… | |
… | |
4 | |
4 | |
5 | #include <assert.h> |
5 | #include <assert.h> |
6 | #include <string.h> |
6 | #include <string.h> |
7 | #include <stdlib.h> |
7 | #include <stdlib.h> |
8 | #include <stdio.h> |
8 | #include <stdio.h> |
|
|
9 | #include <limits.h> |
9 | #include <float.h> |
10 | #include <float.h> |
10 | |
11 | |
11 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
12 | #if defined(__BORLANDC__) || defined(_MSC_VER) |
12 | # define snprintf _snprintf // C compilers have this in stdio.h |
13 | # define snprintf _snprintf // C compilers have this in stdio.h |
13 | #endif |
14 | #endif |
… | |
… | |
15 | // 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 |
16 | // guarentees, though. if it breaks, you get to keep the pieces. |
17 | // guarentees, though. if it breaks, you get to keep the pieces. |
17 | #ifndef UTF8_MAXBYTES |
18 | #ifndef UTF8_MAXBYTES |
18 | # define UTF8_MAXBYTES 13 |
19 | # define UTF8_MAXBYTES 13 |
19 | #endif |
20 | #endif |
|
|
21 | |
|
|
22 | #define IVUV_MAXCHARS (sizeof (UV) * CHAR_BIT * 28 / 93 + 2) |
20 | |
23 | |
21 | #define F_ASCII 0x00000001UL |
24 | #define F_ASCII 0x00000001UL |
22 | #define F_LATIN1 0x00000002UL |
25 | #define F_LATIN1 0x00000002UL |
23 | #define F_UTF8 0x00000004UL |
26 | #define F_UTF8 0x00000004UL |
24 | #define F_INDENT 0x00000008UL |
27 | #define F_INDENT 0x00000008UL |
… | |
… | |
28 | #define F_ALLOW_NONREF 0x00000100UL |
31 | #define F_ALLOW_NONREF 0x00000100UL |
29 | #define F_SHRINK 0x00000200UL |
32 | #define F_SHRINK 0x00000200UL |
30 | #define F_ALLOW_BLESSED 0x00000400UL |
33 | #define F_ALLOW_BLESSED 0x00000400UL |
31 | #define F_CONV_BLESSED 0x00000800UL |
34 | #define F_CONV_BLESSED 0x00000800UL |
32 | #define F_RELAXED 0x00001000UL |
35 | #define F_RELAXED 0x00001000UL |
|
|
36 | #define F_ALLOW_UNKNOWN 0x00002000UL |
33 | |
37 | |
34 | #define F_MAXDEPTH 0xf8000000UL |
38 | #define F_MAXDEPTH 0xf8000000UL |
35 | #define S_MAXDEPTH 27 |
39 | #define S_MAXDEPTH 27 |
36 | #define F_MAXSIZE 0x01f00000UL |
40 | #define F_MAXSIZE 0x01f00000UL |
37 | #define S_MAXSIZE 20 |
41 | #define S_MAXSIZE 20 |
… | |
… | |
75 | #endif |
79 | #endif |
76 | |
80 | |
77 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
81 | static HV *json_stash, *json_boolean_stash; // JSON::XS:: |
78 | static SV *json_true, *json_false; |
82 | static SV *json_true, *json_false; |
79 | |
83 | |
|
|
84 | enum { |
|
|
85 | INCR_M_WS = 0, // initial whitespace skipping, must be 0 |
|
|
86 | INCR_M_STR, // inside string |
|
|
87 | INCR_M_BS, // inside backslash |
|
|
88 | INCR_M_JSON // outside anything, count nesting |
|
|
89 | }; |
|
|
90 | |
|
|
91 | #define INCR_DONE(json) (!(json)->incr_nest && (json)->incr_mode == INCR_M_JSON) |
|
|
92 | |
80 | typedef struct { |
93 | typedef struct { |
81 | U32 flags; |
94 | U32 flags; |
82 | SV *cb_object; |
95 | SV *cb_object; |
83 | HV *cb_sk_object; |
96 | HV *cb_sk_object; |
|
|
97 | |
|
|
98 | // for the incremental parser |
|
|
99 | SV *incr_text; // the source text so far |
|
|
100 | STRLEN incr_pos; // the current offset into the text |
|
|
101 | int incr_nest; // {[]}-nesting level |
|
|
102 | int incr_mode; |
84 | } JSON; |
103 | } JSON; |
85 | |
104 | |
86 | ///////////////////////////////////////////////////////////////////////////// |
105 | ///////////////////////////////////////////////////////////////////////////// |
87 | // utility functions |
