This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Better documentation for internal SV types
[perl5.git] / pod / perlguts.pod
CommitLineData
a0d0e21e
LW
1=head1 NAME
2
954c1994 3perlguts - Introduction to the Perl API
a0d0e21e
LW
4
5=head1 DESCRIPTION
6
b3b6085d 7This document attempts to describe how to use the Perl API, as well as
06f6df17 8to provide some info on the basic workings of the Perl core. It is far
b3b6085d
PP
9from complete and probably contains many errors. Please refer any
10questions or comments to the author below.
a0d0e21e 11
0a753a76
PP
12=head1 Variables
13
5f05dabc 14=head2 Datatypes
a0d0e21e
LW
15
16Perl has three typedefs that handle Perl's three main data types:
17
18 SV Scalar Value
19 AV Array Value
20 HV Hash Value
21
d1b91892 22Each typedef has specific routines that manipulate the various data types.
a0d0e21e
LW
23
24=head2 What is an "IV"?
25
954c1994 26Perl uses a special typedef IV which is a simple signed integer type that is
5f05dabc 27guaranteed to be large enough to hold a pointer (as well as an integer).
954c1994 28Additionally, there is the UV, which is simply an unsigned IV.
a0d0e21e 29
d1b91892 30Perl also uses two special typedefs, I32 and I16, which will always be at
954c1994 31least 32-bits and 16-bits long, respectively. (Again, there are U32 and U16,
20dbd849
NC
32as well.) They will usually be exactly 32 and 16 bits long, but on Crays
33they will both be 64 bits.
a0d0e21e 34
54310121 35=head2 Working with SVs
a0d0e21e 36
20dbd849
NC
37An SV can be created and loaded with one command. There are five types of
38values that can be loaded: an integer value (IV), an unsigned integer
39value (UV), a double (NV), a string (PV), and another scalar (SV).
61984ee1
KW
40("PV" stands for "Pointer Value". You might think that it is misnamed
41because it is described as pointing only to strings. However, it is
42possible to have it point to other things. For example, inversion
43lists, used in regular expression data structures, are scalars, each
44consisting of an array of UVs which are accessed through PVs. But,
45using it for non-strings requires care, as the underlying assumption of
46much of the internals is that PVs are just for strings. Often, for
47example, a trailing NUL is tacked on automatically. The non-string use
48is documented only in this paragraph.)
a0d0e21e 49
20dbd849 50The seven routines are:
a0d0e21e
LW
51
52 SV* newSViv(IV);
20dbd849 53 SV* newSVuv(UV);
a0d0e21e 54 SV* newSVnv(double);
06f6df17
RGS
55 SV* newSVpv(const char*, STRLEN);
56 SV* newSVpvn(const char*, STRLEN);
46fc3d4c 57 SV* newSVpvf(const char*, ...);
a0d0e21e
LW
58 SV* newSVsv(SV*);
59
06f6df17
RGS
60C<STRLEN> is an integer type (Size_t, usually defined as size_t in
61F<config.h>) guaranteed to be large enough to represent the size of
62any string that perl can handle.
63
64In the unlikely case of a SV requiring more complex initialisation, you
65can create an empty SV with newSV(len). If C<len> is 0 an empty SV of
66type NULL is returned, else an SV of type PV is returned with len + 1 (for
67the NUL) bytes of storage allocated, accessible via SvPVX. In both cases
da8c5729 68the SV has the undef value.
