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 1.86 by root, Tue May 27 05:31:39 2008 UTC vs.
Revision 1.108 by root, Tue Aug 17 23:27:36 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
… …
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 && (json)->incr_mode == INCR_M_JSON) 89#define INCR_DONE(json) ((json)->incr_nest <= 0 && (json)->incr_mode == INCR_M_JSON)
85 90
86typedef struct { 91typedef struct {
87 U32 flags; 92 U32 flags;
88 U32 max_depth; 93 U32 max_depth;
89 STRLEN max_size; 94 STRLEN max_size;
… …
92 HV *cb_sk_object; 97 HV *cb_sk_object;
93 98
94 // for the incremental parser 99 // for the incremental parser
95 SV *incr_text; // the source text so far 100 SV *incr_text; // the source text so far
96 STRLEN incr_pos; // the current offset into the text 101 STRLEN incr_pos; // the current offset into the text
97 unsigned char incr_nest; // {[]}-nesting level 102 int incr_nest; // {[]}-nesting level
98 unsigned char incr_mode; 103 unsigned char incr_mode;
99} JSON; 104} JSON;
100 105
101INLINE void 106INLINE void
102json_init (JSON *json) 107json_init (JSON *json)
… …
121 126
122INLINE void 127INLINE void
123shrink (SV *sv) 128shrink (SV *sv)
124{ 129{
125 sv_utf8_downgrade (sv, 1); 130 sv_utf8_downgrade (sv, 1);
131
126 if (SvLEN (sv) > SvCUR (sv) + 1) 132 if (SvLEN (sv) > SvCUR (sv) + 1)
127 { 133 {
128#ifdef SvPV_shrink_to_cur 134#ifdef SvPV_shrink_to_cur
129 SvPV_shrink_to_cur (sv); 135 SvPV_shrink_to_cur (sv);
130#elif defined (SvPV_renew) 136#elif defined (SvPV_renew)
… …
174 *s++ = 0x80 | ( ch & 0x3f); 180 *s++ = 0x80 | ( ch & 0x3f);
175 181
176 return s; 182 return s;
177} 183}
178 184
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}
179///////////////////////////////////////////////////////////////////////////// 288/////////////////////////////////////////////////////////////////////////////
180// encoder 289// encoder
181 290
182// structure used for encoding JSON 291// structure used for encoding JSON
183typedef struct 292typedef struct
… …
193INLINE void 302INLINE void
194need (enc_t *enc, STRLEN len) 303need (enc_t *enc, STRLEN len)
195{ 304{
196 if (expect_false (enc->cur + len >= enc->end)) 305 if (expect_false (enc->cur + len >= enc->end))
197 { 306 {
198 STRLEN cur = enc->cur - SvPVX (enc->sv); 307 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
199 SvGROW (enc->sv, cur + len + 1); 308 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
200 enc->cur = SvPVX (enc->sv) + cur; 309 enc->cur = SvPVX (enc->sv) + cur;
201 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 310 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
202 } 311 }
203} 312}
204 313
… …
279 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 388 (int)((uch - 0x10000) % 0x400 + 0xDC00));
280 enc->cur += 12; 389 enc->cur += 12;
281 } 390 }
282 else 391 else
283 { 392 {
284 static char hexdigit [16] = "0123456789abcdef";
285 need (enc, len += 5); 393 need (enc, len += 5);
286 *enc->cur++ = '\\'; 394 *enc->cur++ = '\\';
287 *enc->cur++ = 'u'; 395 *enc->cur++ = 'u';
288 *enc->cur++ = hexdigit [ uch >> 12 ]; 396 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
289 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 397 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
290 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 398 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
291 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 399 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
292 } 400 }
293 401
294 str += clen; 402 str += clen;
295 } 403 }
296 else if (enc->json.flags & F_LATIN1) 404 else if (enc->json.flags & F_LATIN1)
… …
460 568
461 encode_ch (enc, '{'); 569 encode_ch (enc, '{');
462 570
463 // for canonical output we have to sort by keys first 571 // for canonical output we have to sort by keys first
464 // actually, this is mostly due to the stupid so-called 572 // actually, this is mostly due to the stupid so-called
465 // security workaround added somewhere in 5.8.x. 573 // security workaround added somewhere in 5.8.x
466 // that randomises hash orderings 574 // that randomises hash orderings
467 if (enc->json.flags & F_CANONICAL) 575 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
468 { 576 {
469 int count = hv_iterinit (hv); 577 int count = hv_iterinit (hv);
470 578
471 if (SvMAGICAL (hv)) 579 if (SvMAGICAL (hv))
472 { 580 {
… …
