ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/CBOR-XS/XS.xs
Revision: 1.34
Committed: Sat Nov 30 16:19:59 2013 UTC (10 years, 5 months ago) by root
Branch: MAIN
Changes since 1.33: +47 -27 lines
Log Message:
*** empty log message ***

File Contents

# User Rev Content
1 root 1.1 #include "EXTERN.h"
2     #include "perl.h"
3     #include "XSUB.h"
4    
5     #include <assert.h>
6     #include <string.h>
7     #include <stdlib.h>
8     #include <stdio.h>
9     #include <limits.h>
10     #include <float.h>
11    
12     #include "ecb.h"
13    
14 root 1.14 // compatibility with perl <5.18
15     #ifndef HvNAMELEN_get
16     # define HvNAMELEN_get(hv) strlen (HvNAME (hv))
17     #endif
18     #ifndef HvNAMELEN
19     # define HvNAMELEN(hv) HvNAMELEN_get (hv)
20     #endif
21     #ifndef HvNAMEUTF8
22     # define HvNAMEUTF8(hv) 0
23     #endif
24 root 1.28 #ifndef SvREFCNT_dec_NN
25     # define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26     #endif
27 root 1.14
28 root 1.9 // known tags
29 root 1.8 enum cbor_tag
30     {
31 root 1.29 // extensions
32     CBOR_TAG_STRINGREF = 25, // http://cbor.schmorp.de/stringref
33     CBOR_TAG_PERL_OBJECT = 26, // http://cbor.schmorp.de/perl-object
34     CBOR_TAG_GENERIC_OBJECT = 27, // http://cbor.schmorp.de/generic-object
35     CBOR_TAG_VALUE_SHAREABLE = 28, // http://cbor.schmorp.de/value-sharing
36     CBOR_TAG_VALUE_SHAREDREF = 29, // http://cbor.schmorp.de/value-sharing
37     CBOR_TAG_STRINGREF_NAMESPACE = 256, // http://cbor.schmorp.de/stringref
38 root 1.18 CBOR_TAG_INDIRECTION = 22098, // http://cbor.schmorp.de/indirection
39 root 1.9
40     // rfc7049
41     CBOR_TAG_DATETIME = 0, // rfc4287, utf-8
42     CBOR_TAG_TIMESTAMP = 1, // unix timestamp, any
43     CBOR_TAG_POS_BIGNUM = 2, // byte string
44     CBOR_TAG_NEG_BIGNUM = 3, // byte string
45     CBOR_TAG_DECIMAL = 4, // decimal fraction, array
46     CBOR_TAG_BIGFLOAT = 5, // array
47    
48     CBOR_TAG_CONV_B64U = 21, // base64url, any
49     CBOR_TAG_CONV_B64 = 22, // base64, any
50     CBOR_TAG_CONV_HEX = 23, // base16, any
51     CBOR_TAG_CBOR = 24, // embedded cbor, byte string
52    
53     CBOR_TAG_URI = 32, // URI rfc3986, utf-8
54     CBOR_TAG_B64U = 33, // base64url rfc4648, utf-8
55     CBOR_TAG_B64 = 34, // base6 rfc46484, utf-8
56     CBOR_TAG_REGEX = 35, // regex pcre/ecma262, utf-8
57     CBOR_TAG_MIME = 36, // mime message rfc2045, utf-8
58 root 1.8
59 root 1.9 CBOR_TAG_MAGIC = 55799 // self-describe cbor
60 root 1.8 };
61    
62 root 1.24 #define F_SHRINK 0x00000001UL
63     #define F_ALLOW_UNKNOWN 0x00000002UL
64 root 1.32 #define F_ALLOW_SHARING 0x00000004UL
65     #define F_PACK_STRINGS 0x00000008UL
66 root 1.1
67     #define INIT_SIZE 32 // initial scalar size to be allocated
68    
69     #define SB do {
70     #define SE } while (0)
71    
72     #define IN_RANGE_INC(type,val,beg,end) \
73     ((unsigned type)((unsigned type)(val) - (unsigned type)(beg)) \
74     <= (unsigned type)((unsigned type)(end) - (unsigned type)(beg)))
75    
76     #define ERR_NESTING_EXCEEDED "cbor text or perl structure exceeds maximum nesting level (max_depth set too low?)"