20dbd849 69
06f6df17 70 SV *sv = newSV(0); /* no storage allocated */
a9b0660e
KW
71 SV *sv = newSV(10); /* 10 (+1) bytes of uninitialised storage
72 * allocated */
20dbd849 73
06f6df17 74To change the value of an I<already-existing> SV, there are eight routines:
a0d0e21e
LW
75
76 void sv_setiv(SV*, IV);
deb3007b 77 void sv_setuv(SV*, UV);
a0d0e21e 78 void sv_setnv(SV*, double);
08105a92 79 void sv_setpv(SV*, const char*);
06f6df17 80 void sv_setpvn(SV*, const char*, STRLEN)
46fc3d4c 81 void sv_setpvf(SV*, const char*, ...);
a9b0660e
KW
82 void sv_vsetpvfn(SV*, const char*, STRLEN, va_list *,
83 SV **, I32, bool *);
a0d0e21e
LW
84 void sv_setsv(SV*, SV*);
85
86Notice that you can choose to specify the length of the string to be
9da1e3b5
MUN
87assigned by using C<sv_setpvn>, C<newSVpvn>, or C<newSVpv>, or you may
88allow Perl to calculate the length by using C<sv_setpv> or by specifying
890 as the second argument to C<newSVpv>. Be warned, though, that Perl will
90determine the string's length by using C<strlen>, which depends on the
a9b0660e
KW
91string terminating with a NUL character, and not otherwise containing
92NULs.
9abd00ed
GS
93
94The arguments of C<sv_setpvf> are processed like C<sprintf>, and the
95formatted output becomes the value.
96
328bf373 97C<sv_vsetpvfn> is an analogue of C<vsprintf>, but it allows you to specify
9abd00ed
GS
98either a pointer to a variable argument list or the address and length of
99an array of SVs. The last argument points to a boolean; on return, if that
100boolean is true, then locale-specific information has been used to format
c2611fb3 101the string, and the string's contents are therefore untrustworthy (see
9abd00ed
GS
102L<perlsec>). This pointer may be NULL if that information is not
103important. Note that this function requires you to specify the length of
104the format.
105
9da1e3b5
MUN
106The C<sv_set*()> functions are not generic enough to operate on values
107that have "magic". See L<Magic Virtual Tables> later in this document.
a0d0e21e 108
a3cb178b
GS
109All SVs that contain strings should be terminated with a NUL character.
110If it is not NUL-terminated there is a risk of
5f05dabc
PP
111core dumps and corruptions from code which passes the string to C
112functions or system calls which expect a NUL-terminated string.
113Perl's own functions typically add a trailing NUL for this reason.
114Nevertheless, you should be very careful when you pass a string stored
115in an SV to a C function or system call.
116
a0d0e21e
LW
117To access the actual value that an SV points to, you can use the macros:
118
119 SvIV(SV*)
954c1994 120 SvUV(SV*)
a0d0e21e
LW
121 SvNV(SV*)
122 SvPV(SV*, STRLEN len)
1fa8b10d 123 SvPV_nolen(SV*)
a0d0e21e 124
954c1994 125which will automatically coerce the actual scalar type into an IV, UV, double,
a0d0e21e
LW
126or string.
127
128In the C<SvPV> macro, the length of the string returned is placed into the
1fa8b10d
JD
129variable C<len> (this is a macro, so you do I<not> use C<&len>). If you do
130not care what the length of the data is, use the C<SvPV_nolen> macro.
131Historically the C<SvPV> macro with the global variable C<PL_na> has been
132used in this case. But that can be quite inefficient because C<PL_na> must
133be accessed in thread-local storage in threaded Perl. In any case, remember
134that Perl allows arbitrary strings of data that may both contain NULs and
135might not be terminated by a NUL.
a0d0e21e 136
ce2f5d8f
KA
137Also remember that C doesn't allow you to safely say C<foo(SvPV(s, len),
138len);>. It might work with your compiler, but it won't work for everyone.
139Break this sort of statement up into separate assignments:
140
1aa6ea50
JC
141 SV *s;
142 STRLEN len;
61955433 143 char *ptr;
1aa6ea50
JC
144 ptr = SvPV(s, len);
145 foo(ptr, len);
ce2f5d8f 146
07fa94a1 147If you want to know if the scalar value is TRUE, you can use:
a0d0e21e
LW
148
149 SvTRUE(SV*)
150
151Although Perl will automatically grow strings for you, if you need to force
152Perl to allocate more memory for your SV, you can use the macro
153
154 SvGROW(SV*, STRLEN newlen)
155
156which will determine if more memory needs to be allocated. If so, it will
157call the function C<sv_grow>. Note that C<SvGROW> can only increase, not
5f05dabc 158decrease, the allocated memory of an SV and that it does not automatically
a9b0660e 159add space for the trailing NUL byte (perl's own string functions typically do
8ebc5c01 160C<SvGROW(sv, len + 1)>).
a0d0e21e
LW
161
162If you have an SV and want to know what kind of data Perl thinks is stored
163in it, you can use the following macros to check the type of SV you have.
