_init (&json);
1901 json.flags |= ix; 2072 json.flags |= ix;
1902 XPUSHs (decode_json (jsonstr, &json, 0)); 2073 XPUSHs (decode_json (jsonstr, &json, 0));
1903} 2074}
1904 2075
1905

Diff Legend

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