77    
78     #ifdef USE_ITHREADS
79     # define CBOR_SLOW 1
80     # define CBOR_STASH (cbor_stash ? cbor_stash : gv_stashpv ("CBOR::XS", 1))
81     #else
82     # define CBOR_SLOW 0
83     # define CBOR_STASH cbor_stash
84     #endif
85    
86 root 1.10 static HV *cbor_stash, *types_boolean_stash, *types_error_stash, *cbor_tagged_stash; // CBOR::XS::
87 root 1.27 static SV *types_true, *types_false, *types_error, *sv_cbor, *default_filter;
88 root 1.1
89     typedef struct {
90     U32 flags;
91     U32 max_depth;
92     STRLEN max_size;
93 root 1.27 SV *filter;
94 root 1.1 } CBOR;
95    
96 root 1.5 ecb_inline void
97 root 1.1 cbor_init (CBOR *cbor)
98     {
99     Zero (cbor, 1, CBOR);
100     cbor->max_depth = 512;
101     }
102    
103 root 1.27 ecb_inline void
104     cbor_free (CBOR *cbor)
105     {
106     SvREFCNT_dec (cbor->filter);
107     }
108    
109 root 1.1 /////////////////////////////////////////////////////////////////////////////
110     // utility functions
111    
112 root 1.5 ecb_inline SV *
113 root 1.1 get_bool (const char *name)
114     {
115     SV *sv = get_sv (name, 1);
116    
117     SvREADONLY_on (sv);
118     SvREADONLY_on (SvRV (sv));
119    
120     return sv;
121     }
122    
123 root 1.5 ecb_inline void
124 root 1.1 shrink (SV *sv)
125     {
126     sv_utf8_downgrade (sv, 1);
127    
128     if (SvLEN (sv) > SvCUR (sv) + 1)
129     {
130     #ifdef SvPV_shrink_to_cur
131     SvPV_shrink_to_cur (sv);
132     #elif defined (SvPV_renew)
133     SvPV_renew (sv, SvCUR (sv) + 1);
134     #endif
135     }
136     }
137    
138 root 1.21 // minimum length of a string to be registered for stringref
139     ecb_inline int
140     minimum_string_length (UV idx)
141     {
142     return idx > 23
143     ? idx > 0xffU
144     ? idx > 0xffffU
145     ? idx > 0xffffffffU
146 root 1.29 ? 11
147     : 7
148 root 1.21 : 5
149     : 4
150     : 3;
151     }
152    
153 root 1.1 /////////////////////////////////////////////////////////////////////////////
154     // encoder
155    
156     // structure used for encoding CBOR
157     typedef struct
158     {
159     char *cur; // SvPVX (sv) + current output position
160     char *end; // SvEND (sv)
161     SV *sv; // result scalar
162     CBOR cbor;
163     U32 depth; // recursion level
164 root 1.20 HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
165     UV stringref_idx;
166 root 1.19 HV *shareable; // ptr => index, or 0
167     UV shareable_idx;
168 root 1.1 } enc_t;
169    
170 root 1.5 ecb_inline void
171 root 1.1 need (enc_t *enc, STRLEN len)
172     {
173 root 1.5 if (ecb_expect_false (enc->cur + len >= enc->end))
174 root 1.1 {
175     STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
176     SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
177     enc->cur = SvPVX (enc->sv) + cur;
178     enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
179     }
180     }
181    
182 root 1.5 ecb_inline void
183 root 1.1 encode_ch (enc_t *enc, char ch)
184     {
185     need (enc, 1);
186     *enc->cur++ = ch;
187     }
188    
189     static void
190     encode_uint (enc_t *enc, int major, UV len)
191     {
192     need (enc, 9);
193    
194 root 1.31 if (ecb_expect_true (len < 24))
195 root 1.1 *enc->cur++ = major | len;
196 root 1.31 else if (ecb_expect_true (len <= 0xff))
197 root 1.1 {
198     *enc->cur++ = major | 24;
199     *enc->cur++ = len;
200     }
201 root 1.4 else if (len <= 0xffff)
202 root 1.1 {
203     *enc->cur++ = major | 25;
204     *enc->cur++ = len >> 8;
205     *enc->cur++ = len;
206     }
207 root 1.4 else if (len <= 0xffffffff)
208 root 1.1 {
209     *enc->cur++ = major | 26;
210     *enc->cur++ = len >> 24;
211     *enc->cur++ = len >> 16;
212     *enc->cur++ = len >> 8;
213     *enc->cur++ = len;
214     }
215 root 1.4 else
216 root 1.1 {
217     *enc->cur++ = major | 27;
218     *enc->cur++ = len >> 56;
219     *enc->cur++ = len >> 48;
220     *enc->cur++ = len >> 40;
221     *enc->cur++ = len >> 32;
222     *enc->cur++ = len >> 24;
223     *enc->cur++ = len >> 16;
224     *enc->cur++ = len >> 8;
225     *enc->cur++ = len;
226     }
227     }
228    
229 root 1.21 ecb_inline void
230     encode_tag (enc_t *enc, UV tag)
231     {
232     encode_uint (enc, 0xc0, tag);
233     }
234    
235 root 1.30 ecb_inline void
236     encode_str (enc_t *enc, int utf8, char *str, STRLEN len)
237     {
238     encode_uint (enc, utf8 ? 0x60 : 0x40, len);
239     need (enc, len);
240     memcpy (enc->cur, str, len);
241     enc->cur += len;
242     }
243    
244 root 1.1 static void
245 root 1.30 encode_strref (enc_t *enc, int utf8, char *str, STRLEN len)
246 root 1.1 {
247 root 1.32 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
248 root 1.21 {
249 root 1.22 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
250 root 1.21
251     if (SvOK (*svp))
252     {
253     // already registered, use stringref
254     encode_tag (enc, CBOR_TAG_STRINGREF);
255     encode_uint (enc, 0x00, SvUV (*svp));
256     return;
257     }
258     else if (len >= minimum_string_length (enc->stringref_idx))
259     {
260     // register only
261     sv_setuv (*svp, enc->stringref_idx);
262     ++enc->stringref_idx;
263     }
264     }
265    
266 root 1.30 encode_str (enc, utf8, str, len);
267 root 1.1 }
268    
269     static void encode_sv (enc_t *enc, SV *sv);
270    
271     static void
272     encode_av (enc_t *enc, AV *av)
273     {
274     int i, len = av_len (av);
275    
276     if (enc->depth >= enc->cbor.max_depth)
277     croak (ERR_NESTING_EXCEEDED);
278    
279     ++enc->depth;
280    
281     encode_uint (enc, 0x80, len + 1);
282    
283     for (i = 0; i <= len; ++i)
284     {
285     SV **svp = av_fetch (av, i, 0);
286     encode_sv (enc, svp ? *svp : &PL_sv_undef);
287     }
288    
289     --enc->depth;
290     }
291    
292     static void
293     encode_hv (enc_t *enc, HV *hv)
294     {
295     HE *he;
296    
297     if (enc->depth >= enc->cbor.max_depth)
298     croak (ERR_NESTING_EXCEEDED);
299    
300     ++enc->depth;
301    
302     int pairs = hv_iterinit (hv);
303     int mg = SvMAGICAL (hv);
304    
305     if (mg)
306     encode_ch (enc, 0xa0 | 31);
307     else
308     encode_uint (enc, 0xa0, pairs);
309    
310     while ((he = hv_iternext (hv)))
311     {
312 root 1.21 if (HeKLEN (he) == HEf_SVKEY)
313     encode_sv (enc, HeSVKEY (he));
314 root 1.1 else
315 root 1.30 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
316 root 1.1
317 root 1.5 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
318 root 1.1 }
319    
320     if (mg)
321     encode_ch (enc, 0xe0 | 31);
322    
323     --enc->depth;
324     }
325    
326     // encode objects, arrays and special \0=false and \1=true values.