106 | // utility functions |
|
|
107 | |
|
|
108 | INLINE SV * |
|
|
109 | get_bool (const char *name) |
|
|
110 | { |
|
|
111 | SV *sv = get_sv (name, 1); |
|
|
112 | |
|
|
113 | SvREADONLY_on (sv); |
|
|
114 | SvREADONLY_on (SvRV (sv)); |
|
|
115 | |
|
|
116 | return sv; |
|
|
117 | } |
88 | |
118 | |
89 | INLINE void |
119 | INLINE void |
90 | shrink (SV *sv) |
120 | shrink (SV *sv) |
91 | { |
121 | { |
92 | sv_utf8_downgrade (sv, 1); |
122 | sv_utf8_downgrade (sv, 1); |
… | |
… | |
123 | // this function takes advantage of this fact, although current gccs |
153 | // this function takes advantage of this fact, although current gccs |
124 | // seem to optimise the check for >= 0x80 away anyways |
154 | // seem to optimise the check for >= 0x80 away anyways |
125 | INLINE unsigned char * |
155 | INLINE unsigned char * |
126 | encode_utf8 (unsigned char *s, UV ch) |
156 | encode_utf8 (unsigned char *s, UV ch) |
127 | { |
157 | { |
128 | if (ch <= 0x7FF) |
158 | if (expect_false (ch < 0x000080)) |
129 | { |
159 | *s++ = ch; |
130 | *s++ = (ch >> 6) | 0xc0; |
160 | else if (expect_true (ch < 0x000800)) |
131 | *s++ = (ch & 0x3f) | 0x80; |
161 | *s++ = 0xc0 | ( ch >> 6), |
132 | } |
162 | *s++ = 0x80 | ( ch & 0x3f); |
133 | else |
163 | else if ( ch < 0x010000) |
134 | s = uvuni_to_utf8_flags (s, ch, 0); |
164 | *s++ = 0xe0 | ( ch >> 12), |
|
|
165 | *s++ = 0x80 | ((ch >> 6) & 0x3f), |
|
|
166 | *s++ = 0x80 | ( ch & 0x3f); |
|
|
167 | else if ( ch < 0x110000) |
|
|
168 | *s++ = 0xf0 | ( ch >> 18), |
|
|
169 | *s++ = 0x80 | ((ch >> 12) & 0x3f), |
|
|
170 | *s++ = 0x80 | ((ch >> 6) & 0x3f), |
|
|
171 | *s++ = 0x80 | ( ch & 0x3f); |
135 | |
172 | |
136 | return s; |
173 | return s; |
137 | } |
174 | } |
138 | |
175 | |
139 | ///////////////////////////////////////////////////////////////////////////// |
176 | ///////////////////////////////////////////////////////////////////////////// |
… | |
… | |
227 | clen = 1; |
264 | clen = 1; |
228 | } |
265 | } |
229 | |
266 | |
230 | if (uch < 0x80/*0x20*/ || uch >= enc->limit) |
267 | if (uch < 0x80/*0x20*/ || uch >= enc->limit) |
231 | { |
268 | { |
232 | if (uch > 0xFFFFUL) |
269 | if (uch >= 0x10000UL) |
233 | { |
270 | { |
234 | if (uch > 0x10FFFFUL) |
271 | if (uch >= 0x110000UL) |
235 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
272 | croak ("out of range codepoint (0x%lx) encountered, unrepresentable in JSON", (unsigned long)uch); |
236 | |
273 | |
237 | need (enc, len += 11); |
274 | need (enc, len += 11); |
238 | sprintf (enc->cur, "\\u%04x\\u%04x", |
275 | sprintf (enc->cur, "\\u%04x\\u%04x", |
239 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
276 | (int)((uch - 0x10000) / 0x400 + 0xD800), |
… | |
… | |
413 | |
450 | |
414 | static void |
451 | static void |
415 | encode_hv (enc_t *enc, HV *hv) |
452 | encode_hv (enc_t *enc, HV *hv) |
416 | { |
453 | { |
417 | HE *he; |
454 | HE *he; |
418 | int count; |
|
|
419 | |
455 | |
420 | if (enc->indent >= enc->maxdepth) |
456 | if (enc->indent >= enc->maxdepth) |
421 | croak ("data structure too deep (hit recursion limit)"); |
457 | croak ("data structure too deep (hit recursion limit)"); |
422 | |
458 | |
423 | encode_ch (enc, '{'); |
459 | encode_ch (enc, '{'); |
… | |
… | |
607 | |
643 | |
608 | if (len == 1 && *pv == '1') |
644 | if (len == 1 && *pv == '1') |
609 | encode_str (enc, "true", 4, 0); |
645 | encode_str (enc, "true", 4, 0); |