164
165 SvIOK(SV*)
166 SvNOK(SV*)
167 SvPOK(SV*)
168
169You can get and set the current length of the string stored in an SV with
170the following macros:
171
172 SvCUR(SV*)
173 SvCUR_set(SV*, I32 val)
174
cb1a09d0
AD
175You can also get a pointer to the end of the string stored in the SV
176with the macro:
177
178 SvEND(SV*)
179
180But note that these last three macros are valid only if C<SvPOK()> is true.
a0d0e21e 181
d1b91892
AD
182If you want to append something to the end of string stored in an C<SV*>,
183you can use the following functions:
184
08105a92 185 void sv_catpv(SV*, const char*);
e65f3abd 186 void sv_catpvn(SV*, const char*, STRLEN);
46fc3d4c 187 void sv_catpvf(SV*, const char*, ...);
a9b0660e
KW
188 void sv_vcatpvfn(SV*, const char*, STRLEN, va_list *, SV **,
189 I32, bool);
d1b91892
AD
190 void sv_catsv(SV*, SV*);
191
192The first function calculates the length of the string to be appended by
193using C<strlen>. In the second, you specify the length of the string
46fc3d4c 194yourself. The third function processes its arguments like C<sprintf> and
9abd00ed
GS
195appends the formatted output. The fourth function works like C<vsprintf>.
196You can specify the address and length of an array of SVs instead of the
197va_list argument. The fifth function extends the string stored in the first
198SV with the string stored in the second SV. It also forces the second SV
199to be interpreted as a string.
200
201The C<sv_cat*()> functions are not generic enough to operate on values that
202have "magic". See L<Magic Virtual Tables> later in this document.
d1b91892 203
a0d0e21e
LW
204If you know the name of a scalar variable, you can get a pointer to its SV
205by using the following:
206
64ace3f8 207 SV* get_sv("package::varname", 0);
a0d0e21e
LW
208
209This returns NULL if the variable does not exist.
210
d1b91892 211If you want to know if this variable (or any other SV) is actually C<defined>,
a0d0e21e
LW
212you can call:
213
214 SvOK(SV*)
215
06f6df17 216The scalar C<undef> value is stored in an SV instance called C<PL_sv_undef>.
9adebda4
SB
217
218Its address can be used whenever an C<SV*> is needed. Make sure that
219you don't try to compare a random sv with C<&PL_sv_undef>. For example
220when interfacing Perl code, it'll work correctly for:
221
222 foo(undef);
223
224But won't work when called as:
225
226 $x = undef;
227 foo($x);
228
229So to repeat always use SvOK() to check whether an sv is defined.
230
231Also you have to be careful when using C<&PL_sv_undef> as a value in
232AVs or HVs (see L<AVs, HVs and undefined values>).
a0d0e21e 233
06f6df17
RGS
234There are also the two values C<PL_sv_yes> and C<PL_sv_no>, which contain
235boolean TRUE and FALSE values, respectively. Like C<PL_sv_undef>, their
236addresses can be used whenever an C<SV*> is needed.
a0d0e21e 237
9cde0e7f 238Do not be fooled into thinking that C<(SV *) 0> is the same as C<&PL_sv_undef>.