750 : enc.json.flags & F_LATIN1 ? 0x000100UL 858 : enc.json.flags & F_LATIN1 ? 0x000100UL
751 : 0x110000UL; 859 : 0x110000UL;
752 860
753 SvPOK_only (enc.sv); 861 SvPOK_only (enc.sv);
754 encode_sv (&enc, scalar); 862 encode_sv (&enc, scalar);
863 encode_nl (&enc);
755 864
756 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 865 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
757 *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
758 867
759 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 868 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
939 else if (expect_true (ch >= 0x20 && ch < 0x80)) 1048 else if (expect_true (ch >= 0x20 && ch < 0x80))
940 *cur++ = ch; 1049 *cur++ = ch;
941 else if (ch >= 0x80) 1050 else if (ch >= 0x80)
942 { 1051 {
943 STRLEN clen; 1052 STRLEN clen;
944 UV uch;
945 1053
946 --dec_cur; 1054 --dec_cur;
947 1055
948 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1056 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
949 if (clen == (STRLEN)-1) 1057 if (clen == (STRLEN)-1)
950 ERR ("malformed UTF-8 character in JSON string"); 1058 ERR ("malformed UTF-8 character in JSON string");
951 1059
952 do 1060 do
953 *cur++ = *dec_cur++; 1061 *cur++ = *dec_cur++;
… …
970 { 1078 {
971 STRLEN len = cur - buf; 1079 STRLEN len = cur - buf;
972 1080
973 if (sv) 1081 if (sv)
974 { 1082 {
975 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
976 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1088 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
977 SvCUR_set (sv, SvCUR (sv) + len); 1089 SvCUR_set (sv, SvCUR (sv) + len);
978 } 1090 }
979 else 1091 else
980 sv = newSVpvn (buf, len); 1092 sv = newSVpvn (buf, len);
… …
1071 1183
1072 // special case the rather common 1..5-digit-int case 1184 // special case the rather common 1..5-digit-int case
1073 if (*start == '-') 1185 if (*start == '-')
1074 switch (len) 1186 switch (len)
1075 { 1187 {
1076 case 2: return newSViv (-( start [1] - '0' * 1)); 1188 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1077 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1189 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)); 1190 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)); 1191 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)); 1192 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1081 } 1193 }
1082 else 1194 else
1083 switch (len) 1195 switch (len)
1084 { 1196 {
1085 case 1: return newSViv ( start [0] - '0' * 1); 1197 case 1: return newSViv ( start [0] - '0' * 1);
1086 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1198 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); 1199 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); 1200 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); 1201 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1090 } 1202 }
1091 1203
1092 { 1204 {
1093 UV uv; 1205 UV uv;
1094 int numtype = grok_number (start, len, &uv); 1206 int numtype = grok_number (start, len, &uv);
… …
1103 } 1215 }
1104 1216
1105 len -= *start == '-' ? 1 : 0; 1217 len -= *start == '-' ? 1 : 0;
1106 1218
1107 // does not fit into IV or UV, try NV 1219 // does not fit into IV or UV, try NV
1108 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1220 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 1221 // fits into NV without loss of precision
1114 return newSVnv (Atof (start)); 1222 return newSVnv (json_atof (start));
1115 1223
1116 // everything else fails, convert it to a string 1224 // everything else fails, convert it to a string
1117 return newSVpvn (start, dec->cur - start); 1225 return newSVpvn (start, dec->cur - start);
1118 } 1226 }
1119 1227
1120 // loss of precision here 1228 // loss of precision here
1121 return newSVnv (Atof (start)); 1229 return newSVnv (json_atof (start));
1122 1230
1123fail: 1231fail:
1124 return 0; 1232 return 0;
1125} 1233}
1126 1234
… …
1295 dSP; 1403 dSP;
1296 int count; 1404 int count;
1297 1405
1298 ENTER; SAVETMPS; PUSHMARK (SP); 1406 ENTER; SAVETMPS; PUSHMARK (SP);
1299 XPUSHs (HeVAL (he)); 1407 XPUSHs (HeVAL (he));
1408 sv_2mortal (sv);
1300 1409
1301 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1410 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1302 1411
1303 if (count == 1) 1412 if (count == 1)
1304 { 1413 {
1305 sv = newSVsv (POPs); 1414 sv = newSVsv (POPs);
1306 FREETMPS; LEAVE; 1415 FREETMPS; LEAVE;