610 | else if (len == 1 && *pv == '0') |
646 | else if (len == 1 && *pv == '0') |
611 | encode_str (enc, "false", 5, 0); |
647 | encode_str (enc, "false", 5, 0); |
|
|
648 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
649 | encode_str (enc, "null", 4, 0); |
612 | else |
650 | else |
613 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
651 | croak ("cannot encode reference to scalar '%s' unless the scalar is 0 or 1", |
614 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
652 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
615 | } |
653 | } |
|
|
654 | else if (enc->json.flags & F_ALLOW_UNKNOWN) |
|
|
655 | encode_str (enc, "null", 4, 0); |
616 | else |
656 | else |
617 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
657 | croak ("encountered %s, but JSON can only represent references to arrays or hashes", |
618 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
658 | SvPV_nolen (sv_2mortal (newRV_inc (sv)))); |
619 | } |
659 | } |
620 | |
660 | |
… | |
… | |
638 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
678 | Gconvert (SvNVX (sv), NV_DIG, 0, enc->cur); |
639 | enc->cur += strlen (enc->cur); |
679 | enc->cur += strlen (enc->cur); |
640 | } |
680 | } |
641 | else if (SvIOKp (sv)) |
681 | else if (SvIOKp (sv)) |
642 | { |
682 | { |
643 | // we assume we can always read an IV as a UV |
683 | // we assume we can always read an IV as a UV and vice versa |
644 | if (SvUV (sv) & ~(UV)0x7fff) |
684 | // we assume two's complement |
645 | { |
685 | // we assume no aliasing issues in the union |
646 | // large integer, use the (rather slow) snprintf way. |
686 | if (SvIsUV (sv) ? SvUVX (sv) <= 59000 |
647 | need (enc, sizeof (UV) * 3); |
687 | : SvIVX (sv) <= 59000 && SvIVX (sv) >= -59000) |
648 | enc->cur += |
|
|
649 | SvIsUV(sv) |
|
|
650 | ? snprintf (enc->cur, sizeof (UV) * 3, "%"UVuf, (UV)SvUVX (sv)) |
|
|
651 | : snprintf (enc->cur, sizeof (UV) * 3, "%"IVdf, (IV)SvIVX (sv)); |
|
|
652 | } |
|
|
653 | else |
|
|
654 | { |
688 | { |
655 | // optimise the "small number case" |
689 | // optimise the "small number case" |
656 | // code will likely be branchless and use only a single multiplication |
690 | // code will likely be branchless and use only a single multiplication |
|
|
691 | // works for numbers up to 59074 |
657 | I32 i = SvIV (sv); |
692 | I32 i = SvIVX (sv); |
658 | U32 u; |
693 | U32 u; |
659 | char digit, nz = 0; |
694 | char digit, nz = 0; |
660 | |
695 | |
661 | need (enc, 6); |
696 | need (enc, 6); |
662 | |
697 | |
… | |
… | |
668 | |
703 | |
669 | // now output digit by digit, each time masking out the integer part |
704 | // now output digit by digit, each time masking out the integer part |
670 | // and multiplying by 5 while moving the decimal point one to the right, |
705 | // and multiplying by 5 while moving the decimal point one to the right, |
671 | // resulting in a net multiplication by 10. |
706 | // resulting in a net multiplication by 10. |
672 | // we always write the digit to memory but conditionally increment |
707 | // we always write the digit to memory but conditionally increment |
673 | // the pointer, to ease the usage of conditional move instructions. |
708 | // the pointer, to enable the use of conditional move instructions. |
674 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffff) * 5; |
709 | digit = u >> 28; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0xfffffffUL) * 5; |
