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 1.85 by root, Sat Apr 5 18:15:46 2008 UTC vs.
Revision 1.113 by root, Thu Aug 11 21:37:41 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 && (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 // (IEEE 754 conformant ones are required to be exact)
266 if (postdp) *expo -= eaccum;
267 *accum += uaccum * Perl_pow (10., *expo);
268 *expo += eaccum;
269}
270
271static NV
272json_atof (const char *s)
273{
274 NV accum = 0.;
275 int expo = 0;
276 int neg = 0;
277
278 if (*s == '-')
279 {
280 ++s;
281 neg = 1;
282 }
283
284 // a recursion depth of ten gives us >>500 bits
285 json_atof_scan1 (s, &accum, &expo, 0, 10);
286
287 return neg ? -accum : accum;
288}
179///////////////////////////////////////////////////////////////////////////// 289/////////////////////////////////////////////////////////////////////////////
180// encoder 290// encoder
181 291
182// structure used for encoding JSON 292// structure used for encoding JSON
183typedef struct 293typedef struct
… …
193INLINE void 303INLINE void
194need (enc_t *enc, STRLEN len) 304need (enc_t *enc, STRLEN len)
195{ 305{
196 if (expect_false (enc->cur + len >= enc->end)) 306 if (expect_false (enc->cur + len >= enc->end))
197 { 307 {
198 STRLEN cur = enc->cur - SvPVX (enc->sv); 308 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
199 SvGROW (enc->sv, cur + len + 1); 309 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
200 enc->cur = SvPVX (enc->sv) + cur; 310 enc->cur = SvPVX (enc->sv) + cur;
201 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 311 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
202 } 312 }
203} 313}
204 314
… …
279 (int)((uch - 0x10000) % 0x400 + 0xDC00)); 389 (int)((uch - 0x10000) % 0x400 + 0xDC00));
280 enc->cur += 12; 390 enc->cur += 12;
281 } 391 }
282 else 392 else
283 { 393 {
284 static char hexdigit [16] = "0123456789abcdef";
285 need (enc, len += 5); 394 need (enc, len += 5);
286 *enc->cur++ = '\\'; 395 *enc->cur++ = '\\';
287 *enc->cur++ = 'u'; 396 *enc->cur++ = 'u';
288 *enc->cur++ = hexdigit [ uch >> 12 ]; 397 *enc->cur++ = PL_hexdigit [ uch >> 12 ];
289 *enc->cur++ = hexdigit [(uch >> 8) & 15]; 398 *enc->cur++ = PL_hexdigit [(uch >> 8) & 15];
290 *enc->cur++ = hexdigit [(uch >> 4) & 15]; 399 *enc->cur++ = PL_hexdigit [(uch >> 4) & 15];
291 *enc->cur++ = hexdigit [(uch >> 0) & 15]; 400 *enc->cur++ = PL_hexdigit [(uch >> 0) & 15];
292 } 401 }
293 402
294 str += clen; 403 str += clen;
295 } 404 }
296 else if (enc->json.flags & F_LATIN1) 405 else if (enc->json.flags & F_LATIN1)
… …
460 569
461 encode_ch (enc, '{'); 570 encode_ch (enc, '{');
462 571
463 // for canonical output we have to sort by keys first 572 // for canonical output we have to sort by keys first
464 // actually, this is mostly due to the stupid so-called 573 // actually, this is mostly due to the stupid so-called
465 // security workaround added somewhere in 5.8.x. 574 // security workaround added somewhere in 5.8.x
466 // that randomises hash orderings 575 // that randomises hash orderings
467 if (enc->json.flags & F_CANONICAL) 576 if (enc->json.flags & F_CANONICAL && !SvRMAGICAL (hv))
468 { 577 {
469 int count = hv_iterinit (hv); 578 int count = hv_iterinit (hv);
470 579
471 if (SvMAGICAL (hv)) 580 if (SvMAGICAL (hv))
472 { 581 {
… …
728 encode_rv (enc, SvRV (sv)); 837 encode_rv (enc, SvRV (sv));