a0d0e21e
LW
239Take this code:
240
241 SV* sv = (SV*) 0;
242 if (I-am-to-return-a-real-value) {
243 sv = sv_2mortal(newSViv(42));
244 }
245 sv_setsv(ST(0), sv);
246
247This code tries to return a new SV (which contains the value 42) if it should
04343c6d 248return a real value, or undef otherwise. Instead it has returned a NULL
a0d0e21e 249pointer which, somewhere down the line, will cause a segmentation violation,
06f6df17
RGS
250bus error, or just weird results. Change the zero to C<&PL_sv_undef> in the
251first line and all will be well.
a0d0e21e
LW
252
253To free an SV that you've created, call C<SvREFCNT_dec(SV*)>. Normally this
3fe9a6f1 254call is not necessary (see L<Reference Counts and Mortality>).
a0d0e21e 255
94dde4fb
SC
256=head2 Offsets
257
258Perl provides the function C<sv_chop> to efficiently remove characters
259from the beginning of a string; you give it an SV and a pointer to
da75cd15 260somewhere inside the PV, and it discards everything before the
94dde4fb
SC
261pointer. The efficiency comes by means of a little hack: instead of
262actually removing the characters, C<sv_chop> sets the flag C<OOK>
263(offset OK) to signal to other functions that the offset hack is in
264effect, and it puts the number of bytes chopped off into the IV field
265of the SV. It then moves the PV pointer (called C<SvPVX>) forward that
00aadd71 266many bytes, and adjusts C<SvCUR> and C<SvLEN>.
94dde4fb
SC
267
268Hence, at this point, the start of the buffer that we allocated lives
269at C<SvPVX(sv) - SvIV(sv)> in memory and the PV pointer is pointing
270into the middle of this allocated storage.
271
272This is best demonstrated by example:
273
274 % ./perl -Ilib -MDevel::Peek -le '$a="12345"; $a=~s/.//; Dump($a)'
275 SV = PVIV(0x8128450) at 0x81340f0
276 REFCNT = 1
277 FLAGS = (POK,OOK,pPOK)
278 IV = 1 (OFFSET)
279 PV = 0x8135781 ( "1" . ) "2345"\0
280 CUR = 4
281 LEN = 5
282
283Here the number of bytes chopped off (1) is put into IV, and
284C<Devel::Peek::Dump> helpfully reminds us that this is an offset. The
285portion of the string between the "real" and the "fake" beginnings is
286shown in parentheses, and the values of C<SvCUR> and C<SvLEN> reflect
287the fake beginning, not the real one.
288
fe854a6f 289Something similar to the offset hack is performed on AVs to enable
319cef53
SC
290efficient shifting and splicing off the beginning of the array; while
291C<AvARRAY> points to the first element in the array that is visible from
292Perl, C<AvALLOC> points to the real start of the C array. These are
293usually the same, but a C<shift> operation can be carried out by
6de131f0 294increasing C<AvARRAY> by one and decreasing C<AvFILL> and C<AvMAX>.
319cef53
SC
295Again, the location of the real start of the C array only comes into
296play when freeing the array. See C<av_shift> in F<av.c>.
297
d1b91892 298=head2 What's Really Stored in an SV?
a0d0e21e
LW
299
300Recall that the usual method of determining the type of scalar you have is
5f05dabc 301to use C<Sv*OK> macros. Because a scalar can be both a number and a string,
d1b91892 302usually these macros will always return TRUE and calling the C<Sv*V>
a0d0e21e
LW
303macros will do the appropriate conversion of string to integer/double or
304integer/double to string.
305
306If you I<really> need to know if you have an integer, double, or string
307pointer in an SV, you can use the following three macros instead:
308
309 SvIOKp(SV*)
310 SvNOKp(SV*)
311 SvPOKp(SV*)
312
313These will tell you if you truly have an integer, double, or string pointer
d1b91892 314stored in your SV. The "p" stands for private.
a0d0e21e 315
da8c5729 316There are various ways in which the private and public flags may differ.
