… | |
… | |
17 | // guarantees, 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 | |
77 | enum { |
80 | enum { |
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 | |
… | |
… | |
184 | return SvUTF8 (sv) |
189 | return SvUTF8 (sv) |
185 | ? utf8_distance (offset, SvPVX (sv)) |
190 | ? utf8_distance (offset, SvPVX (sv)) |
186 | : offset - SvPVX (sv); |
191 | : offset - SvPVX (sv); |
187 | } |
192 | } |
188 | |
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 | if (postdp) *expo -= eaccum; |
|
|
266 | *accum += uaccum * Perl_pow (10., *expo); |
|
|
267 | *expo += eaccum; |
|
|
268 | } |
|
|
269 | |
|
|
270 | static NV |
|
|
271 | json_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 | } |
189 | ///////////////////////////////////////////////////////////////////////////// |
288 | ///////////////////////////////////////////////////////////////////////////// |
190 | // encoder |
289 | // encoder |
191 | |
290 | |
192 | // structure used for encoding JSON |
291 | // structure used for encoding JSON |
193 | typedef struct |
292 | typedef struct |
… | |
… | |
949 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
1048 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
950 | *cur++ = ch; |
1049 | *cur++ = ch; |
951 | else if (ch >= 0x80) |
1050 | else if (ch >= 0x80) |
952 | { |
1051 | { |
953 | STRLEN clen; |
1052 | STRLEN clen; |
954 | UV uch; |
|
|
955 | |
1053 | |
956 | --dec_cur; |
1054 | --dec_cur; |
957 | |
1055 | |
958 | uch = decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
1056 | decode_utf8 (dec_cur, dec->end - dec_cur, &clen); |
959 | if (clen == (STRLEN)-1) |
1057 | if (clen == (STRLEN)-1) |
960 | ERR ("malformed UTF-8 character in JSON string"); |
1058 | ERR ("malformed UTF-8 character in JSON string"); |
961 | |
1059 | |
962 | do |
1060 | do |
963 | *cur++ = *dec_cur++; |
1061 | *cur++ = *dec_cur++; |
… | |
… | |
1117 | } |
1215 | } |
1118 | |
1216 | |
1119 | len -= *start == '-' ? 1 : 0; |
1217 | len -= *start == '-' ? 1 : 0; |
1120 | |
1218 | |
1121 | // does not fit into IV or UV, try NV |
1219 | // does not fit into IV or UV, try NV |
1122 | if ((sizeof (NV) == sizeof (double) && DBL_DIG >= len) |
1220 | if (len <= NV_DIG) |
1123 | #if defined (LDBL_DIG) |
|
|
1124 | || (sizeof (NV) == sizeof (long double) && LDBL_DIG >= len) |
|
|
1125 | #endif |
|
|
1126 | ) |
|
|
1127 | // fits into NV without loss of precision |
1221 | // fits into NV without loss of precision |
1128 | return newSVnv (Atof (start)); |
1222 | return newSVnv (json_atof (start)); |
1129 | |
1223 | |
1130 | // everything else fails, convert it to a string |
1224 | // everything else fails, convert it to a string |
1131 | return newSVpvn (start, dec->cur - start); |
1225 | return newSVpvn (start, dec->cur - start); |
1132 | } |
1226 | } |
1133 | |
1227 | |
1134 | // loss of precision here |
1228 | // loss of precision here |
1135 | return newSVnv (Atof (start)); |
1229 | return newSVnv (json_atof (start)); |
1136 | |
1230 | |
1137 | fail: |
1231 | fail: |
1138 | return 0; |
1232 | return 0; |
1139 | } |
1233 | } |
1140 | |
1234 | |
… | |
… | |
1309 | dSP; |
1403 | dSP; |
1310 | int count; |
1404 | int count; |
1311 | |
1405 | |
1312 | ENTER; SAVETMPS; PUSHMARK (SP); |
1406 | ENTER; SAVETMPS; PUSHMARK (SP); |
1313 | XPUSHs (HeVAL (he)); |
1407 | XPUSHs (HeVAL (he)); |
|
|
1408 | sv_2mortal (sv); |
1314 | |
1409 | |