729 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) 838 else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN)
730 encode_str (enc, "null", 4, 0); 839 encode_str (enc, "null", 4, 0);
731 else 840 else
732 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", 841 croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this",
733 SvPV_nolen (sv), SvFLAGS (sv)); 842 SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
734} 843}
735 844
736static SV * 845static SV *
737encode_json (SV *scalar, JSON *json) 846encode_json (SV *scalar, JSON *json)
738{ 847{
… …
750 : enc.json.flags & F_LATIN1 ? 0x000100UL 859 : enc.json.flags & F_LATIN1 ? 0x000100UL
751 : 0x110000UL; 860 : 0x110000UL;
752 861
753 SvPOK_only (enc.sv); 862 SvPOK_only (enc.sv);
754 encode_sv (&enc, scalar); 863 encode_sv (&enc, scalar);
864 encode_nl (&enc);
755 865
756 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); 866 SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
757 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings 867 *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
758 868
759 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8))) 869 if (!(enc.json.flags & (F_ASCII | F_LATIN1 | F_UTF8)))
… …
939 else if (expect_true (ch >= 0x20 && ch < 0x80)) 1049 else if (expect_true (ch >= 0x20 && ch < 0x80))
940 *cur++ = ch; 1050 *cur++ = ch;
941 else if (ch >= 0x80) 1051 else if (ch >= 0x80)
942 { 1052 {
943 STRLEN clen; 1053 STRLEN clen;
944 UV uch;
945 1054
946 --dec_cur; 1055 --dec_cur;
947 1056
948 uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); 1057 decode_utf8 (dec_cur, dec->end - dec_cur, &clen);
949 if (clen == (STRLEN)-1) 1058 if (clen == (STRLEN)-1)
950 ERR ("malformed UTF-8 character in JSON string"); 1059 ERR ("malformed UTF-8 character in JSON string");
951 1060
952 do 1061 do
953 *cur++ = *dec_cur++; 1062 *cur++ = *dec_cur++;
… …
970 { 1079 {
971 STRLEN len = cur - buf; 1080 STRLEN len = cur - buf;
972 1081
973 if (sv) 1082 if (sv)
974 { 1083 {
975 SvGROW (sv, SvCUR (sv) + len + 1); 1084 STRLEN cur = SvCUR (sv);
1085
1086 if (SvLEN (sv) <= cur + len)
1087 SvGROW (sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1088
976 memcpy (SvPVX (sv) + SvCUR (sv), buf, len); 1089 memcpy (SvPVX (sv) + SvCUR (sv), buf, len);
977 SvCUR_set (sv, SvCUR (sv) + len); 1090 SvCUR_set (sv, SvCUR (sv) + len);
978 } 1091 }
979 else 1092 else
980 sv = newSVpvn (buf, len); 1093 sv = newSVpvn (buf, len);
… …
1071 1184
1072 // special case the rather common 1..5-digit-int case 1185 // special case the rather common 1..5-digit-int case
1073 if (*start == '-') 1186 if (*start == '-')
1074 switch (len) 1187 switch (len)
1075 { 1188 {
1076 case 2: return newSViv (-( start [1] - '0' * 1)); 1189 case 2: return newSViv (-(IV)( start [1] - '0' * 1));
1077 case 3: return newSViv (-( start [1] * 10 + start [2] - '0' * 11)); 1190 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)); 1191 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)); 1192 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)); 1193 case 6: return newSViv (-(IV)(start [1] * 10000 + start [2] * 1000 + start [3] * 100 + start [4] * 10 + start [5] - '0' * 11111));
1081 } 1194 }
1082 else 1195 else
1083 switch (len) 1196 switch (len)
1084 { 1197 {
1085 case 1: return newSViv ( start [0] - '0' * 1); 1198 case 1: return newSViv ( start [0] - '0' * 1);