327     static void
328     encode_rv (enc_t *enc, SV *sv)
329     {
330 root 1.19 SvGETMAGIC (sv);
331 root 1.1
332 root 1.19 svtype svt = SvTYPE (sv);
333 root 1.18
334 root 1.5 if (ecb_expect_false (SvOBJECT (sv)))
335 root 1.1 {
336 root 1.10 HV *boolean_stash = !CBOR_SLOW || types_boolean_stash
337     ? types_boolean_stash
338     : gv_stashpv ("Types::Serialiser::Boolean", 1);
339     HV *error_stash = !CBOR_SLOW || types_error_stash
340     ? types_error_stash
341     : gv_stashpv ("Types::Serialiser::Error", 1);
342 root 1.6 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
343     ? cbor_tagged_stash
344     : gv_stashpv ("CBOR::XS::Tagged" , 1);
345 root 1.1
346 root 1.11 HV *stash = SvSTASH (sv);
347    
348     if (stash == boolean_stash)
349 root 1.34 {
350     encode_ch (enc, SvIV (sv) ? 0xe0 | 21 : 0xe0 | 20);
351     return;
352     }
353 root 1.11 else if (stash == error_stash)
354 root 1.34 {
355     encode_ch (enc, 0xe0 | 23);
356     return;
357     }
358 root 1.11 else if (stash == tagged_stash)
359 root 1.6 {
360     if (svt != SVt_PVAV)
361     croak ("encountered CBOR::XS::Tagged object that isn't an array");
362    
363     encode_uint (enc, 0xc0, SvUV (*av_fetch ((AV *)sv, 0, 1)));
364     encode_sv (enc, *av_fetch ((AV *)sv, 1, 1));
365 root 1.34
366     return;
367 root 1.6 }
368 root 1.34 }
369    
370     if (ecb_expect_false (SvREFCNT (sv) > 1)
371     && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
372     {
373     if (!enc->shareable)
374     enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
375    
376     SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
377    
378     if (SvOK (*svp))
379     {
380     encode_tag (enc, CBOR_TAG_VALUE_SHAREDREF);
381     encode_uint (enc, 0x00, SvUV (*svp));
382     return;
383     }
384     else
385     {
386     sv_setuv (*svp, enc->shareable_idx);
387     ++enc->shareable_idx;
388     encode_tag (enc, CBOR_TAG_VALUE_SHAREABLE);
389     }
390     }
391    
392     if (ecb_expect_false (SvOBJECT (sv)))
393     {
394     HV *stash = SvSTASH (sv);
395     GV *method;
396    
397     if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
398 root 1.1 {
399 root 1.11 dSP;
400    
401     ENTER; SAVETMPS; PUSHMARK (SP);
402 root 1.6 // we re-bless the reference to get overload and other niceties right
403 root 1.11 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
404    
405     PUTBACK;
406     // G_SCALAR ensures that return value is 1
407     call_sv ((SV *)GvCV (method), G_SCALAR);
408     SPAGAIN;
409    
410     // catch this surprisingly common error
411     if (SvROK (TOPs) && SvRV (TOPs) == sv)
412     croak ("%s::TO_CBOR method returned same object as was passed instead of a new one", HvNAME (stash));
413    
414     encode_sv (enc, POPs);
415    
416     PUTBACK;
417    
418     FREETMPS; LEAVE;
419     }
420     else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
421     {
422     dSP;
423 root 1.6
424 root 1.11 ENTER; SAVETMPS; PUSHMARK (SP);
425     EXTEND (SP, 2);
426     // we re-bless the reference to get overload and other niceties right
427     PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
428     PUSHs (sv_cbor);
429 root 1.1
430 root 1.11 PUTBACK;
431     int count = call_sv ((SV *)GvCV (method), G_ARRAY);
432     SPAGAIN;
433 root 1.6
434 root 1.11 // catch this surprisingly common error
435     if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
436     croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
437 root 1.6
438 root 1.18 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
439 root 1.11 encode_uint (enc, 0x80, count + 1);
440 root 1.30 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
441 root 1.6
442 root 1.11 while (count)
443     encode_sv (enc, SP[1 - count--]);
444 root 1.6
445 root 1.11 PUTBACK;
446 root 1.6
447 root 1.11 FREETMPS; LEAVE;
448 root 1.1 }
449 root 1.11 else
450     croak ("encountered object '%s', but no TO_CBOR or FREEZE methods available on it",
451     SvPV_nolen (sv_2mortal (newRV_inc (sv))));
452 root 1.1 }
453     else if (svt == SVt_PVHV)
454     encode_hv (enc, (HV *)sv);
455     else if (svt == SVt_PVAV)
456     encode_av (enc, (AV *)sv);
457 root 1.18 else
458 root 1.1 {
459 root 1.18 encode_tag (enc, CBOR_TAG_INDIRECTION);
460     encode_sv (enc, sv);
461 root 1.1 }
462     }
463    
464     static void
465     encode_nv (enc_t *enc, SV *sv)
466     {
467     double nv = SvNVX (sv);
468    
469     need (enc, 9);
470    
471 root 1.5 if (ecb_expect_false (nv == (U32)nv))
472 root 1.1 encode_uint (enc, 0x00, (U32)nv);
473     //TODO: maybe I32?