9e9796d6
JH
317For example, a tied SV may have a valid underlying value in the IV slot
318(so SvIOKp is true), but the data should be accessed via the FETCH
319routine rather than directly, so SvIOK is false. Another is when
d7f8936a 320numeric conversion has occurred and precision has been lost: only the
9e9796d6
JH
321private flag is set on 'lossy' values. So when an NV is converted to an
322IV with loss, SvIOKp, SvNOKp and SvNOK will be set, while SvIOK wont be.
323
07fa94a1 324In general, though, it's best to use the C<Sv*V> macros.
a0d0e21e 325
54310121 326=head2 Working with AVs
a0d0e21e 327
07fa94a1
JO
328There are two ways to create and load an AV. The first method creates an
329empty AV:
a0d0e21e
LW
330
331 AV* newAV();
332
54310121 333The second method both creates the AV and initially populates it with SVs:
a0d0e21e
LW
334
335 AV* av_make(I32 num, SV **ptr);
336
5f05dabc 337The second argument points to an array containing C<num> C<SV*>'s. Once the
54310121 338AV has been created, the SVs can be destroyed, if so desired.
a0d0e21e 339
da8c5729 340Once the AV has been created, the following operations are possible on it:
a0d0e21e
LW
341
342 void av_push(AV*, SV*);
343 SV* av_pop(AV*);
344 SV* av_shift(AV*);
345 void av_unshift(AV*, I32 num);
346
347These should be familiar operations, with the exception of C<av_unshift>.
348This routine adds C<num> elements at the front of the array with the C<undef>
349value. You must then use C<av_store> (described below) to assign values
350to these new elements.
351
352Here are some other functions:
353
5f05dabc 354 I32 av_len(AV*);
a0d0e21e 355 SV** av_fetch(AV*, I32 key, I32 lval);
a0d0e21e 356 SV** av_store(AV*, I32 key, SV* val);
a0d0e21e 357
da8c5729 358The C<av_len> function returns the highest index value in an array (just
5f05dabc
PP
359like $#array in Perl). If the array is empty, -1 is returned. The
360C<av_fetch> function returns the value at index C<key>, but if C<lval>
361is non-zero, then C<av_fetch> will store an undef value at that index.
04343c6d
GS
362The C<av_store> function stores the value C<val> at index C<key>, and does
363not increment the reference count of C<val>. Thus the caller is responsible
364for taking care of that, and if C<av_store> returns NULL, the caller will
365have to decrement the reference count to avoid a memory leak. Note that
366C<av_fetch> and C<av_store> both return C<SV**>'s, not C<SV*>'s as their
367return value.
d1b91892 368
da8c5729
MH
369A few more:
370
a0d0e21e 371 void av_clear(AV*);
a0d0e21e 372 void av_undef(AV*);
cb1a09d0 373 void av_extend(AV*, I32 key);
5f05dabc
PP
374
375The C<av_clear> function deletes all the elements in the AV* array, but
376does not actually delete the array itself. The C<av_undef> function will
377delete all the elements in the array plus the array itself. The
adc882cf
GS
378C<av_extend> function extends the array so that it contains at least C<key+1>
379elements. If C<key+1> is less than the currently allocated length of the array,
380then nothing is done.
a0d0e21e
LW
381
382If you know the name of an array variable, you can get a pointer to its AV
383by using the following:
384
cbfd0a87 385 AV* get_av("package::varname", 0);
a0d0e21e
LW
386
387This returns NULL if the variable does not exist.
388
04343c6d
GS
389See L<Understanding the Magic of Tied Hashes and Arrays> for more
390information on how to use the array access functions on tied arrays.