1086 case 2: return newSViv ( start [0] * 10 + start [1] - '0' * 11); 1199 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); 1200 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); 1201 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); 1202 case 5: return newSViv ( start [0] * 10000 + start [1] * 1000 + start [2] * 100 + start [3] * 10 + start [4] - '0' * 11111);
1090 } 1203 }
1091 1204
1092 { 1205 {
1093 UV uv; 1206 UV uv;
1094 int numtype = grok_number (start, len, &uv); 1207 int numtype = grok_number (start, len, &uv);
… …
1103 } 1216 }
1104 1217
1105 len -= *start == '-' ? 1 : 0; 1218 len -= *start == '-' ? 1 : 0;
1106 1219
1107 // does not fit into IV or UV, try NV 1220 // does not fit into IV or UV, try NV
1108 if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) 1221 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 1222 // fits into NV without loss of precision
1114 return newSVnv (Atof (start)); 1223 return newSVnv (json_atof (start));
1115 1224
1116 // everything else fails, convert it to a string 1225 // everything else fails, convert it to a string
1117 return newSVpvn (start, dec->cur - start); 1226 return newSVpvn (start, dec->cur - start);
1118 } 1227 }
1119 1228
1120 // loss of precision here 1229 // loss of precision here
1121 return newSVnv (Atof (start)); 1230 return newSVnv (json_atof (start));
1122 1231
1123fail: 1232fail:
1124 return 0; 1233 return 0;
1125} 1234}
1126 1235
… …
1295 dSP; 1404 dSP;
1296 int count; 1405 int count;
1297 1406
1298 ENTER; SAVETMPS; PUSHMARK (SP); 1407 ENTER; SAVETMPS; PUSHMARK (SP);
1299 XPUSHs (HeVAL (he)); 1408 XPUSHs (HeVAL (he));
1409 sv_2mortal (sv);
1300 1410
1301 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; 1411 PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN;
1302 1412
1303 if (count == 1) 1413 if (count == 1)
1304 { 1414 {
1305 sv = newSVsv (POPs); 1415 sv = newSVsv (POPs);
1306 FREETMPS; LEAVE; 1416 FREETMPS; LEAVE;
1307 return sv; 1417 return sv;
1308 } 1418 }
1309 1419
1420 SvREFCNT_inc (sv);
1310 FREETMPS; LEAVE; 1421 FREETMPS; LEAVE;
1311 } 1422 }
1312 } 1423 }
1313 1424
1314 if (dec->json.cb_object) 1425 if (dec->json.cb_object)
… …
1404fail: 1515fail:
1405 return 0; 1516 return 0;
1406} 1517}
1407 1518
1408static SV * 1519static SV *
1409decode_json (SV *string, JSON *json, STRLEN *offset_return) 1520decode_json (SV *string, JSON *json, char **offset_return)
1410{ 1521{
1411 dec_t dec; 1522 dec_t dec;
1412 STRLEN offset;
1413 SV *sv; 1523 SV *sv;
1414 1524
1525 /* work around bugs in 5.10 where manipulating magic values
1526 * will perl ignore the magic in subsequent accesses.
1527 * also make a copy of non-PV values, to get them into a clean
1528 * state (SvPV should do that, but it's buggy, see below).
1529 */
1415 SvGETMAGIC (string); 1530 /*SvGETMAGIC (string);*/
1531 if (SvMAGICAL (string) || !SvPOK (string))
1532 string = sv_2mortal (newSVsv (string));
1533
1416 SvUPGRADE (string, SVt_PV); 1534 SvUPGRADE (string, SVt_PV);
1417 1535
1536 /* work around a bug in perl 5.10, which causes SvCUR to fail an
1537 * assertion with -DDEBUGGING, although SvCUR is documented to
1538 * return the xpv_cur field which certainly exists after upgrading.
1539 * according to nicholas clark, calling SvPOK fixes this.
1540 * But it doesn't fix it, so try another workaround, call SvPV_nolen
1541 * and hope for the best.
1542 * Damnit, SvPV_nolen still trips over yet another assertion. This
1543 * assertion business is seriously broken, try yet another workaround
1544 * for the broken -DDEBUGGING.
