… | |
… | |
100 | #define F_SHRINK 0x00000001UL |
100 | #define F_SHRINK 0x00000001UL |
101 | #define F_ALLOW_UNKNOWN 0x00000002UL |
101 | #define F_ALLOW_UNKNOWN 0x00000002UL |
102 | #define F_ALLOW_SHARING 0x00000004UL |
102 | #define F_ALLOW_SHARING 0x00000004UL |
103 | #define F_ALLOW_CYCLES 0x00000008UL |
103 | #define F_ALLOW_CYCLES 0x00000008UL |
104 | #define F_PACK_STRINGS 0x00000010UL |
104 | #define F_PACK_STRINGS 0x00000010UL |
|
|
105 | #define F_TEXT_KEYS 0x00000020UL |
105 | #define F_UTF8_STRINGS 0x00000020UL |
106 | #define F_TEXT_STRINGS 0x00000040UL |
106 | #define F_VALIDATE_UTF8 0x00000040UL |
107 | #define F_VALIDATE_UTF8 0x00000080UL |
107 | |
108 | |
108 | #define INIT_SIZE 32 // initial scalar size to be allocated |
109 | #define INIT_SIZE 32 // initial scalar size to be allocated |
109 | |
110 | |
110 | #define SB do { |
111 | #define SB do { |
111 | #define SE } while (0) |
112 | #define SE } while (0) |
… | |
… | |
277 | encode_tag (enc_t *enc, UV tag) |
278 | encode_tag (enc_t *enc, UV tag) |
278 | { |
279 | { |
279 | encode_uint (enc, MAJOR_TAG, tag); |
280 | encode_uint (enc, MAJOR_TAG, tag); |
280 | } |
281 | } |
281 | |
282 | |
|
|
283 | // exceptional (hopefully) slow path for byte strings that need to be utf8-encoded |
|
|
284 | ecb_noinline static void |
|
|
285 | encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len) |
|
|
286 | { |
|
|
287 | STRLEN ulen = len; |
|
|
288 | U8 *p, *pend = (U8 *)str + len; |
|
|
289 | |
|
|
290 | for (p = (U8 *)str; p < pend; ++p) |
|
|
291 | ulen += *p >> 7; // count set high bits |
|
|
292 | |
|
|
293 | encode_uint (enc, MAJOR_TEXT, ulen); |
|
|
294 | |
|
|
295 | need (enc, ulen); |
|
|
296 | for (p = (U8 *)str; p < pend; ++p) |
|
|
297 | if (*p < 0x80) |
|
|
298 | *enc->cur++ = *p; |
|
|
299 | else |
|
|
300 | { |
|
|
301 | *enc->cur++ = 0xc0 + (*p >> 6); |
|
|
302 | *enc->cur++ = 0x80 + (*p & 63); |
|
|
303 | } |
|
|
304 | } |
|
|
305 | |
282 | ecb_inline void |
306 | ecb_inline void |
283 | encode_str (enc_t *enc, int utf8, char *str, STRLEN len) |
307 | encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) |
284 | { |
308 | { |
285 | if (ecb_expect_false (enc->cbor.flags & F_UTF8_STRINGS)) |
309 | if (ecb_expect_false (upgrade_utf8)) |
286 | if (!utf8) |
310 | if (!utf8) |
287 | { |
311 | { |
288 | SV *sv = sv_newmortal (); |
|
|
289 | char *s; STRLEN l; |
|
|
290 | |
|
|
291 | sv_setpvn (sv, str, len); |
|
|
292 | |
|
|
293 | s = SvPVutf8 (sv, l); |
|
|
294 | encode_str (enc, 1, s, l); |
312 | encode_str_utf8 (enc, utf8, str, len); |
295 | return; |
313 | return; |
296 | } |
314 | } |
297 | |
315 | |
298 | encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); |
316 | encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); |
299 | need (enc, len); |
317 | need (enc, len); |
300 | memcpy (enc->cur, str, len); |
318 | memcpy (enc->cur, str, len); |
301 | enc->cur += len; |
319 | enc->cur += len; |
302 | } |
320 | } |
303 | |
321 | |
304 | static void |
322 | ecb_inline void |
305 | encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) |
323 | encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) |
306 | { |
324 | { |
307 | if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) |
325 | if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) |
308 | { |
326 | { |
309 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
327 | SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); |
310 | |
328 | |
… | |
… | |
321 | sv_setuv (*svp, enc->stringref_idx); |
339 | sv_setuv (*svp, enc->stringref_idx); |
322 | ++enc->stringref_idx; |
340 | ++enc->stringref_idx; |
323 | } |
341 | } |
324 | } |
342 | } |
325 | |
343 | |
326 | encode_str (enc, utf8, str, len); |
344 | encode_str (enc, upgrade_utf8, utf8, str, len); |
327 | } |
345 | } |
328 | |
346 | |
329 | static void encode_sv (enc_t *enc, SV *sv); |
347 | static void encode_sv (enc_t *enc, SV *sv); |
330 | |
348 | |
331 | static void |
349 | static void |
… | |
… | |
377 | while ((he = hv_iternext (hv))) |
395 | while ((he = hv_iternext (hv))) |
378 | { |
396 | { |
379 | if (HeKLEN (he) == HEf_SVKEY) |
397 | if (HeKLEN (he) == HEf_SVKEY) |
380 | encode_sv (enc, HeSVKEY (he)); |
398 | encode_sv (enc, HeSVKEY (he)); |