391
54310121 392=head2 Working with HVs
a0d0e21e
LW
393
394To create an HV, you use the following routine:
395
396 HV* newHV();
397
da8c5729 398Once the HV has been created, the following operations are possible on it:
a0d0e21e 399
08105a92
GS
400 SV** hv_store(HV*, const char* key, U32 klen, SV* val, U32 hash);
401 SV** hv_fetch(HV*, const char* key, U32 klen, I32 lval);
a0d0e21e 402
5f05dabc
PP
403The C<klen> parameter is the length of the key being passed in (Note that
404you cannot pass 0 in as a value of C<klen> to tell Perl to measure the
405length of the key). The C<val> argument contains the SV pointer to the
54310121 406scalar being stored, and C<hash> is the precomputed hash value (zero if
5f05dabc
PP
407you want C<hv_store> to calculate it for you). The C<lval> parameter
408indicates whether this fetch is actually a part of a store operation, in
409which case a new undefined value will be added to the HV with the supplied
410key and C<hv_fetch> will return as if the value had already existed.
a0d0e21e 411
5f05dabc
PP
412Remember that C<hv_store> and C<hv_fetch> return C<SV**>'s and not just
413C<SV*>. To access the scalar value, you must first dereference the return
414value. However, you should check to make sure that the return value is
415not NULL before dereferencing it.
a0d0e21e 416
da8c5729
MH
417The first of these two functions checks if a hash table entry exists, and the
418second deletes it.
a0d0e21e 419
08105a92
GS
420 bool hv_exists(HV*, const char* key, U32 klen);
421 SV* hv_delete(HV*, const char* key, U32 klen, I32 flags);
a0d0e21e 422
5f05dabc
PP
423If C<flags> does not include the C<G_DISCARD> flag then C<hv_delete> will
424create and return a mortal copy of the deleted value.
425
a0d0e21e
LW
426And more miscellaneous functions:
427
428 void hv_clear(HV*);
a0d0e21e 429 void hv_undef(HV*);
5f05dabc
PP
430
431Like their AV counterparts, C<hv_clear> deletes all the entries in the hash
432table but does not actually delete the hash table. The C<hv_undef> deletes
433both the entries and the hash table itself.
a0d0e21e 434
a9b0660e 435Perl keeps the actual data in a linked list of structures with a typedef of HE.
d1b91892
AD
436These contain the actual key and value pointers (plus extra administrative
437overhead). The key is a string pointer; the value is an C<SV*>. However,
438once you have an C<HE*>, to get the actual key and value, use the routines
439specified below.
440
a0d0e21e
LW
441 I32 hv_iterinit(HV*);
442 /* Prepares starting point to traverse hash table */
443 HE* hv_iternext(HV*);
444 /* Get the next entry, and return a pointer to a
445 structure that has both the key and value */
446 char* hv_iterkey(HE* entry, I32* retlen);
447 /* Get the key from an HE structure and also return
448 the length of the key string */
cb1a09d0 449 SV* hv_iterval(HV*, HE* entry);
d1be9408 450 /* Return an SV pointer to the value of the HE
a0d0e21e 451 structure */
cb1a09d0 452 SV* hv_iternextsv(HV*, char** key, I32* retlen);
d1b91892
AD
453 /* This convenience routine combines hv_iternext,
454 hv_iterkey, and hv_iterval. The key and retlen
455 arguments are return values for the key and its
456 length. The value is returned in the SV* argument */
a0d0e21e
LW
457
458If you know the name of a hash variable, you can get a pointer to its HV
459by using the following:
460
6673a63c 461 HV* get_hv("package::varname", 0);
a0d0e21e
LW
462
463This returns NULL if the variable does not exist.
464
a43e7901 465The hash algorithm is defined in the C<PERL_HASH> macro:
a0d0e21e 466
a43e7901 467 PERL_HASH(hash, key, klen)
ab192400 468
a43e7901
YO
469The exact implementation of this macro varies by architecture and version
470of perl, and the return value may change per invocation, so the value
471is only valid for the duration of a single perl process.
a0d0e21e 472
04343c6d
GS
473See L<Understanding the Magic of Tied Hashes and Arrays> for more
474information on how to use the hash access functions on tied hashes.