1307 return sv; 1416 return sv;
1308 } 1417 }
1309 1418
1419 SvREFCNT_inc (sv);
1310 FREETMPS; LEAVE; 1420 FREETMPS; LEAVE;
1311 } 1421 }
1312 } 1422 }
1313 1423
1314 if (dec->json.cb_object) 1424 if (dec->json.cb_object)
… …
1404fail: 1514fail:
1405 return 0; 1515 return 0;
1406} 1516}
1407 1517
1408static SV * 1518static SV *
1409decode_json (SV *string, JSON *json, STRLEN *offset_return) 1519decode_json (SV *string, JSON *json, char **offset_return)
1410{ 1520{
1411 dec_t dec; 1521 dec_t dec;
1412 STRLEN offset;
1413 SV *sv; 1522 SV *sv;
1414 1523
1524 /* work around bugs in 5.10 where manipulating magic values
1525 * will perl ignore the magic in subsequent accesses
1526 */
1415 SvGETMAGIC (string); 1527 /*SvGETMAGIC (string);*/
1528 if (SvMAGICAL (string))
1529 string = sv_2mortal (newSVsv (string));
1530
1416 SvUPGRADE (string, SVt_PV); 1531 SvUPGRADE (string, SVt_PV);
1417 1532
1533 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1534 * assertion with -DDEBUGGING, although SvCUR is documented to
1535 * return the xpv_cur field which certainly exists after upgrading.
1536 * according to nicholas clark, calling SvPOK fixes this.
1537 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1538 * and hope for the best.
1539 * Damnit, SvPV_nolen still trips over yet another assertion. This
1540 * assertion business is seriously broken, try yet another workaround
1541 * for the broken -DDEBUGGING.
1542 */
1543 {
1544#ifdef DEBUGGING
1545 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1546#else
1547 STRLEN offset = SvCUR (string);
1548#endif
1549
1418 if (SvCUR (string) > json->max_size && json->max_size) 1550 if (offset > json->max_size && json->max_size)
1419 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1551 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1420 (unsigned long)SvCUR (string), (unsigned long)json->max_size); 1552 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1553 }
1421 1554
1422 if (json->flags & F_UTF8) 1555 if (DECODE_WANTS_OCTETS (json))
1423 sv_utf8_downgrade (string, 0); 1556 sv_utf8_downgrade (string, 0);
1424 else 1557 else
1425 sv_utf8_upgrade (string); 1558 sv_utf8_upgrade (string);
1426 1559
1427 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1560 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
… …
1437 1570
1438 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1571 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1439 1572
1440 decode_ws (&dec); 1573 decode_ws (&dec);
1441 sv = decode_sv (&dec); 1574 sv = decode_sv (&dec);
1575
1576 if (offset_return)
1577 *offset_return = dec.cur;
1442 1578
1443 if (!(offset_return || !sv)) 1579 if (!(offset_return || !sv))
1444 { 1580 {
1445 // check for trailing garbage 1581 // check for trailing garbage
1446 decode_ws (&dec); 1582 decode_ws (&dec);
… …
1449 { 1585 {
1450 dec.err = "garbage after JSON object"; 1586 dec.err = "garbage after JSON object";
1451 SvREFCNT_dec (sv); 1587 SvREFCNT_dec (sv);
1452 sv = 0; 1588 sv = 0;
1453 } 1589 }
1454 }
1455
1456 if (offset_return || !sv)
1457 {
1458 offset = dec.json.flags & F_UTF8
1459 ? dec.cur - SvPVX (string)
1460 : utf8_distance (dec.cur, SvPVX (string));
1461
1462 if (offset_return)
1463 *offset_return = offset;
1464 } 1590 }
1465 1591
1466 if (!sv) 1592 if (!sv)
1467 { 1593 {
1468 SV *uni = sv_newmortal (); 1594 SV *uni = sv_newmortal ();
… …
1474 SAVEVPTR (PL_curcop); 1600 SAVEVPTR (PL_curcop);
1475 PL_curcop = &cop; 1601 PL_curcop = &cop;
1476 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1602 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1477 LEAVE; 1603 LEAVE;
1478 1604
1479 croak ("%s, at character offset %d [\"%s\"]", 1605 croak ("%s, at character offset %d (before \"%s\")",
1480 dec.err, 1606 dec.err,
1481 (int)offset, 1607 ptr_to_index (string, dec.cur),
1482 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1608 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1483 } 1609 }
1484 1610
1485 sv = sv_2mortal (sv); 1611 sv = sv_2mortal (sv);
1486 1612
… …
1495 1621
1496static void 1622static void
1497incr_parse (JSON *self) 1623incr_parse (JSON *self)
1498{ 1624{
1499 const char *p = SvPVX (self->incr_text) + self->incr_pos; 1625 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1626
1627 // the state machine here is a bit convoluted and could be simplified a lot
1628 // but this would make it slower, so...