1315 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1410 | PUTBACK; count = call_sv (HeVAL (cb), G_ARRAY); SPAGAIN; |
1316 | |
1411 | |
1317 | if (count == 1) |
1412 | if (count == 1) |
1318 | { |
1413 | { |
1319 | sv = newSVsv (POPs); |
1414 | sv = newSVsv (POPs); |
1320 | FREETMPS; LEAVE; |
1415 | FREETMPS; LEAVE; |
1321 | return sv; |
1416 | return sv; |
1322 | } |
1417 | } |
1323 | |
1418 | |
|
|
1419 | SvREFCNT_inc (sv); |
1324 | FREETMPS; LEAVE; |
1420 | FREETMPS; LEAVE; |
1325 | } |
1421 | } |
1326 | } |
1422 | } |
1327 | |
1423 | |
1328 | if (dec->json.cb_object) |
1424 | if (dec->json.cb_object) |
… | |
… | |
1424 | { |
1520 | { |
1425 | dec_t dec; |
1521 | dec_t dec; |
1426 | SV *sv; |
1522 | SV *sv; |
1427 | |
1523 | |
1428 | /* work around bugs in 5.10 where manipulating magic values |
1524 | /* work around bugs in 5.10 where manipulating magic values |
1429 | * will perl ignore the magic in subsequent accesses |
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). |
1430 | */ |
1528 | */ |
1431 | /*SvGETMAGIC (string);*/ |
1529 | /*SvGETMAGIC (string);*/ |
1432 | if (SvMAGICAL (string)) |
1530 | if (SvMAGICAL (string) || !SvPOK (string)) |
1433 | string = sv_2mortal (newSVsv (string)); |
1531 | string = sv_2mortal (newSVsv (string)); |
1434 | |
1532 | |
1435 | SvUPGRADE (string, SVt_PV); |
1533 | SvUPGRADE (string, SVt_PV); |
1436 | |
1534 | |
1437 | /* 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 |
… | |
… | |
1454 | if (offset > json->max_size && json->max_size) |
1552 | if (offset > json->max_size && json->max_size) |
1455 | 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", |
1456 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
1554 | (unsigned long)SvCUR (string), (unsigned long)json->max_size); |
1457 | } |
1555 | } |
1458 | |
1556 | |
1459 | if (json->flags & F_UTF8) |
1557 | if (DECODE_WANTS_OCTETS (json)) |
1460 | sv_utf8_downgrade (string, 0); |
1558 | sv_utf8_downgrade (string, 0); |
1461 | else |
1559 | else |
1462 | sv_utf8_upgrade (string); |
1560 | sv_utf8_upgrade (string); |
1463 | |
1561 | |
1464 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
1562 | SvGROW (string, SvCUR (string) + 1); // should basically be a NOP |
… | |
… | |
1526 | static void |
1624 | static void |
1527 | incr_parse (JSON *self) |
1625 | incr_parse (JSON *self) |
1528 | { |
1626 | { |
1529 | const char *p = SvPVX (self->incr_text) + self->incr_pos; |
1627 | const char *p = SvPVX (self->incr_text) + self->incr_pos; |
1530 | |
1628 | |
|
|
1629 | // the state machine here is a bit convoluted and could be simplified a lot |
|
|
1630 | // but this would make it slower, so... |
|
|
1631 | |
1531 | for (;;) |
1632 | for (;;) |
1532 | { |
1633 | { |
1533 | //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 |
1534 | switch (self->incr_mode) |
1635 | switch (self->incr_mode) |
1535 | { |
1636 | { |
1536 | // only used for intiial whitespace skipping |
1637 | // only used for initial whitespace skipping |
1537 | case INCR_M_WS: |
1638 | case INCR_M_WS: |
1538 | for (;;) |
1639 | for (;;) |
1539 | { |
1640 | { |
1540 | if (*p > 0x20) |
1641 | if (*p > 0x20) |
1541 | { |
1642 | { |
|
|
1643 | if (*p == '#') |
|
|
1644 | { |
|
|
1645 | self->incr_mode = INCR_M_C0; |
|
|
1646 | goto incr_m_c; |
|
|
1647 | } |
|
|
1648 | else |
|
|
1649 | { |
1542 | self->incr_mode = INCR_M_JSON; |
1650 | self->incr_mode = INCR_M_JSON; |
1543 | goto incr_m_json; |
1651 | goto incr_m_json; |