474 root 1.5 else if (ecb_expect_false (nv == (float)nv))
475 root 1.1 {
476     uint32_t fp = ecb_float_to_binary32 (nv);
477    
478     *enc->cur++ = 0xe0 | 26;
479    
480     if (!ecb_big_endian ())
481     fp = ecb_bswap32 (fp);
482    
483     memcpy (enc->cur, &fp, 4);
484     enc->cur += 4;
485     }
486     else
487     {
488     uint64_t fp = ecb_double_to_binary64 (nv);
489    
490     *enc->cur++ = 0xe0 | 27;
491    
492     if (!ecb_big_endian ())
493     fp = ecb_bswap64 (fp);
494    
495     memcpy (enc->cur, &fp, 8);
496     enc->cur += 8;
497     }
498     }
499    
500     static void
501     encode_sv (enc_t *enc, SV *sv)
502     {
503     SvGETMAGIC (sv);
504    
505     if (SvPOKp (sv))
506     {
507     STRLEN len;
508     char *str = SvPV (sv, len);
509 root 1.30 encode_strref (enc, SvUTF8 (sv), str, len);
510 root 1.1 }
511     else if (SvNOKp (sv))
512     encode_nv (enc, sv);
513     else if (SvIOKp (sv))
514     {
515     if (SvIsUV (sv))
516     encode_uint (enc, 0x00, SvUVX (sv));
517     else if (SvIVX (sv) >= 0)
518     encode_uint (enc, 0x00, SvIVX (sv));
519     else
520     encode_uint (enc, 0x20, -(SvIVX (sv) + 1));
521     }
522     else if (SvROK (sv))
523     encode_rv (enc, SvRV (sv));
524     else if (!SvOK (sv))
525     encode_ch (enc, 0xe0 | 22);
526     else if (enc->cbor.flags & F_ALLOW_UNKNOWN)
527     encode_ch (enc, 0xe0 | 23);
528     else
529     croak ("encountered perl type (%s,0x%x) that CBOR cannot handle, check your input data",
530     SvPV_nolen (sv), (unsigned int)SvFLAGS (sv));
531     }
532    
533     static SV *
534     encode_cbor (SV *scalar, CBOR *cbor)
535     {
536 root 1.18 enc_t enc = { };
537 root 1.1
538     enc.cbor = *cbor;
539     enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
540     enc.cur = SvPVX (enc.sv);
541     enc.end = SvEND (enc.sv);
542    
543     SvPOK_only (enc.sv);
544 root 1.20
545 root 1.32 if (cbor->flags & F_PACK_STRINGS)
546 root 1.20 {
547     encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
548     enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
549     enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
550     }
551    
552 root 1.1 encode_sv (&enc, scalar);
553    
554     SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
555     *SvEND (enc.sv) = 0; // many xs functions expect a trailing 0 for text strings
556    
557     if (enc.cbor.flags & F_SHRINK)
558     shrink (enc.sv);
559    
560     return enc.sv;
561     }
562    
563     /////////////////////////////////////////////////////////////////////////////
564     // decoder
565    
566     // structure used for decoding CBOR
567     typedef struct
568     {
569     U8 *cur; // current parser pointer
570     U8 *end; // end of input string
571     const char *err; // parse error, if != 0
572     CBOR cbor;
573     U32 depth; // recursion depth
574     U32 maxdepth; // recursion depth limit
575 root 1.19 AV *shareable;
576 root 1.20 AV *stringref;
577 root 1.27 SV *decode_tagged;
578 root 1.1 } dec_t;
579    
580     #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
581    
582 root 1.5 #define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data")
583 root 1.1
584     #define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED)
585     #define DEC_DEC_DEPTH --dec->depth
586    
587     static UV
588     decode_uint (dec_t *dec)
589     {
590     switch (*dec->cur & 31)
591     {
592     case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
593     case 8: case 9: case 10: case 11: case 12: case 13: case 14: case 15:
594     case 16: case 17: case 18: case 19: case 20: case 21: case 22: case 23:
595     return *dec->cur++ & 31;
596    
597     case 24:
598     WANT (2);
599     dec->cur += 2;
600     return dec->cur[-1];
601    
602     case 25:
603     WANT (3);
604     dec->cur += 3;
605     return (((UV)dec->cur[-2]) << 8)
606     | ((UV)dec->cur[-1]);
607    
608     case 26:
609     WANT (5);
610     dec->cur += 5;
611     return (((UV)dec->cur[-4]) << 24)
612     | (((UV)dec->cur[-3]) << 16)
613     | (((UV)dec->cur[-2]) << 8)
614     | ((UV)dec->cur[-1]);
615    
616     case 27:
617     WANT (9);
618     dec->cur += 9;
619 root 1.34
620     return
621     #if UVSIZE < 8
622     0
623     #else
624     (((UV)dec->cur[-8]) << 56)
625 root 1.1 | (((UV)dec->cur[-7]) << 48)
626     | (((UV)dec->cur[-6]) << 40)
627     | (((UV)dec->cur[-5]) << 32)
628 root 1.34 #endif
629 root 1.1 | (((UV)dec->cur[-4]) << 24)
630     | (((UV)dec->cur[-3]) << 16)