1500 1629
1501 for (;;) 1630 for (;;)
1502 { 1631 {
1503 //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 1632 //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
1504 switch (self->incr_mode) 1633 switch (self->incr_mode)
1505 { 1634 {
1506 // only used for intiial whitespace skipping 1635 // only used for initial whitespace skipping
1507 case INCR_M_WS: 1636 case INCR_M_WS:
1508 for (;;) 1637 for (;;)
1509 { 1638 {
1510 if (*p > 0x20) 1639 if (*p > 0x20)
1511 { 1640 {
1641 if (*p == '#')
1642 {
1643 self->incr_mode = INCR_M_C0;
1644 goto incr_m_c;
1645 }
1646 else
1647 {
1512 self->incr_mode = INCR_M_JSON; 1648 self->incr_mode = INCR_M_JSON;
1513 goto incr_m_json; 1649 goto incr_m_json;
1650 }
1514 } 1651 }
1515 else if (!*p) 1652 else if (!*p)
1516 goto interrupt; 1653 goto interrupt;
1517 1654
1518 ++p; 1655 ++p;
… …
1524 goto interrupt; 1661 goto interrupt;
1525 1662
1526 ++p; 1663 ++p;
1527 self->incr_mode = INCR_M_STR; 1664 self->incr_mode = INCR_M_STR;
1528 goto incr_m_str; 1665 goto incr_m_str;
1666
1667 // inside #-style comments
1668 case INCR_M_C0:
1669 case INCR_M_C1:
1670 incr_m_c:
1671 for (;;)
1672 {
1673 if (*p == '\n')
1674 {
1675 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1676 break;
1677 }
1678 else if (!*p)
1679 goto interrupt;
1680
1681 ++p;
1682 }
1683
1684 break;
1529 1685
1530 // inside a string 1686 // inside a string
1531 case INCR_M_STR: 1687 case INCR_M_STR:
1532 incr_m_str: 1688 incr_m_str:
1533 for (;;) 1689 for (;;)
… …
1590 croak (ERR_NESTING_EXCEEDED); 1746 croak (ERR_NESTING_EXCEEDED);
1591 break; 1747 break;
1592 1748
1593 case ']': 1749 case ']':
1594 case '}': 1750 case '}':
1595 if (!--self->incr_nest) 1751 if (--self->incr_nest <= 0)
1596 goto interrupt; 1752 goto interrupt;
1753 break;
1754
1755 case '#':
1756 self->incr_mode = INCR_M_C1;
1757 goto incr_m_c;
1597 } 1758 }
1598 } 1759 }
1599 } 1760 }
1600 1761
1601 modechange: 1762 modechange:
1602 ; 1763 ;
1603 } 1764 }
1604 1765
1605interrupt: 1766interrupt:
1606 self->incr_pos = p - SvPVX (self->incr_text); 1767 self->incr_pos = p - SvPVX (self->incr_text);
1768 //printf ("interrupt<%.*s>\n", self->incr_pos, SvPVX(self->incr_text));//D
1607 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D 1769 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1608} 1770}
1609 1771
1610///////////////////////////////////////////////////////////////////////////// 1772/////////////////////////////////////////////////////////////////////////////
1611// XS interface functions 1773// XS interface functions
… …
1626 json_stash = gv_stashpv ("JSON::XS" , 1); 1788 json_stash = gv_stashpv ("JSON::XS" , 1);
1627 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1789 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1628 1790
1629 json_true = get_bool ("JSON::XS::true"); 1791 json_true = get_bool ("JSON::XS::true");
1630 json_false = get_bool ("JSON::XS::false"); 1792 json_false = get_bool ("JSON::XS::false");
1793
1794 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1631} 1795}
1632 1796
1633PROTOTYPES: DISABLE 1797PROTOTYPES: DISABLE
1634 1798
1635void CLONE (...) 1799void CLONE (...)