675 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffff) * 5; |
710 | digit = u >> 27; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x7ffffffUL) * 5; |
676 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffff) * 5; |
711 | digit = u >> 26; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x3ffffffUL) * 5; |
677 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffff) * 5; |
712 | digit = u >> 25; *enc->cur = digit + '0'; enc->cur += (nz = nz || digit); u = (u & 0x1ffffffUL) * 5; |
678 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
713 | digit = u >> 24; *enc->cur = digit + '0'; enc->cur += 1; // correctly generate '0' |
679 | } |
714 | } |
|
|
715 | else |
|
|
716 | { |
|
|
717 | // large integer, use the (rather slow) snprintf way. |
|
|
718 | need (enc, IVUV_MAXCHARS); |
|
|
719 | enc->cur += |
|
|
720 | SvIsUV(sv) |
|
|
721 | ? snprintf (enc->cur, IVUV_MAXCHARS, "%"UVuf, (UV)SvUVX (sv)) |
|
|
722 | : snprintf (enc->cur, IVUV_MAXCHARS, "%"IVdf, (IV)SvIVX (sv)); |
|
|
723 | } |
680 | } |
724 | } |
681 | else if (SvROK (sv)) |
725 | else if (SvROK (sv)) |
682 | encode_rv (enc, SvRV (sv)); |
726 | encode_rv (enc, SvRV (sv)); |
683 | else if (!SvOK (sv)) |
727 | else if (!SvOK (sv) || enc->json.flags & F_ALLOW_UNKNOWN) |
684 | encode_str (enc, "null", 4, 0); |
728 | encode_str (enc, "null", 4, 0); |
685 | else |
729 | else |
686 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
730 | croak ("encountered perl type (%s,0x%x) that JSON cannot handle, you might want to report this", |
687 | SvPV_nolen (sv), SvFLAGS (sv)); |
731 | SvPV_nolen (sv), SvFLAGS (sv)); |
688 | } |
732 | } |
… | |
… | |
701 | enc.end = SvEND (enc.sv); |
745 | enc.end = SvEND (enc.sv); |
702 | enc.indent = 0; |
746 | enc.indent = 0; |
703 | enc.maxdepth = DEC_DEPTH (enc.json.flags); |
747 | enc.maxdepth = DEC_DEPTH (enc.json.flags); |
704 | enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
748 | enc.limit = enc.json.flags & F_ASCII ? 0x000080UL |
705 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
749 | : enc.json.flags & F_LATIN1 ? 0x000100UL |
706 | : 0x10FFFFUL; |
750 | : 0x110000UL; |
707 | |
751 | |
708 | SvPOK_only (enc.sv); |
752 | SvPOK_only (enc.sv); |
709 | encode_sv (&enc, scalar); |
753 | encode_sv (&enc, scalar); |
710 | |
754 | |
711 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
755 | SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv)); |
… | |
… | |
889 | default: |
933 | default: |
890 | --dec_cur; |
934 | --dec_cur; |
891 | ERR ("illegal backslash escape sequence in string"); |
935 | ERR ("illegal backslash escape sequence in string"); |
892 | } |
936 | } |
893 | } |
937 | } |
894 | else if (expect_true (ch >= 0x20 && ch <= 0x7f)) |
938 | else if (expect_true (ch >= 0x20 && ch < 0x80)) |
895 | *cur++ = ch; |
939 | *cur++ = ch; |
896 | else if (ch >= 0x80) |
940 | else if (ch >= 0x80) |
897 | { |
941 | { |
898 | STRLEN clen; |
942 | STRLEN clen; |
899 | UV uch; |
943 | UV uch; |
… | |
… | |
1157 | char *p = dec->cur; |
1201 | char *p = dec->cur; |
1158 | char *e = p + 24; // only try up to 24 bytes |
1202 | char *e = p + 24; // only try up to 24 bytes |
1159 | |
1203 | |
1160 | for (;;) |
1204 | for (;;) |
1161 | { |
1205 | { |
1162 | // the >= 0x80 is true on most architectures |
1206 | // the >= 0x80 is false on most architectures |