631     | (((UV)dec->cur[-2]) << 8)
632     | ((UV)dec->cur[-1]);
633    
634     default:
635     ERR ("corrupted CBOR data (unsupported integer minor encoding)");
636     }
637    
638     fail:
639     return 0;
640     }
641    
642     static SV *decode_sv (dec_t *dec);
643    
644     static SV *
645     decode_av (dec_t *dec)
646     {
647     AV *av = newAV ();
648    
649     DEC_INC_DEPTH;
650    
651     if ((*dec->cur & 31) == 31)
652     {
653     ++dec->cur;
654    
655     for (;;)
656     {
657     WANT (1);
658    
659 root 1.2 if (*dec->cur == (0xe0 | 31))
660 root 1.1 {
661     ++dec->cur;
662     break;
663     }
664    
665     av_push (av, decode_sv (dec));
666     }
667     }
668     else
669     {
670     int i, len = decode_uint (dec);
671    
672     av_fill (av, len - 1);
673    
674     for (i = 0; i < len; ++i)
675     AvARRAY (av)[i] = decode_sv (dec);
676     }
677    
678     DEC_DEC_DEPTH;
679     return newRV_noinc ((SV *)av);
680    
681     fail:
682     SvREFCNT_dec (av);
683     DEC_DEC_DEPTH;
684     return &PL_sv_undef;
685     }
686    
687 root 1.16 static void
688     decode_he (dec_t *dec, HV *hv)
689     {
690     // for speed reasons, we specialcase single-string
691 root 1.21 // byte or utf-8 strings as keys, but only when !stringref
692    
693 root 1.23 if (ecb_expect_true (!dec->stringref))
694 root 1.21 if (*dec->cur >= 0x40 && *dec->cur <= 0x40 + 27)
695     {
696     I32 len = decode_uint (dec);
697     char *key = (char *)dec->cur;
698 root 1.16
699 root 1.21 dec->cur += len;
700    
701     if (ecb_expect_false (dec->stringref))
702     av_push (dec->stringref, newSVpvn (key, len));
703    
704     hv_store (hv, key, len, decode_sv (dec), 0);
705 root 1.16
706 root 1.21 return;
707     }
708     else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)
709     {
710     I32 len = decode_uint (dec);
711     char *key = (char *)dec->cur;
712 root 1.16
713 root 1.21 dec->cur += len;
714 root 1.20
715 root 1.21 if (ecb_expect_false (dec->stringref))
716     av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));
717 root 1.16
718 root 1.21 hv_store (hv, key, -len, decode_sv (dec), 0);
719 root 1.16
720 root 1.21 return;
721     }
722 root 1.20
723 root 1.21 SV *k = decode_sv (dec);
724     SV *v = decode_sv (dec);
725 root 1.16
726 root 1.21 hv_store_ent (hv, k, v, 0);
727     SvREFCNT_dec (k);
728 root 1.16 }
729    
730 root 1.1 static SV *
731     decode_hv (dec_t *dec)
732     {
733     HV *hv = newHV ();
734    
735     DEC_INC_DEPTH;
736    
737     if ((*dec->cur & 31) == 31)
738     {
739     ++dec->cur;
740    
741     for (;;)
742     {
743     WANT (1);
744    
745 root 1.2 if (*dec->cur == (0xe0 | 31))
746 root 1.1 {
747     ++dec->cur;
748     break;
749     }
750    
751 root 1.16 decode_he (dec, hv);
752 root 1.1 }
753     }
754     else
755     {
756 root 1.16 int pairs = decode_uint (dec);
757 root 1.1
758 root 1.16 while (pairs--)
759     decode_he (dec, hv);
760 root 1.1 }
761    
762     DEC_DEC_DEPTH;
763     return newRV_noinc ((SV *)hv);
764    
765     fail:
766     SvREFCNT_dec (hv);
767     DEC_DEC_DEPTH;
768     return &PL_sv_undef;
769     }
770    
771     static SV *
772     decode_str (dec_t *dec, int utf8)
773     {
774 root 1.6 SV *sv = 0;
775 root 1.1
776     if ((*dec->cur & 31) == 31)
777     {
778 root 1.33 // indefinite length strings
779 root 1.1 ++dec->cur;
780    
781 root 1.33 unsigned char major = *dec->cur & 0xe0;
782    
783 root 1.1 sv = newSVpvn ("", 0);
784    
785     for (;;)
786     {
787     WANT (1);
788    
789 root 1.33 if ((*dec->cur ^ major) >= 31)
790     if (*dec->cur == (0xe0 | 31))
791     {
792     ++dec->cur;
793     break;
794     }
795     else
796     ERR ("corrupted CBOR data (invalid chunks in indefinite length string)");
797    
798     STRLEN len = decode_uint (dec);
799 root 1.1
800 root 1.33 WANT (len);
801     sv_catpvn (sv, dec->cur, len);
802     dec->cur += len;
803 root 1.1 }
804     }
805     else
806     {
807     STRLEN len = decode_uint (dec);
808    
809     WANT (len);
810     sv = newSVpvn (dec->cur, len);
811     dec->cur += len;
812 root 1.25
813     if (ecb_expect_false (dec->stringref)
814     && SvCUR (sv) >= minimum_string_length (AvFILLp (dec->stringref) + 1))
815     av_push (dec->stringref, SvREFCNT_inc_NN (sv));
816 root 1.1 }
817    
818     if (utf8)
819     SvUTF8_on (sv);
820    
821     return sv;
822    
823     fail:
824 root 1.6 SvREFCNT_dec (sv);
825 root 1.1 return &PL_sv_undef;
826     }
827    