475
1e422769
PP
476=head2 Hash API Extensions
477
478Beginning with version 5.004, the following functions are also supported:
479
480 HE* hv_fetch_ent (HV* tb, SV* key, I32 lval, U32 hash);
481 HE* hv_store_ent (HV* tb, SV* key, SV* val, U32 hash);
c47ff5f1 482
1e422769
PP
483 bool hv_exists_ent (HV* tb, SV* key, U32 hash);
484 SV* hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash);
c47ff5f1 485
1e422769
PP
486 SV* hv_iterkeysv (HE* entry);
487
488Note that these functions take C<SV*> keys, which simplifies writing
489of extension code that deals with hash structures. These functions
490also allow passing of C<SV*> keys to C<tie> functions without forcing
491you to stringify the keys (unlike the previous set of functions).
492
493They also return and accept whole hash entries (C<HE*>), making their
494use more efficient (since the hash number for a particular string
4a4eefd0
GS
495doesn't have to be recomputed every time). See L<perlapi> for detailed
496descriptions.
1e422769
PP
497
498The following macros must always be used to access the contents of hash
499entries. Note that the arguments to these macros must be simple
500variables, since they may get evaluated more than once. See
4a4eefd0 501L<perlapi> for detailed descriptions of these macros.
1e422769
PP
502
503 HePV(HE* he, STRLEN len)
504 HeVAL(HE* he)
505 HeHASH(HE* he)
506 HeSVKEY(HE* he)
507 HeSVKEY_force(HE* he)
508 HeSVKEY_set(HE* he, SV* sv)
509
510These two lower level macros are defined, but must only be used when
511dealing with keys that are not C<SV*>s:
512
513 HeKEY(HE* he)
514 HeKLEN(HE* he)
515
04343c6d
GS
516Note that both C<hv_store> and C<hv_store_ent> do not increment the
517reference count of the stored C<val>, which is the caller's responsibility.
518If these functions return a NULL value, the caller will usually have to
519decrement the reference count of C<val> to avoid a memory leak.
1e422769 520
a9381218
MHM
521=head2 AVs, HVs and undefined values
522
523Sometimes you have to store undefined values in AVs or HVs. Although
524this may be a rare case, it can be tricky. That's because you're
525used to using C<&PL_sv_undef> if you need an undefined SV.
526
527For example, intuition tells you that this XS code:
528
529 AV *av = newAV();
530 av_store( av, 0, &PL_sv_undef );
531
532is equivalent to this Perl code:
533
534 my @av;
535 $av[0] = undef;
536
537Unfortunately, this isn't true. AVs use C<&PL_sv_undef> as a marker
538for indicating that an array element has not yet been initialized.
539Thus, C<exists $av[0]> would be true for the above Perl code, but
540false for the array generated by the XS code.
541
542Other problems can occur when storing C<&PL_sv_undef> in HVs:
543
544 hv_store( hv, "key", 3, &PL_sv_undef, 0 );
545
546This will indeed make the value C<undef>, but if you try to modify
547the value of C<key>, you'll get the following error:
548
549 Modification of non-creatable hash value attempted
550
551In perl 5.8.0, C<&PL_sv_undef> was also used to mark placeholders
552in restricted hashes. This caused such hash entries not to appear
553when iterating over the hash or when checking for the keys
554with the C<hv_exists> function.
555
8abccac8
AR
556You can run into similar problems when you store C<&PL_sv_yes> or
557C<&PL_sv_no> into AVs or HVs. Trying to modify such elements
a9381218
MHM
558will give you the following error:
559
560 Modification of a read-only value attempted
561
562To make a long story short, you can use the special variables
8abccac8 563C<&PL_sv_undef>, C<&PL_sv_yes> and C<&PL_sv_no> with AVs and
a9381218
MHM
564HVs, but you have to make sure you know what you're doing.