1163 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1207 | if (p == e || *p < 0x20 || *p >= 0x80 || *p == '\\') |
1164 | { |
1208 | { |
1165 | // slow path, back up and use decode_str |
1209 | // slow path, back up and use decode_str |
1166 | SV *key = decode_str (dec); |
1210 | SV *key = decode_str (dec); |
1167 | if (!key) |
1211 | if (!key) |
… | |
… | |
1298 | |
1342 | |
1299 | static SV * |
1343 | static SV * |
1300 | decode_sv (dec_t *dec) |
1344 | decode_sv (dec_t *dec) |
1301 | { |
1345 | { |
1302 | // the beauty of JSON: you need exactly one character lookahead |
1346 | // the beauty of JSON: you need exactly one character lookahead |
1303 | // to parse anything. |
1347 | // to parse everything. |
1304 | switch (*dec->cur) |
1348 | switch (*dec->cur) |
1305 | { |
1349 | { |
1306 | case '"': ++dec->cur; return decode_str (dec); |
1350 | case '"': ++dec->cur; return decode_str (dec); |
1307 | case '[': ++dec->cur; return decode_av (dec); |
1351 | case '[': ++dec->cur; return decode_av (dec); |
1308 | case '{': ++dec->cur; return decode_hv (dec); |
1352 | case '{': ++dec->cur; return decode_hv (dec); |
1309 | |
1353 | |
1310 | case '-': |
1354 | case '-': |
1311 | case '0': case '1': case '2': case '3': case '4': |
1355 | case '0': case '1': case '2': case '3': case '4': |
1312 | case '5': case '6': case '7': case '8': case '9': |
1356 | case '5': case '6': case '7': case '8': case '9': |
1313 | return decode_num (dec); |
1357 | return decode_num (dec); |
… | |
… | |
1315 | case 't': |
1359 | case 't': |
1316 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1360 | if (dec->end - dec->cur >= 4 && !memcmp (dec->cur, "true", 4)) |
1317 | { |
1361 | { |
1318 | dec->cur += 4; |
1362 | dec->cur += 4; |
1319 | #if JSON_SLOW |
1363 | #if JSON_SLOW |
1320 | json_true = get_sv ("JSON::XS::true", 1); SvREADONLY_on (json_true); |
1364 | json_true = get_bool ("JSON::XS::true"); |
1321 | #endif |
1365 | #endif |
1322 | return SvREFCNT_inc (json_true); |
1366 | return newSVsv (json_true); |
1323 | } |
1367 | } |
1324 | else |
1368 | else |
1325 | ERR ("'true' expected"); |
1369 | ERR ("'true' expected"); |
1326 | |
1370 | |
1327 | break; |
1371 | break; |
… | |
… | |
1329 | case 'f': |
1373 | case 'f': |
1330 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1374 | if (dec->end - dec->cur >= 5 && !memcmp (dec->cur, "false", 5)) |
1331 | { |
1375 | { |
1332 | dec->cur += 5; |
1376 | dec->cur += 5; |
1333 | #if JSON_SLOW |
1377 | #if JSON_SLOW |
1334 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1378 | json_false = get_bool ("JSON::XS::false"); |
1335 | #endif |
1379 | #endif |
1336 | return SvREFCNT_inc (json_false); |
1380 | return newSVsv (json_false); |
1337 | } |
1381 | } |
1338 | else |
1382 | else |
1339 | ERR ("'false' expected"); |
1383 | ERR ("'false' expected"); |
1340 | |
1384 | |
1341 | break; |
1385 | break; |
… | |
… | |
1359 | fail: |
1403 | fail: |
1360 | return 0; |
1404 | return 0; |
1361 | } |
1405 | } |
1362 | |
1406 | |
1363 | static SV * |
1407 | static SV * |
1364 | decode_json (SV *string, JSON *json, UV *offset_return) |
1408 | decode_json (SV *string, JSON *json, STRLEN *offset_return) |
1365 | { |
1409 | { |
1366 | dec_t dec; |
1410 | dec_t dec; |
1367 | UV offset; |
1411 | STRLEN offset; |
1368 | SV *sv; |
1412 | SV *sv; |
1369 | |
1413 | |
1370 | SvGETMAGIC (string); |
1414 | SvGETMAGIC (string); |
1371 | SvUPGRADE (string, SVt_PV); |