828     static SV *
829 root 1.3 decode_tagged (dec_t *dec)
830     {
831 root 1.19 SV *sv = 0;
832 root 1.3 UV tag = decode_uint (dec);
833 root 1.19
834     WANT (1);
835 root 1.3
836 root 1.18 switch (tag)
837     {
838     case CBOR_TAG_MAGIC:
839 root 1.20 sv = decode_sv (dec);
840     break;
841 root 1.18
842     case CBOR_TAG_INDIRECTION:
843 root 1.20 sv = newRV_noinc (decode_sv (dec));
844     break;
845    
846     case CBOR_TAG_STRINGREF_NAMESPACE:
847     {
848     ENTER; SAVETMPS;
849    
850     SAVESPTR (dec->stringref);
851     dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
852    
853     sv = decode_sv (dec);
854    
855     FREETMPS; LEAVE;
856     }
857     break;
858    
859     case CBOR_TAG_STRINGREF:
860     {
861     if ((*dec->cur >> 5) != 0)
862     ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
863    
864     UV idx = decode_uint (dec);
865    
866     if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
867     ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
868    
869     sv = newSVsv (AvARRAY (dec->stringref)[idx]);
870     }
871     break;
872 root 1.11
873 root 1.19 case CBOR_TAG_VALUE_SHAREABLE:
874     {
875     if (ecb_expect_false (!dec->shareable))
876     dec->shareable = (AV *)sv_2mortal ((SV *)newAV ());
877    
878     sv = newSV (0);
879     av_push (dec->shareable, SvREFCNT_inc_NN (sv));
880 root 1.18
881 root 1.19 SV *osv = decode_sv (dec);
882     sv_setsv (sv, osv);
883     SvREFCNT_dec_NN (osv);
884     }
885 root 1.20 break;
886 root 1.18
887     case CBOR_TAG_VALUE_SHAREDREF:
888 root 1.17 {
889 root 1.19 if ((*dec->cur >> 5) != 0)
890     ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
891 root 1.18
892 root 1.19 UV idx = decode_uint (dec);
893    
894 root 1.20 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
895 root 1.18 ERR ("corrupted CBOR data (sharedref index out of bounds)");
896    
897 root 1.20 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
898 root 1.17 }
899 root 1.20 break;
900 root 1.17
901 root 1.18 case CBOR_TAG_PERL_OBJECT:
902     {
903 root 1.19 sv = decode_sv (dec);
904    
905 root 1.18 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
906     ERR ("corrupted CBOR data (non-array perl object)");
907    
908     AV *av = (AV *)SvRV (sv);
909     int len = av_len (av) + 1;
910     HV *stash = gv_stashsv (*av_fetch (av, 0, 1), 0);
911    
912     if (!stash)
913     ERR ("cannot decode perl-object (package does not exist)");
914    
915     GV *method = gv_fetchmethod_autoload (stash, "THAW", 0);
916    
917     if (!method)
918     ERR ("cannot decode perl-object (package does not have a THAW method)");
919    
920     dSP;
921    
922     ENTER; SAVETMPS; PUSHMARK (SP);
923     EXTEND (SP, len + 1);
924     // we re-bless the reference to get overload and other niceties right
925     PUSHs (*av_fetch (av, 0, 1));
926     PUSHs (sv_cbor);
927    
928     int i;
929    
930     for (i = 1; i < len; ++i)
931     PUSHs (*av_fetch (av, i, 1));
932    
933     PUTBACK;
934     call_sv ((SV *)GvCV (method), G_SCALAR | G_EVAL);
935     SPAGAIN;
936    
937     if (SvTRUE (ERRSV))
938     {
939     FREETMPS; LEAVE;
940     ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
941     }
942    
943     SvREFCNT_dec (sv);
944     sv = SvREFCNT_inc (POPs);
945 root 1.11
946 root 1.18 PUTBACK;
947 root 1.11
948 root 1.18 FREETMPS; LEAVE;
949     }
950 root 1.20 break;
951 root 1.9
952 root 1.18 default:
953     {
954 root 1.19 sv = decode_sv (dec);
955    
956 root 1.27 dSP;
957     ENTER; SAVETMPS; PUSHMARK (SP);
958     EXTEND (SP, 2);
959     PUSHs (newSVuv (tag));
960     PUSHs (sv);
961    
962     PUTBACK;
963     int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
964     SPAGAIN;
965    
966     if (SvTRUE (ERRSV))
967     {
968     FREETMPS; LEAVE;
969     ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV))));
970     }
971    
972     if (count)
973     {
974     SvREFCNT_dec (sv);
975     sv = SvREFCNT_inc (POPs);
976     }
977     else
978     {
979     AV *av = newAV ();
980     av_push (av, newSVuv (tag));
981     av_push (av, sv);
982    
983     HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
984     ? cbor_tagged_stash
985     : gv_stashpv ("CBOR::XS::Tagged" , 1);
986     sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
987     }
988 root 1.7
989 root 1.27 PUTBACK;
990    
991     FREETMPS; LEAVE;
992 root 1.18 }
993 root 1.20 break;
994 root 1.11 }
995 root 1.9
996 root 1.20 return sv;
997    
998 root 1.9 fail:
999     SvREFCNT_dec (sv);
1000     return &PL_sv_undef;
1001 root 1.3 }
1002    
1003     static SV *
1004 root 1.1 decode_sv (dec_t *dec)
1005     {
1006     WANT (1);
1007    
1008     switch (*dec->cur >> 5)
1009     {
1010     case 0: // unsigned int
1011     return newSVuv (decode_uint (dec));
1012     case 1: // negative int
1013     return newSViv (-1 - (IV)decode_uint (dec));
1014     case 2: // octet string
1015     return decode_str (dec, 0);
1016     case 3: // utf-8 string
1017     return decode_str (dec, 1);
1018     case 4: // array
1019     return decode_av (dec);
1020     case 5: // map
1021     return decode_hv (dec);
1022     case 6: // tag
1023 root 1.3 return decode_tagged (dec);
1024 root 1.1 case 7: // misc
1025     switch (*dec->cur++ & 31)
1026     {
1027     case 20:
1028     #if CBOR_SLOW
1029 root 1.10 types_false = get_bool ("Types::Serialiser::false");
1030 root 1.1 #endif
1031 root 1.10 return newSVsv (types_false);
1032 root 1.1 case 21:
1033     #if CBOR_SLOW
1034 root 1.10 types_true = get_bool ("Types::Serialiser::true");
1035 root 1.1 #endif
1036 root 1.10 return newSVsv (types_true);
1037 root 1.1 case 22:
1038     return newSVsv (&PL_sv_undef);
1039 root 1.10 case 23:
1040     #if CBOR_SLOW
1041     types_error = get_bool ("Types::Serialiser::error");
1042     #endif
1043     return newSVsv (types_error);
1044 root 1.1
1045     case 25:
1046 root 1.2 {
1047     WANT (2);
1048    
1049     uint16_t fp = (dec->cur[0] << 8) | dec->cur[1];
1050     dec->cur += 2;
1051    
1052     return newSVnv (ecb_binary16_to_float (fp));
1053     }
1054 root 1.1
1055     case 26:
1056     {
1057     uint32_t fp;
1058     WANT (4);
1059     memcpy (&fp, dec->cur, 4);
1060     dec->cur += 4;
1061    
1062     if (!ecb_big_endian ())
1063     fp = ecb_bswap32 (fp);
1064    
1065     return newSVnv (ecb_binary32_to_float (fp));
1066     }
1067    
1068     case 27:
1069     {
1070     uint64_t fp;
1071     WANT (8);
1072     memcpy (&fp, dec->cur, 8);
1073     dec->cur += 8;
1074    
1075     if (!ecb_big_endian ())
1076     fp = ecb_bswap64 (fp);
1077    
1078     return newSVnv (ecb_binary64_to_double (fp));
1079     }
1080    
1081     // 0..19 unassigned
1082     // 24 reserved + unassigned (reserved values are not encodable)
1083     default:
1084     ERR ("corrupted CBOR data (reserved/unassigned major 7 value)");
1085     }
1086    
1087     break;
1088     }
1089    
1090     fail:
1091     return &PL_sv_undef;
1092     }
1093    
1094     static SV *
1095     decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1096     {
1097 root 1.18 dec_t dec = { };
1098 root 1.1 SV *sv;
1099 root 1.16 STRLEN len;
1100     char *data = SvPVbyte (string, len);
1101 root 1.1
1102 root 1.16 if (len > cbor->max_size && cbor->max_size)
1103     croak ("attempted decode of CBOR text of %lu bytes size, but max_size is set to %lu",
1104     (unsigned long)len, (unsigned long)cbor->max_size);
1105 root 1.1
1106     dec.cbor = *cbor;
1107 root 1.16 dec.cur = (U8 *)data;
1108     dec.end = (U8 *)data + len;
1109 root 1.1
1110     sv = decode_sv (&dec);
1111    
1112     if (offset_return)
1113     *offset_return = dec.cur;
1114    
1115     if (!(offset_return || !sv))
1116 root 1.2 if (dec.cur != dec.end && !dec.err)
1117     dec.err = "garbage after CBOR object";
1118    
1119     if (dec.err)
1120 root 1.1 {
1121 root 1.2 SvREFCNT_dec (sv);
1122 root 1.16 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1123 root 1.1 }
1124    
1125     sv = sv_2mortal (sv);
1126    
1127     return sv;
1128     }
1129    
1130     /////////////////////////////////////////////////////////////////////////////
1131     // XS interface functions
1132    
1133     MODULE = CBOR::XS PACKAGE = CBOR::XS
1134    
1135     BOOT:
1136     {
1137     cbor_stash = gv_stashpv ("CBOR::XS" , 1);
1138 root 1.6 cbor_tagged_stash = gv_stashpv ("CBOR::XS::Tagged" , 1);
1139 root 1.1
1140 root 1.10 types_boolean_stash = gv_stashpv ("Types::Serialiser::Boolean", 1);
1141     types_error_stash = gv_stashpv ("Types::Serialiser::Error" , 1);
1142    
1143     types_true = get_bool ("Types::Serialiser::true" );
1144     types_false = get_bool ("Types::Serialiser::false");
1145     types_error = get_bool ("Types::Serialiser::error");
1146 root 1.11
1147 root 1.27 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1148    
1149 root 1.11 sv_cbor = newSVpv ("CBOR", 0);
1150     SvREADONLY_on (sv_cbor);
1151 root 1.1 }
1152    
1153     PROTOTYPES: DISABLE
1154    
1155     void CLONE (...)