381 | else |
399 | else |
382 | encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
400 | encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he)); |
383 | |
401 | |
384 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
402 | encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); |
385 | } |
403 | } |
386 | |
404 | |
387 | if (mg) |
405 | if (mg) |
… | |
… | |
505 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
523 | if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) |
506 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
524 | croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); |
507 | |
525 | |
508 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
526 | encode_tag (enc, CBOR_TAG_PERL_OBJECT); |
509 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
527 | encode_uint (enc, MAJOR_ARRAY, count + 1); |
510 | encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
528 | encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); |
511 | |
529 | |
512 | while (count) |
530 | while (count) |
513 | encode_sv (enc, SP[1 - count--]); |
531 | encode_sv (enc, SP[1 - count--]); |
514 | |
532 | |
515 | PUTBACK; |
533 | PUTBACK; |
… | |
… | |
574 | |
592 | |
575 | if (SvPOKp (sv)) |
593 | if (SvPOKp (sv)) |
576 | { |
594 | { |
577 | STRLEN len; |
595 | STRLEN len; |
578 | char *str = SvPV (sv, len); |
596 | char *str = SvPV (sv, len); |
579 | encode_strref (enc, SvUTF8 (sv), str, len); |
597 | encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len); |
580 | } |
598 | } |
581 | else if (SvNOKp (sv)) |
599 | else if (SvNOKp (sv)) |
582 | encode_nv (enc, sv); |
600 | encode_nv (enc, sv); |
583 | else if (SvIOKp (sv)) |
601 | else if (SvIOKp (sv)) |
584 | { |
602 | { |
… | |
… | |
924 | sv = newRV_noinc (decode_sv (dec)); |
942 | sv = newRV_noinc (decode_sv (dec)); |
925 | break; |
943 | break; |
926 | |
944 | |
927 | case CBOR_TAG_STRINGREF_NAMESPACE: |
945 | case CBOR_TAG_STRINGREF_NAMESPACE: |
928 | { |
946 | { |
|
|
947 | // do nmot use SAVETMPS/FREETMPS, as these will |
|
|
948 | // erase mortalised caches, e.g. "shareable" |
929 | ENTER; SAVETMPS; |
949 | ENTER; |
930 | |
950 | |
931 | SAVESPTR (dec->stringref); |
951 | SAVESPTR (dec->stringref); |
932 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
952 | dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); |
933 | |
953 | |
934 | sv = decode_sv (dec); |
954 | sv = decode_sv (dec); |
935 | |
955 | |
936 | FREETMPS; LEAVE; |
956 | LEAVE; |
937 | } |
957 | } |
938 | break; |
958 | break; |
939 | |
959 | |
940 | case CBOR_TAG_STRINGREF: |
960 | case CBOR_TAG_STRINGREF: |
941 | { |
961 | { |
… | |
… | |
1404 | shrink = F_SHRINK |
1424 | shrink = F_SHRINK |
1405 | allow_unknown = F_ALLOW_UNKNOWN |
1425 | allow_unknown = F_ALLOW_UNKNOWN |
1406 | allow_sharing = F_ALLOW_SHARING |
1426 | allow_sharing = F_ALLOW_SHARING |
1407 | allow_cycles = F_ALLOW_CYCLES |
1427 | allow_cycles = F_ALLOW_CYCLES |
1408 | pack_strings = F_PACK_STRINGS |
1428 | pack_strings = F_PACK_STRINGS |
|
|
1429 | text_keys = F_TEXT_KEYS |
1409 | utf8_strings = F_UTF8_STRINGS |
1430 | text_strings = F_TEXT_STRINGS |
1410 | validate_utf8 = F_VALIDATE_UTF8 |
1431 | validate_utf8 = F_VALIDATE_UTF8 |
1411 | PPCODE: |
1432 | PPCODE: |
1412 | { |
1433 | { |
1413 | if (enable) |
1434 | if (enable) |
1414 | self->flags |= ix; |
1435 | self->flags |= ix; |
… | |
… | |
1423 | get_shrink = F_SHRINK |
1444 | get_shrink = F_SHRINK |
1424 | get_allow_unknown = F_ALLOW_UNKNOWN |
1445 | get_allow_unknown = F_ALLOW_UNKNOWN |
1425 | get_allow_sharing = F_ALLOW_SHARING |
1446 | get_allow_sharing = F_ALLOW_SHARING |
1426 | get_allow_cycles = F_ALLOW_CYCLES |
1447 | get_allow_cycles = F_ALLOW_CYCLES |
1427 | get_pack_strings = F_PACK_STRINGS |
1448 | get_pack_strings = F_PACK_STRINGS |
|
|
1449 | get_text_keys = F_TEXT_KEYS |
|
|
1450 | get_text_strings = F_TEXT_STRINGS |
1428 | get_validate_utf8 = F_VALIDATE_UTF8 |
1451 | get_validate_utf8 = F_VALIDATE_UTF8 |
1429 | PPCODE: |
1452 | PPCODE: |
1430 | XPUSHs (boolSV (self->flags & ix)); |
1453 | XPUSHs (boolSV (self->flags & ix)); |
1431 | |
1454 | |
1432 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |
1455 | void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) |