1415 | SvUPGRADE (string, SVt_PV); |
1372 | |
1416 | |
… | |
… | |
1445 | |
1489 | |
1446 | return sv; |
1490 | return sv; |
1447 | } |
1491 | } |
1448 | |
1492 | |
1449 | ///////////////////////////////////////////////////////////////////////////// |
1493 | ///////////////////////////////////////////////////////////////////////////// |
|
|
1494 | // incremental parser |
|
|
1495 | |
|
|
1496 | static void |
|
|
1497 | incr_parse (JSON *self) |
|
|
1498 | { |
|
|
1499 | const char *p = SvPVX (self->incr_text) + self->incr_pos; |
|
|
1500 | |
|
|
1501 | for (;;) |
|
|
1502 | { |
|
|
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 |
|
|
1504 | switch (self->incr_mode) |
|
|
1505 | { |
|
|
1506 | // only used for intiial whitespace skipping |
|
|
1507 | case INCR_M_WS: |
|
|
1508 | for (;;) |
|
|
1509 | { |
|
|
1510 | if (*p > 0x20) |
|
|
1511 | { |
|
|
1512 | self->incr_mode = INCR_M_JSON; |
|
|
1513 | goto incr_m_json; |
|
|
1514 | } |
|
|
1515 | else if (!*p) |
|
|
1516 | goto interrupt; |
|
|
1517 | |
|
|
1518 | ++p; |
|
|
1519 | } |
|
|
1520 | |
|
|
1521 | // skip a single char inside a string (for \\-processing) |
|
|
1522 | case INCR_M_BS: |
|
|
1523 | if (!*p) |
|
|
1524 | goto interrupt; |
|
|
1525 | |
|
|
1526 | ++p; |
|
|
1527 | self->incr_mode = INCR_M_STR; |
|
|
1528 | goto incr_m_str; |
|
|
1529 | |
|
|
1530 | // inside a string |
|
|
1531 | case INCR_M_STR: |
|
|
1532 | incr_m_str: |
|
|
1533 | for (;;) |
|
|
1534 | { |
|
|
1535 | if (*p == '"') |
|
|
1536 | { |
|
|
1537 | ++p; |
|
|
1538 | self->incr_mode = INCR_M_JSON; |
|
|
1539 | |
|
|
1540 | if (!self->incr_nest) |
|
|
1541 | goto interrupt; |
|
|
1542 | |
|
|
1543 | goto incr_m_json; |
|
|
1544 | } |
|
|
1545 | else if (*p == '\\') |
|
|
1546 | { |
|
|
1547 | ++p; // "virtually" consumes character after \ |
|
|
1548 | |
|
|
1549 | if (!*p) // if at end of string we have to switch modes |
|
|
1550 | { |
|
|
1551 | self->incr_mode = INCR_M_BS; |
|
|
1552 | goto interrupt; |
|
|
1553 | } |
|
|
1554 | } |
|
|
1555 | else if (!*p) |
|
|
1556 | goto interrupt; |
|
|
1557 | |
|
|
1558 | ++p; |
|
|
1559 | } |
|
|
1560 | |
|
|
1561 | // after initial ws, outside string |
|
|
1562 | case INCR_M_JSON: |
|
|
1563 | incr_m_json: |
|
|
1564 | for (;;) |
|
|
1565 | { |
|
|
1566 | switch (*p++) |
|
|
1567 | { |
|
|
1568 | case 0: |
|
|
1569 | --p; |
|
|
1570 | goto interrupt; |
|
|
1571 | |
|
|
1572 | case 0x09: |
|
|
1573 | case 0x0a: |
|
|
1574 | case 0x0d: |
|
|
1575 | case 0x20: |
|
|
1576 | if (!self->incr_nest) |
|
|
1577 | { |
|
|
1578 | --p; // do not eat the whitespace, let the next round do it |
|
|
1579 | goto interrupt; |
|
|
1580 | } |
|
|
1581 | break; |
|
|
1582 | |
|
|
1583 | case '"': |
|
|
1584 | self->incr_mode = INCR_M_STR; |
|
|
1585 | goto incr_m_str; |
|
|
1586 | |
|
|
1587 | case '[': |
|
|
1588 | case '{': |
|
|
1589 | ++self->incr_nest; |
|
|
1590 | break; |
|
|
1591 | |
|
|
1592 | case ']': |
|
|
1593 | case '}': |
|
|
1594 | if (!--self->incr_nest) |
|
|
1595 | goto interrupt; |
|
|
1596 | } |
|
|
1597 | } |
|
|
1598 | } |
|
|
1599 | |
|
|
1600 | modechange: |
|
|
1601 | ; |
|
|
1602 | } |
|
|
1603 | |
|
|
1604 | interrupt: |
|
|
1605 | self->incr_pos = p - SvPVX (self->incr_text); |
|
|
1606 | //printf ("return pos %d mode %d nest %d\n", self->incr_pos, self->incr_mode, self->incr_nest);//D |
|
|
1607 | } |
|
|
1608 | |
|
|