… …
1755 XPUSHs (decode_json (jsonstr, self, 0)); 1919 XPUSHs (decode_json (jsonstr, self, 0));
1756 1920
1757void decode_prefix (JSON *self, SV *jsonstr) 1921void decode_prefix (JSON *self, SV *jsonstr)
1758 PPCODE: 1922 PPCODE:
1759{ 1923{
1760 STRLEN offset; 1924 char *offset;
1761 EXTEND (SP, 2); 1925 EXTEND (SP, 2);
1762 PUSHs (decode_json (jsonstr, self, &offset)); 1926 PUSHs (decode_json (jsonstr, self, &offset));
1763 PUSHs (sv_2mortal (newSVuv (offset))); 1927 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1764} 1928}
1765 1929
1766void incr_parse (JSON *self, SV *jsonstr = 0) 1930void incr_parse (JSON *self, SV *jsonstr = 0)
1767 PPCODE: 1931 PPCODE:
1768{ 1932{
1769 if (!self->incr_text) 1933 if (!self->incr_text)
1770 self->incr_text = newSVpvn ("", 0); 1934 self->incr_text = newSVpvn ("", 0);
1935
1936 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
1937 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
1938 if (DECODE_WANTS_OCTETS (self))
1939 {
1940 if (self->incr_pos)
1941 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
1942 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
1943
1944 sv_utf8_downgrade (self->incr_text, 0);
1945 }
1946 else
1947 {
1948 sv_utf8_upgrade (self->incr_text);
1949
1950 if (self->incr_pos)
1951 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1952 - (U8 *)SvPVX (self->incr_text);
1953 }
1771 1954
1772 // append data, if any 1955 // append data, if any
1773 if (jsonstr) 1956 if (jsonstr)
1774 { 1957 {
1958 /* make sure both strings have same encoding */
1775 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) 1959 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
1960 if (SvUTF8 (jsonstr))
1961 sv_utf8_downgrade (jsonstr, 0);
1776 { 1962 else
1777 /* utf-8-ness differs, need to upgrade */
1778 sv_utf8_upgrade (self->incr_text); 1963 sv_utf8_upgrade (jsonstr);
1779 1964
1780 if (self->incr_pos) 1965 /* and then just blindly append */
1781 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1782 - (U8 *)SvPVX (self->incr_text);
1783 }
1784
1785 { 1966 {
1786 STRLEN len; 1967 STRLEN len;
1787 const char *str = SvPV (jsonstr, len); 1968 const char *str = SvPV (jsonstr, len);
1788 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); 1969 STRLEN cur = SvCUR (self->incr_text);
1970
1971 if (SvLEN (self->incr_text) <= cur + len)
1972 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1973
1789 Move (str, SvEND (self->incr_text), len, char); 1974 Move (str, SvEND (self->incr_text), len, char);
1790 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 1975 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len);
1791 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there 1976 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1792 } 1977 }
1793 } 1978 }
1794 1979
1795 if (GIMME_V != G_VOID) 1980 if (GIMME_V != G_VOID)
1796 do 1981 do
1797 { 1982 {
1798 STRLEN offset; 1983 char *offset;
1799 1984
1800 if (!INCR_DONE (self)) 1985 if (!INCR_DONE (self))
1801 { 1986 {
1802 incr_parse (self); 1987 incr_parse (self);
1803 1988
… …
1809 break; 1994 break;
1810 } 1995 }
1811 1996
1812 XPUSHs (decode_json (self->incr_text, self, &offset)); 1997 XPUSHs (decode_json (self->incr_text, self, &offset));
1813 1998
1814 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1815 self->incr_pos -= offset; 1999 self->incr_pos -= offset - SvPVX (self->incr_text);
1816 self->incr_nest = 0; 2000 self->incr_nest = 0;
1817 self->incr_mode = 0; 2001 self->incr_mode = 0;
2002
2003 sv_chop (self->incr_text, offset);
1818 } 2004 }
1819 while (GIMME_V == G_ARRAY); 2005 while (GIMME_V == G_ARRAY);
1820} 2006}
1821 2007
1822SV *incr_text (JSON *self) 2008SV *incr_text (JSON *self)
… …
1883 json_init (&json); 2069 json_init (&json);
1884 json.flags |= ix; 2070 json.flags |= ix;
1885 XPUSHs (decode_json (jsonstr, &json, 0)); 2071 XPUSHs (decode_json (jsonstr, &json, 0));
1886} 2072}
1887 2073
1888

Diff Legend

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