1156     CODE:
1157 root 1.10 cbor_stash = 0;
1158     cbor_tagged_stash = 0;
1159     types_error_stash = 0;
1160     types_boolean_stash = 0;
1161 root 1.1
1162     void new (char *klass)
1163     PPCODE:
1164     {
1165     SV *pv = NEWSV (0, sizeof (CBOR));
1166     SvPOK_only (pv);
1167     cbor_init ((CBOR *)SvPVX (pv));
1168     XPUSHs (sv_2mortal (sv_bless (
1169     newRV_noinc (pv),
1170     strEQ (klass, "CBOR::XS") ? CBOR_STASH : gv_stashpv (klass, 1)
1171     )));
1172     }
1173    
1174     void shrink (CBOR *self, int enable = 1)
1175     ALIAS:
1176     shrink = F_SHRINK
1177     allow_unknown = F_ALLOW_UNKNOWN
1178 root 1.18 allow_sharing = F_ALLOW_SHARING
1179 root 1.32 pack_strings = F_PACK_STRINGS
1180 root 1.1 PPCODE:
1181     {
1182     if (enable)
1183     self->flags |= ix;
1184     else
1185     self->flags &= ~ix;
1186    
1187     XPUSHs (ST (0));
1188     }
1189    
1190     void get_shrink (CBOR *self)
1191     ALIAS:
1192     get_shrink = F_SHRINK
1193     get_allow_unknown = F_ALLOW_UNKNOWN
1194 root 1.18 get_allow_sharing = F_ALLOW_SHARING
1195 root 1.32 get_pack_strings = F_PACK_STRINGS
1196 root 1.1 PPCODE:
1197     XPUSHs (boolSV (self->flags & ix));
1198    
1199     void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
1200     PPCODE:
1201     self->max_depth = max_depth;
1202     XPUSHs (ST (0));
1203    
1204     U32 get_max_depth (CBOR *self)
1205     CODE:
1206     RETVAL = self->max_depth;
1207     OUTPUT:
1208     RETVAL
1209    
1210     void max_size (CBOR *self, U32 max_size = 0)
1211     PPCODE:
1212     self->max_size = max_size;
1213     XPUSHs (ST (0));
1214    
1215     int get_max_size (CBOR *self)
1216     CODE:
1217     RETVAL = self->max_size;
1218     OUTPUT:
1219     RETVAL
1220    
1221 root 1.27 void filter (CBOR *self, SV *filter = 0)
1222     PPCODE:
1223     SvREFCNT_dec (self->filter);
1224     self->filter = filter ? newSVsv (filter) : filter;
1225     XPUSHs (ST (0));
1226    
1227     SV *get_filter (CBOR *self)
1228     CODE:
1229     RETVAL = self->filter ? self->filter : NEWSV (0, 0);
1230     OUTPUT:
1231     RETVAL
1232    
1233 root 1.1 void encode (CBOR *self, SV *scalar)
1234     PPCODE:
1235     PUTBACK; scalar = encode_cbor (scalar, self); SPAGAIN;
1236     XPUSHs (scalar);
1237    
1238     void decode (CBOR *self, SV *cborstr)
1239     PPCODE:
1240     PUTBACK; cborstr = decode_cbor (cborstr, self, 0); SPAGAIN;
1241     XPUSHs (cborstr);
1242    
1243     void decode_prefix (CBOR *self, SV *cborstr)
1244     PPCODE:
1245     {
1246     SV *sv;
1247     char *offset;
1248     PUTBACK; sv = decode_cbor (cborstr, self, &offset); SPAGAIN;
1249     EXTEND (SP, 2);
1250     PUSHs (sv);
1251     PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1252     }
1253    
1254 root 1.27 void DESTROY (CBOR *self)
1255     PPCODE:
1256     cbor_free (self);
1257    
1258 root 1.1 PROTOTYPES: ENABLE
1259    
1260     void encode_cbor (SV *scalar)
1261     PPCODE:
1262     {
1263     CBOR cbor;
1264     cbor_init (&cbor);
1265     PUTBACK; scalar = encode_cbor (scalar, &cbor); SPAGAIN;
1266     XPUSHs (scalar);
1267     }
1268    
1269     void decode_cbor (SV *cborstr)
1270     PPCODE:
1271     {
1272     CBOR cbor;
1273     cbor_init (&cbor);
1274     PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1275     XPUSHs (cborstr);
1276     }
1277