1545 */
1546 {
1547#ifdef DEBUGGING
1548 STRLEN offset = SvOK (string) ? sv_len (string) : 0;
1549#else
1550 STRLEN offset = SvCUR (string);
1551#endif
1552
1418 if (SvCUR (string) > json->max_size && json->max_size) 1553 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", 1554 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); 1555 (unsigned long)SvCUR (string), (unsigned long)json->max_size);
1556 }
1421 1557
1422 if (json->flags & F_UTF8) 1558 if (DECODE_WANTS_OCTETS (json))
1423 sv_utf8_downgrade (string, 0); 1559 sv_utf8_downgrade (string, 0);
1424 else 1560 else
1425 sv_utf8_upgrade (string); 1561 sv_utf8_upgrade (string);
1426 1562
1427 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP 1563 SvGROW (string, SvCUR (string) + 1); // should basically be a NOP
… …
1437 1573
1438 *dec.end = 0; // this should basically be a nop, too, but make sure it's there 1574 *dec.end = 0; // this should basically be a nop, too, but make sure it's there
1439 1575
1440 decode_ws (&dec); 1576 decode_ws (&dec);
1441 sv = decode_sv (&dec); 1577 sv = decode_sv (&dec);
1578
1579 if (offset_return)
1580 *offset_return = dec.cur;
1442 1581
1443 if (!(offset_return || !sv)) 1582 if (!(offset_return || !sv))
1444 { 1583 {
1445 // check for trailing garbage 1584 // check for trailing garbage
1446 decode_ws (&dec); 1585 decode_ws (&dec);
… …
1449 { 1588 {
1450 dec.err = "garbage after JSON object"; 1589 dec.err = "garbage after JSON object";
1451 SvREFCNT_dec (sv); 1590 SvREFCNT_dec (sv);
1452 sv = 0; 1591 sv = 0;
1453 } 1592 }
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 } 1593 }
1465 1594
1466 if (!sv) 1595 if (!sv)
1467 { 1596 {
1468 SV *uni = sv_newmortal (); 1597 SV *uni = sv_newmortal ();
… …
1474 SAVEVPTR (PL_curcop); 1603 SAVEVPTR (PL_curcop);
1475 PL_curcop = &cop; 1604 PL_curcop = &cop;
1476 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ); 1605 pv_uni_display (uni, dec.cur, dec.end - dec.cur, 20, UNI_DISPLAY_QQ);
1477 LEAVE; 1606 LEAVE;
1478 1607
1479 croak ("%s, at character offset %d [\"%s\"]", 1608 croak ("%s, at character offset %d (before \"%s\")",
1480 dec.err, 1609 dec.err,
1481 (int)offset, 1610 (int)ptr_to_index (string, dec.cur),
1482 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)"); 1611 dec.cur != dec.end ? SvPV_nolen (uni) : "(end of string)");
1483 } 1612 }
1484 1613
1485 sv = sv_2mortal (sv); 1614 sv = sv_2mortal (sv);
1486 1615
… …
1495 1624
1496static void 1625static void
1497incr_parse (JSON *self) 1626incr_parse (JSON *self)
1498{ 1627{
1499 const char *p = SvPVX (self->incr_text) + self->incr_pos; 1628 const char *p = SvPVX (self->incr_text) + self->incr_pos;
1629
1630 // the state machine here is a bit convoluted and could be simplified a lot
1631 // but this would make it slower, so...
1500 1632
1501 for (;;) 1633 for (;;)
1502 { 1634 {
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 1635 //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) 1636 switch (self->incr_mode)
1505 { 1637 {
1506 // only used for intiial whitespace skipping 1638 // only used for initial whitespace skipping
1507 case INCR_M_WS: 1639 case INCR_M_WS:
1508 for (;;) 1640 for (;;)
1509 { 1641 {
1510 if (*p > 0x20) 1642 if (*p > 0x20)
1511 { 1643 {
1644 if (*p == '#')
1645 {
1646 self->incr_mode = INCR_M_C0;
1647 goto incr_m_c;
1648 }
1649 else
1650 {
1512 self->incr_mode = INCR_M_JSON; 1651 self->incr_mode = INCR_M_JSON;
1513 goto incr_m_json; 1652 goto incr_m_json;
1653 }
1514 } 1654 }
1515 else if (!*p) 1655 else if (!*p)
1516 goto interrupt; 1656 goto interrupt;
1517 1657
1518 ++p; 1658 ++p;
… …
1524 goto interrupt; 1664 goto interrupt;
1525 1665
1526 ++p; 1666 ++p;
1527 self->incr_mode = INCR_M_STR; 1667 self->incr_mode = INCR_M_STR;