1609 | ///////////////////////////////////////////////////////////////////////////// |
1450 | // XS interface functions |
1610 | // XS interface functions |
1451 | |
1611 | |
1452 | MODULE = JSON::XS PACKAGE = JSON::XS |
1612 | MODULE = JSON::XS PACKAGE = JSON::XS |
1453 | |
1613 | |
1454 | BOOT: |
1614 | BOOT: |
… | |
… | |
1463 | : -1; |
1623 | : -1; |
1464 | |
1624 | |
1465 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1625 | json_stash = gv_stashpv ("JSON::XS" , 1); |
1466 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
1626 | json_boolean_stash = gv_stashpv ("JSON::XS::Boolean", 1); |
1467 | |
1627 | |
1468 | json_true = get_sv ("JSON::XS::true" , 1); SvREADONLY_on (json_true ); |
1628 | json_true = get_bool ("JSON::XS::true"); |
1469 | json_false = get_sv ("JSON::XS::false", 1); SvREADONLY_on (json_false); |
1629 | json_false = get_bool ("JSON::XS::false"); |
1470 | } |
1630 | } |
1471 | |
1631 | |
1472 | PROTOTYPES: DISABLE |
1632 | PROTOTYPES: DISABLE |
1473 | |
1633 | |
1474 | void CLONE (...) |
1634 | void CLONE (...) |
… | |
… | |
1502 | allow_nonref = F_ALLOW_NONREF |
1662 | allow_nonref = F_ALLOW_NONREF |
1503 | shrink = F_SHRINK |
1663 | shrink = F_SHRINK |
1504 | allow_blessed = F_ALLOW_BLESSED |
1664 | allow_blessed = F_ALLOW_BLESSED |
1505 | convert_blessed = F_CONV_BLESSED |
1665 | convert_blessed = F_CONV_BLESSED |
1506 | relaxed = F_RELAXED |
1666 | relaxed = F_RELAXED |
|
|
1667 | allow_unknown = F_ALLOW_UNKNOWN |
1507 | PPCODE: |
1668 | PPCODE: |
1508 | { |
1669 | { |
1509 | if (enable) |
1670 | if (enable) |
1510 | self->flags |= ix; |
1671 | self->flags |= ix; |
1511 | else |
1672 | else |
… | |
… | |
1526 | get_allow_nonref = F_ALLOW_NONREF |
1687 | get_allow_nonref = F_ALLOW_NONREF |
1527 | get_shrink = F_SHRINK |
1688 | get_shrink = F_SHRINK |
1528 | get_allow_blessed = F_ALLOW_BLESSED |
1689 | get_allow_blessed = F_ALLOW_BLESSED |
1529 | get_convert_blessed = F_CONV_BLESSED |
1690 | get_convert_blessed = F_CONV_BLESSED |
1530 | get_relaxed = F_RELAXED |
1691 | get_relaxed = F_RELAXED |
|
|
1692 | get_allow_unknown = F_ALLOW_UNKNOWN |
1531 | PPCODE: |
1693 | PPCODE: |
1532 | XPUSHs (boolSV (self->flags & ix)); |
1694 | XPUSHs (boolSV (self->flags & ix)); |
1533 | |
1695 | |
1534 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
1696 | void max_depth (JSON *self, UV max_depth = 0x80000000UL) |
1535 | PPCODE: |
1697 | PPCODE: |
… | |
… | |
1614 | XPUSHs (decode_json (jsonstr, self, 0)); |
1776 | XPUSHs (decode_json (jsonstr, self, 0)); |
1615 | |
1777 | |
1616 | void decode_prefix (JSON *self, SV *jsonstr) |
1778 | void decode_prefix (JSON *self, SV *jsonstr) |
1617 | PPCODE: |
1779 | PPCODE: |
1618 | { |
1780 | { |
1619 | UV offset; |
1781 | STRLEN offset; |
1620 | EXTEND (SP, 2); |
1782 | EXTEND (SP, 2); |
1621 | PUSHs (decode_json (jsonstr, self, &offset)); |
1783 | PUSHs (decode_json (jsonstr, self, &offset)); |
1622 | PUSHs (sv_2mortal (newSVuv (offset))); |
1784 | PUSHs (sv_2mortal (newSVuv (offset))); |
|
|
1785 | } |
|
|
1786 | |
|
|
1787 | void incr_parse (JSON *self, SV *jsonstr = 0) |
|
|
1788 | PPCODE: |
|
|
1789 | { |
|
|
1790 | if (!self->incr_text) |
|
|
1791 | self->incr_text = newSVpvn ("", 0); |
|
|
1792 | |
|
|
1793 | // append data, if any |
|
|
1794 | if (jsonstr) |
|
|
1795 | { |
|
|
1796 | if (SvUTF8 (jsonstr) && !SvUTF8 (self->incr_text)) |
|
|
1797 | { |
|
|
1798 | /* utf-8-ness differs, need to upgrade */ |
|
