… | |
… | |
42 | #define INIT_SIZE 32 // initial scalar size to be allocated |
42 | #define INIT_SIZE 32 // initial scalar size to be allocated |
43 | #define INDENT_STEP 3 // spaces per indentation level |
43 | #define INDENT_STEP 3 // spaces per indentation level |
44 | |
44 | |
45 | #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 |
46 | |
46 | |
|
|
47 | #define DECODE_WANTS_OCTETS(json) ((json)->flags & F_UTF8) |
|
|
48 | |
47 | #define SB do { |
49 | #define SB do { |
48 | #define SE } while (0) |
50 | #define SE } while (0) |
49 | |
51 | |
50 | #if __GNUC__ >= 3 |
52 | #if __GNUC__ >= 3 |
51 | # define expect(expr,value) __builtin_expect ((expr), (value)) |
53 | # define expect(expr,value) __builtin_expect ((expr), (value)) |
… | |
… | |
187 | return SvUTF8 (sv) |
189 | return SvUTF8 (sv) |
188 | ? utf8_distance (offset, SvPVX (sv)) |
190 | ? utf8_distance (offset, SvPVX (sv)) |
189 | : offset - SvPVX (sv); |
191 | : offset - SvPVX (sv); |
190 | } |
192 | } |
191 | |
193 | |
|
|
194 | ///////////////////////////////////////////////////////////////////////////// |
|
|
195 | // fp hell |
|
|
196 | |
|
|
197 | // scan a group of digits, and a trailing exponent |
|
|
198 | static void |
|
|
199 | json_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 | |
|
|
271 | static NV |
|
|
272 | json_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 | } |
192 | ///////////////////////////////////////////////////////////////////////////// |
289 | ///////////////////////////////////////////////////////////////////////////// |
193 | // encoder |
290 | // encoder |
194 | |
291 | |
195 | // structure used for encoding JSON |
292 | // structure used for encoding JSON |
196 | typedef struct |
293 | typedef struct |
… | |
… | |
1119 | } |
1216 | } |
1120 | |
1217 | |
1121 | len -= *start == '-' ? 1 : 0; |
1218 | len -= *start == '-' ? 1 : 0; |
1122 | |
1219 | |
1123 | // does not fit into IV or UV, try NV |
1220 | // does not fit into IV or UV, try NV |
1124 | if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) |
1221 | if (len <= NV_DIG) |
1125 | #if defined (LDBL_DIG) |
|
|
1126 | || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len) |
|
|
1127 | #endif |
|
|
1128 | ) |
|
|
1129 | // fits into NV without loss of precision |
1222 | // fits into NV without loss of precision |
1130 | return newSVnv (Atof (start)); |
1223 | return newSVnv (json_atof (start)); |
1131 | |
1224 | |
1132 | // everything else fails, convert it to a string |
1225 | // everything else fails, convert it to a string |
1133 | return newSVpvn (start, dec->cur - start); |
1226 | return newSVpvn (start, dec->cur - start); |
1134 | } |
1227 | } |
1135 | |
1228 | |
1136 | // loss of precision here |
1229 | // loss of precision here |
1137 | return newSVnv (Atof (start)); |
1230 | return newSVnv (json_atof (start)); |
1138 | |
1231 | |
1139 | fail: |
1232 | fail: |
1140 | return 0; |
1233 | return 0; |
1141 | } |
1234 | } |
1142 | |
1235 | |
… | |
… | |
1311 | dSP; |
1404 | dSP; |
1312 | int count; |
1405 | int count; |
1313 | |
1406 | |
1314 | ENTER; SAVETMPS; PUSHMARK (SP); |
1407 | ENTER; SAVETMPS; PUSHMARK (SP); |
1315 | XPUSHs (HeVAL (he)); |
1408 | XPUSHs (HeVAL (he)); |
|
|
1409 | sv_2mortal (sv); |
1316 | |
1410 | |
1317 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1411 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1318 | |
1412 | |
1319 | if (count == 1) |
1413 | if (count == 1) |
1320 | { |
1414 | { |
1321 | sv = newSVsv (POPs); |
1415 | sv = newSVsv (POPs); |
1322 | FREETMPS; LEAVE; |
1416 | FREETMPS; LEAVE; |
1323 | return sv; |
1417 | return sv; |
1324 | } |
1418 | } |
1325 | |
1419 | |
|
|
1420 | SvREFCNT_inc (sv); |
1326 | FREETMPS; LEAVE; |
1421 | FREETMPS; LEAVE; |
1327 | } |
1422 | } |
1328 | } |
1423 | } |
1329 | |
1424 | |
1330 | if (dec->json.cb_object) |
1425 | if (dec->json.cb_object) |
… | |
… | |
1426 | { |
1521 | { |
1427 | dec_t dec; |
1522 | dec_t dec; |
1428 | SV *sv; |
1523 | SV *sv; |
1429 | |
1524 | |
1430 | /* work around bugs in 5.10 where manipulating magic values |