1528 goto incr_m_str; 1668 goto incr_m_str;
1669
1670 // inside #-style comments
1671 case INCR_M_C0:
1672 case INCR_M_C1:
1673 incr_m_c:
1674 for (;;)
1675 {
1676 if (*p == '\n')
1677 {
1678 self->incr_mode = self->incr_mode == INCR_M_C0 ? INCR_M_WS : INCR_M_JSON;
1679 break;
1680 }
1681 else if (!*p)
1682 goto interrupt;
1683
1684 ++p;
1685 }
1686
1687 break;
1529 1688
1530 // inside a string 1689 // inside a string
1531 case INCR_M_STR: 1690 case INCR_M_STR:
1532 incr_m_str: 1691 incr_m_str:
1533 for (;;) 1692 for (;;)
… …
1590 croak (ERR_NESTING_EXCEEDED); 1749 croak (ERR_NESTING_EXCEEDED);
1591 break; 1750 break;
1592 1751
1593 case ']': 1752 case ']':
1594 case '}': 1753 case '}':
1595 if (!--self->incr_nest) 1754 if (--self->incr_nest <= 0)
1596 goto interrupt; 1755 goto interrupt;
1756 break;
1757
1758 case '#':
1759 self->incr_mode = INCR_M_C1;
1760 goto incr_m_c;
1597 } 1761 }
1598 } 1762 }
1599 } 1763 }
1600 1764
1601 modechange: 1765 modechange:
1602 ; 1766 ;
1603 } 1767 }
1604 1768
1605interrupt: 1769interrupt:
1606 self->incr_pos = p - SvPVX (self->incr_text); 1770 self->incr_pos = p - SvPVX (self->incr_text);
1771 //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 1772 //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D
1608} 1773}
1609 1774
1610///////////////////////////////////////////////////////////////////////////// 1775/////////////////////////////////////////////////////////////////////////////
1611// XS interface functions 1776// XS interface functions
… …
1626 json_stash = gv_stashpv ("JSON::XS" , 1); 1791 json_stash = gv_stashpv ("JSON::XS" , 1);
1627 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); 1792 json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1);
1628 1793
1629 json_true = get_bool ("JSON::XS::true"); 1794 json_true = get_bool ("JSON::XS::true");
1630 json_false = get_bool ("JSON::XS::false"); 1795 json_false = get_bool ("JSON::XS::false");
1796
1797 CvNODEBUG_on (get_cv ("JSON::XS::incr_text", 0)); /* the debugger completely breaks lvalue subs */
1631} 1798}
1632 1799
1633PROTOTYPES: DISABLE 1800PROTOTYPES: DISABLE
1634 1801
1635void CLONE (...) 1802void CLONE (...)
… …
1755 XPUSHs (decode_json (jsonstr, self, 0)); 1922 XPUSHs (decode_json (jsonstr, self, 0));
1756 1923
1757void decode_prefix (JSON *self, SV *jsonstr) 1924void decode_prefix (JSON *self, SV *jsonstr)
1758 PPCODE: 1925 PPCODE:
1759{ 1926{
1760 STRLEN offset; 1927 char *offset;
1761 EXTEND (SP, 2); 1928 EXTEND (SP, 2);
1762 PUSHs (decode_json (jsonstr, self, &offset)); 1929 PUSHs (decode_json (jsonstr, self, &offset));
1763 PUSHs (sv_2mortal (newSVuv (offset))); 1930 PUSHs (sv_2mortal (newSVuv (ptr_to_index (jsonstr, offset))));
1764} 1931}
1765 1932
1766void incr_parse (JSON *self, SV *jsonstr = 0) 1933void incr_parse (JSON *self, SV *jsonstr = 0)
1767 PPCODE: 1934 PPCODE:
1768{ 1935{
1769 if (!self->incr_text) 1936 if (!self->incr_text)
1770 self->incr_text = newSVpvn ("", 0); 1937 self->incr_text = newSVpvn ("", 0);
1938
1939 /* if utf8-ness doesn't match the decoder, need to upgrade/downgrade */
1940 if (!DECODE_WANTS_OCTETS (self) == !SvUTF8 (self->incr_text))
1941 if (DECODE_WANTS_OCTETS (self))
1942 {
1943 if (self->incr_pos)
1944 self->incr_pos = utf8_length ((U8 *)SvPVX (self->incr_text),
1945 (U8 *)SvPVX (self->incr_text) + self->incr_pos);
1946
1947 sv_utf8_downgrade (self->incr_text, 0);
1948 }
1949 else
1950 {
1951 sv_utf8_upgrade (self->incr_text);
1952
1953 if (self->incr_pos)
1954 self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos)
1955 - (U8 *)SvPVX (self->incr_text);
1956 }
1771 1957
1772 // append data, if any 1958 // append data, if any
1773 if (jsonstr) 1959 if (jsonstr)
1774 { 1960 {