|
1799 | sv_utf8_upgrade (self->incr_text); |
|
|
1800 | |
|
|
1801 | if (self->incr_pos) |
|
|
1802 | self->incr_pos = utf8_hop ((U8 *)SvPVX (self->incr_text), self->incr_pos) |
|
|
1803 | - (U8 *)SvPVX (self->incr_text); |
|
|
1804 | } |
|
|
1805 | |
|
|
1806 | { |
|
|
1807 | STRLEN len; |
|
|
1808 | const char *str = SvPV (jsonstr, len); |
|
|
1809 | SvGROW (self->incr_text, SvCUR (self->incr_text) + len + 1); |
|
|
1810 | Move (str, SvEND (self->incr_text), len, char); |
|
|
1811 | SvCUR_set (self->incr_text, SvCUR (self->incr_text) + len); |
|
|
1812 | *SvEND (self->incr_text) = 0; // this should basically be a nop, too, but make sure it's there |
|
|
1813 | } |
|
|
1814 | } |
|
|
1815 | |
|
|
1816 | if (GIMME_V != G_VOID) |
|
|
1817 | do |
|
|
1818 | { |
|
|
1819 | STRLEN offset; |
|
|
1820 | |
|
|
1821 | if (!INCR_DONE (self)) |
|
|
1822 | { |
|
|
1823 | incr_parse (self); |
|
|
1824 | if (!INCR_DONE (self)) |
|
|
1825 | break; |
|
|
1826 | } |
|
|
1827 | |
|
|
1828 | XPUSHs (decode_json (self->incr_text, self, &offset)); |
|
|
1829 | |
|
|
1830 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + offset); |
|
|
1831 | self->incr_pos -= offset; |
|
|
1832 | self->incr_nest = 0; |
|
|
1833 | self->incr_mode = 0; |
|
|
1834 | } |
|
|
1835 | while (GIMME_V == G_ARRAY); |
|
|
1836 | } |
|
|
1837 | |
|
|
1838 | SV *incr_text (JSON *self) |
|
|
1839 | ATTRS: lvalue |
|
|
1840 | CODE: |
|
|
1841 | { |
|
|
1842 | if (self->incr_pos) |
|
|
1843 | croak ("incr_text can not be called when the incremental parser already started parsing"); |
|
|
1844 | |
|
|
1845 | RETVAL = self->incr_text ? SvREFCNT_inc (self->incr_text) : &PL_sv_undef; |
|
|
1846 | } |
|
|
1847 | OUTPUT: |
|
|
1848 | RETVAL |
|
|
1849 | |
|
|
1850 | void incr_skip (JSON *self) |
|
|
1851 | CODE: |
|
|
1852 | { |
|
|
1853 | if (self->incr_pos) |
|
|
1854 | { |
|
|
1855 | sv_chop (self->incr_text, SvPV_nolen (self->incr_text) + self->incr_pos); |
|
|
1856 | self->incr_pos = 0; |
|
|
1857 | self->incr_nest = 0; |
|
|
1858 | self->incr_mode = 0; |
|
|
1859 | } |
1623 | } |
1860 | } |
1624 | |
1861 | |
1625 | void DESTROY (JSON *self) |
1862 | void DESTROY (JSON *self) |
1626 | CODE: |
1863 | CODE: |
1627 | SvREFCNT_dec (self->cb_sk_object); |
1864 | SvREFCNT_dec (self->cb_sk_object); |
1628 | SvREFCNT_dec (self->cb_object); |
1865 | SvREFCNT_dec (self->cb_object); |
|
|
1866 | SvREFCNT_dec (self->incr_text); |
1629 | |
1867 | |
1630 | PROTOTYPES: ENABLE |
1868 | PROTOTYPES: ENABLE |
1631 | |
1869 | |
1632 | void encode_json (SV *scalar) |
1870 | void encode_json (SV *scalar) |
|
|
1871 | ALIAS: |
|
|
1872 | to_json_ = 0 |
|
|
1873 | encode_json = F_UTF8 |
1633 | PPCODE: |
1874 | PPCODE: |
1634 | { |
1875 | { |
1635 | JSON json = { F_DEFAULT | F_UTF8 }; |
1876 | JSON json = { F_DEFAULT | ix }; |
1636 | XPUSHs (encode_json (scalar, &json)); |
1877 | XPUSHs (encode_json (scalar, &json)); |
1637 | } |
1878 | } |
1638 | |
1879 | |
1639 | void decode_json (SV *jsonstr) |
1880 | void decode_json (SV *jsonstr) |
|
|
1881 | ALIAS: |
|
|
1882 | from_json_ = 0 |
|
|
1883 | decode_json = F_UTF8 |
1640 | PPCODE: |
1884 | PPCODE: |
1641 | { |
1885 | { |
1642 | JSON json = { F_DEFAULT | F_UTF8 }; |
1886 | JSON json = { F_DEFAULT | ix }; |
1643 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1887 | XPUSHs (decode_json (jsonstr, &json, 0)); |
1644 | } |
1888 | } |
1645 | |
1889 | |
|
|
1890 | |