1525 | /* work around bugs in 5.10 where manipulating magic values |
1431 | * will perl ignore the magic in subsequent accesses |
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). |
1432 | */ |
1529 | */ |
1433 | /*SvGETMAGIC (string);*/ |
1530 | /*SvGETMAGIC (string);*/ |
1434 | if (SvMAGICAL (string)) |
1531 | if (SvMAGICAL (string) || !SvPOK (string)) |
1435 | string = sv_2mortal (newSVsv (string)); |
1532 | string = sv_2mortal (newSVsv (string)); |
1436 | |
1533 | |
1437 | SvUPGRADE (string, SVt_PV); |
1534 | SvUPGRADE (string, SVt_PV); |
1438 | |
1535 | |
1439 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
1536 | /* work around a bug in perl 5.10, which causes SvCUR to fail an |
… | |
… | |
1456 | if (offset > json->max_size && json->max_size) |
1553 | if (offset > json->max_size && json->max_size) |
1457 | 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", |
1458 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
1555 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
1459 | } |
1556 | } |
1460 | |
1557 | |
1461 | if (json->flags & F_UTF8) |
1558 | if (DECODE_WANTS_OCTETS (json)) |
1462 | sv_utf8_downgrade (string, 0); |
1559 | sv_utf8_downgrade (string, 0); |
1463 | else |
1560 | else |
1464 | sv_utf8_upgrade (string); |
1561 | sv_utf8_upgrade (string); |
1465 | |
1562 | |
1466 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1563 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
… | |
… | |
1837 | PPCODE: |
1934 | PPCODE: |
1838 | { |
1935 | { |
1839 | if (!self->incr_text) |
1936 | if (!self->incr_text) |
1840 | self->incr_text = newSVpvn ("", 0); |
1937 | self->incr_text = newSVpvn ("", 0); |
1841 | |
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 | } |
|
|
1957 | |
1842 | // append data, if any |
1958 | // append data, if any |
1843 | if (jsonstr) |
1959 | if (jsonstr) |
1844 | { |
1960 | { |
|
|
1961 | /* make sure both strings have same encoding */ |
|
|
1962 | if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text)) |
1845 | if (SvUTF8 (jsonstr)) |
1963 | if (SvUTF8 (jsonstr)) |
|
|
1964 | sv_utf8_downgrade (jsonstr, 0); |
1846 | { |
1965 | else |
1847 | if (!SvUTF8 (self->incr_text)) |
|
|
1848 | { |
|
|
1849 | /* utf-8-ness differs, need to upgrade */ |
|
|
1850 | sv_utf8_upgrade (self->incr_text); |
|
|
1851 | |
|
|
1852 | if (self->incr_pos) |
|
|
1853 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
|
|
1854 | - (U8 *)SvPVX (self->incr_text); |
|
|
1855 | } |
|
|
1856 | } |
|
|
1857 | else if (SvUTF8 (self->incr_text)) |
|
|
1858 | sv_utf8_upgrade (jsonstr); |
1966 | sv_utf8_upgrade (jsonstr); |
1859 | |
1967 | |
|
|
1968 | /* and then just blindly append */ |
1860 | { |
1969 | { |
1861 | STRLEN len; |
1970 | STRLEN len; |
1862 | const char *str = SvPV (jsonstr, len); |
1971 | const char *str = SvPV (jsonstr, len); |
1863 | STRLEN cur = SvCUR (self->incr_text); |
1972 | STRLEN cur = SvCUR (self->incr_text); |
1864 | |
1973 | |
… | |
… | |
1883 | if (self->incr_pos > self->max_size && self->max_size) |
1992 | if (self->incr_pos > self->max_size && self->max_size) |
1884 | 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", |
1885 | (unsigned long)self->incr_pos, (unsigned long)self->max_size); |
1994 | (unsigned long)self->incr_pos, (unsigned long)self->max_size); |
1886 | |
1995 | |
1887 | 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) |
|
|
2000 | SvCUR_set (self->incr_text, 0); |
|
|
2001 | |
1888 | break; |
2002 | break; |
|
|
2003 | } |
1889 | } |
2004 | } |
1890 | |
2005 | |
1891 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
2006 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
1892 | |
2007 | |
1893 | self->incr_pos -= offset - SvPVX (self->incr_text); |
2008 | self->incr_pos -= offset - SvPVX (self->incr_text); |
… | |
… | |
1963 | json_init (&json); |
2078 | json_init (&json); |
1964 | json.flags |= ix; |
2079 | json.flags |= ix; |
1965 | XPUSHs (decode_json (jsonstr, &json, 0)); |
2080 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1966 | } |
2081 | } |
1967 | |
2082 | |
1968 | |
|
|