1961 /* make sure both strings have same encoding */
1775 if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) 1962 if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text))
1963 if (SvUTF8 (jsonstr))
1964 sv_utf8_downgrade (jsonstr, 0);
1776 { 1965 else
1777 /* utf-8-ness differs, need to upgrade */
1778 sv_utf8_upgrade (self->incr_text); 1966 sv_utf8_upgrade (jsonstr);
1779 1967
1780 if (self->incr_pos) 1968 /* 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 { 1969 {
1786 STRLEN len; 1970 STRLEN len;
1787 const char *str = SvPV (jsonstr, len); 1971 const char *str = SvPV (jsonstr, len);
1788 SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); 1972 STRLEN cur = SvCUR (self->incr_text);
1973
1974 if (SvLEN (self->incr_text) <= cur + len)
1975 SvGROW (self->incr_text, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
1976
1789 Move (str, SvEND (self->incr_text), len, char); 1977 Move (str, SvEND (self->incr_text), len, char);
1790 SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); 1978 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 1979 *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there
1792 } 1980 }
1793 } 1981 }
1794 1982
1795 if (GIMME_V != G_VOID) 1983 if (GIMME_V != G_VOID)
1796 do 1984 do
1797 { 1985 {
1798 STRLEN offset; 1986 char *offset;
1799 1987
1800 if (!INCR_DONE (self)) 1988 if (!INCR_DONE (self))
1801 { 1989 {
1802 incr_parse (self); 1990 incr_parse (self);
1803 1991
1804 if (self->incr_pos > self->max_size && self->max_size) 1992 if (self->incr_pos > self->max_size && self->max_size)
1805 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu", 1993 croak ("attempted decode of JSON text of %lu bytes size, but max_size is set to %lu",
1806 (unsigned long)self->incr_pos, (unsigned long)self->max_size); 1994 (unsigned long)self->incr_pos, (unsigned long)self->max_size);
1807 1995
1808 if (!INCR_DONE (self)) 1996 if (!INCR_DONE (self))
1997 {
1998 // as an optimisation, do not accumulate white space in the incr buffer
1999 if (self->incr_mode == INCR_M_WS && self->incr_pos)
2000 {
2001 self->incr_pos = 0;
2002 SvCUR_set (self->incr_text, 0);
2003 }
2004
1809 break; 2005 break;
2006 }
1810 } 2007 }
1811 2008
1812 XPUSHs (decode_json (self->incr_text, self, &offset)); 2009 XPUSHs (decode_json (self->incr_text, self, &offset));
1813 2010
1814 sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset);
1815 self->incr_pos -= offset; 2011 self->incr_pos -= offset - SvPVX (self->incr_text);
1816 self->incr_nest = 0; 2012 self->incr_nest = 0;
1817 self->incr_mode = 0; 2013 self->incr_mode = 0;
2014
2015 sv_chop (self->incr_text, offset);
1818 } 2016 }
1819 while (GIMME_V == G_ARRAY); 2017 while (GIMME_V == G_ARRAY);
1820} 2018}
1821 2019
1822SV *incr_text (JSON *self) 2020SV *incr_text (JSON *self)
… …
1841 self->incr_nest = 0; 2039 self->incr_nest = 0;
1842 self->incr_mode = 0; 2040 self->incr_mode = 0;
1843 } 2041 }
1844} 2042}
1845 2043
2044void incr_reset (JSON *self)
2045 CODE:
2046{
2047 SvREFCNT_dec (self->incr_text);
2048 self->incr_text = 0;
2049 self->incr_pos = 0;
2050 self->incr_nest = 0;
2051 self->incr_mode = 0;
2052}
2053
1846void DESTROY (JSON *self) 2054void DESTROY (JSON *self)
1847 CODE: 2055 CODE:
1848 SvREFCNT_dec (self->cb_sk_object); 2056 SvREFCNT_dec (self->cb_sk_object);
1849 SvREFCNT_dec (self->cb_object); 2057 SvREFCNT_dec (self->cb_object);
1850 SvREFCNT_dec (self->incr_text); 2058 SvREFCNT_dec (self->incr_text);
… …
1873 json_init (&json); 2081 json_init (&json);
1874 json.flags |= ix; 2082 json.flags |= ix;
1875 XPUSHs (decode_json (jsonstr, &json, 0)); 2083 XPUSHs (decode_json (jsonstr, &json, 0));
1876} 2084}
1877 2085
1878

Diff Legend

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