|
|
1652 | } |
1544 | } |
1653 | } |
1545 | else if (!*p) |
1654 | else if (!*p) |
1546 | goto interrupt; |
1655 | goto interrupt; |
1547 | |
1656 | |
1548 | ++p; |
1657 | ++p; |
… | |
… | |
1554 | goto interrupt; |
1663 | goto interrupt; |
1555 | |
1664 | |
1556 | ++p; |
1665 | ++p; |
1557 | self->incr_mode = INCR_M_STR; |
1666 | self->incr_mode = INCR_M_STR; |
1558 | 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; |
1559 | |
1687 | |
1560 | // inside a string |
1688 | // inside a string |
1561 | case INCR_M_STR: |
1689 | case INCR_M_STR: |
1562 | incr_m_str: |
1690 | incr_m_str: |
1563 | for (;;) |
1691 | for (;;) |
… | |
… | |
1622 | |
1750 | |
1623 | case ']': |
1751 | case ']': |
1624 | case '}': |
1752 | case '}': |
1625 | if (--self->incr_nest <= 0) |
1753 | if (--self->incr_nest <= 0) |
1626 | goto interrupt; |
1754 | goto interrupt; |
|
|
1755 | break; |
|
|
1756 | |
|
|
1757 | case '#': |
|
|
1758 | self->incr_mode = INCR_M_C1; |
|
|
1759 | goto incr_m_c; |
1627 | } |
1760 | } |
1628 | } |
1761 | } |
1629 | } |
1762 | } |
1630 | |
1763 | |
1631 | modechange: |
1764 | modechange: |
1632 | ; |
1765 | ; |
1633 | } |
1766 | } |
1634 | |
1767 | |
1635 | interrupt: |
1768 | interrupt: |
1636 | 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 |
1637 | //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 |
1638 | } |
1772 | } |
1639 | |
1773 | |
1640 | ///////////////////////////////////////////////////////////////////////////// |
1774 | ///////////////////////////////////////////////////////////////////////////// |
1641 | // XS interface functions |
1775 | // XS interface functions |
… | |
… | |
1799 | PPCODE: |
1933 | PPCODE: |
1800 | { |
1934 | { |
1801 | if (!self->incr_text) |
1935 | if (!self->incr_text) |
1802 | self->incr_text = newSVpvn ("", 0); |
1936 | self->incr_text = newSVpvn ("", 0); |
1803 | |
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 | } |
|
|
1956 | |
1804 | // append data, if any |
1957 | // append data, if any |
1805 | if (jsonstr) |
1958 | if (jsonstr) |
1806 | { |
1959 | { |
|
|
1960 | /* make sure both strings have same encoding */ |
|
|
1961 | if (SvUTF8 (jsonstr) != SvUTF8 (self->incr_text)) |
1807 | if (SvUTF8 (jsonstr)) |
1962 | if (SvUTF8 (jsonstr)) |
|
|
1963 | sv_utf8_downgrade (jsonstr, 0); |
1808 | { |
1964 | else |
1809 | if (!SvUTF8 (self->incr_text)) |
|
|
1810 | { |
|
|
1811 | /* utf-8-ness differs, need to upgrade */ |
|
|
1812 | sv_utf8_upgrade (self->incr_text); |
|
|
1813 | |
|
|
1814 | if (self->incr_pos) |
|
|
1815 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
|
|
1816 | - (U8 *)SvPVX (self->incr_text); |
|
|
1817 | } |
|
|
1818 | } |
|
|
1819 | else if (SvUTF8 (self->incr_text)) |
|
|
1820 | sv_utf8_upgrade (jsonstr); |
1965 | sv_utf8_upgrade (jsonstr); |
1821 | |
1966 | |
|
|
1967 | /* and then just blindly append */ |
1822 | { |
1968 | { |
1823 | STRLEN len; |
1969 | STRLEN len; |
1824 | const char *str = SvPV (jsonstr, len); |
1970 | const char *str = SvPV (jsonstr, len); |
1825 | STRLEN cur = SvCUR (self->incr_text); |
1971 | STRLEN cur = SvCUR (self->incr_text); |
1826 | |
1972 | |
… | |
… | |
1925 | json_init (&json); |
2071 | json_init (&json); |
1926 | json.flags |= ix; |
2072 | json.flags |= ix; |
1927 | XPUSHs (decode_json (jsonstr, &json, 0)); |
2073 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1928 | } |
2074 | } |
1929 | |
2075 | |
1930 | |
|
|