565
566Generally, if you want to store an undefined value in an AV
567or HV, you should not use C<&PL_sv_undef>, but rather create a
568new undefined value using the C<newSV> function, for example:
569
570 av_store( av, 42, newSV(0) );
571 hv_store( hv, "foo", 3, newSV(0), 0 );
572
a0d0e21e
LW
573=head2 References
574
d1b91892 575References are a special type of scalar that point to other data types
a9b0660e 576(including other references).
a0d0e21e 577
07fa94a1 578To create a reference, use either of the following functions:
a0d0e21e 579
5f05dabc
PP
580 SV* newRV_inc((SV*) thing);
581 SV* newRV_noinc((SV*) thing);
a0d0e21e 582
5f05dabc 583The C<thing> argument can be any of an C<SV*>, C<AV*>, or C<HV*>. The
07fa94a1
JO
584functions are identical except that C<newRV_inc> increments the reference
585count of the C<thing>, while C<newRV_noinc> does not. For historical
586reasons, C<newRV> is a synonym for C<newRV_inc>.
587
588Once you have a reference, you can use the following macro to dereference
589the reference:
a0d0e21e
LW
590
591 SvRV(SV*)
592
593then call the appropriate routines, casting the returned C<SV*> to either an
d1b91892 594C<AV*> or C<HV*>, if required.
a0d0e21e 595
d1b91892 596To determine if an SV is a reference, you can use the following macro:
a0d0e21e
LW
597
598 SvROK(SV*)
599
07fa94a1
JO
600To discover what type of value the reference refers to, use the following
601macro and then check the return value.
d1b91892
AD
602
603 SvTYPE(SvRV(SV*))
604
605The most useful types that will be returned are:
606
a5e62da0
FC
607 < SVt_PVAV Scalar
608 SVt_PVAV Array
609 SVt_PVHV Hash
610 SVt_PVCV Code
611 SVt_PVGV Glob (possibly a file handle)
612
613See L<perlapi/svtype> for more details.
d1b91892 614
cb1a09d0
AD
615=head2 Blessed References and Class Objects
616
06f6df17 617References are also used to support object-oriented programming. In perl's
cb1a09d0
AD
618OO lexicon, an object is simply a reference that has been blessed into a
619package (or class). Once blessed, the programmer may now use the reference
620to access the various methods in the class.
621
622A reference can be blessed into a package with the following function:
623
624 SV* sv_bless(SV* sv, HV* stash);
625
06f6df17
RGS
626The C<sv> argument must be a reference value. The C<stash> argument
627specifies which class the reference will belong to. See
2ae324a7 628L<Stashes and Globs> for information on converting class names into stashes.
cb1a09d0
AD
629
630/* Still under construction */
631
ddd2cc91
DM
632The following function upgrades rv to reference if not already one.
633Creates a new SV for rv to point to. If C<classname> is non-null, the SV
634is blessed into the specified class. SV is returned.
cb1a09d0 635
08105a92 636 SV* newSVrv(SV* rv, const char* classname);
cb1a09d0 637
ddd2cc91
DM
638The following three functions copy integer, unsigned integer or double
639into an SV whose reference is C<rv>. SV is blessed if C<classname> is
640non-null.
cb1a09d0 641
08105a92 642 SV* sv_setref_iv(SV* rv, const char* classname, IV iv);
e1c57cef 643 SV* sv_setref_uv(SV* rv, const char* classname, UV uv);
08105a92 644 SV* sv_setref_nv(SV* rv, const char* classname, NV iv);
cb1a09d0 645
ddd2cc91
DM
646The following function copies the pointer value (I<the address, not the
647string!>) into an SV whose reference is rv. SV is blessed if C<classname>
648is non-null.
cb1a09d0 649
ddd2cc91 650 SV* sv_setref_pv(SV* rv, const char* classname, void* pv);
cb1a09d0 651
a9b0660e 652The following function copies a string into an SV whose reference is C<rv>.
ddd2cc91
DM
653Set length to 0 to let Perl calculate the string length. SV is blessed if
654C<classname> is non-null.
cb1a09d0 655
a9b0660e
KW
656 SV* sv_setref_pvn(SV* rv, const char* classname, char* pv,
657 STRLEN length);
cb1a09d0 658
ddd2cc91
DM
659The following function tests whether the SV is blessed into the specified
660class. It does not check inheritance relationships.
9abd00ed 661
08105a92 662 int sv_isa(SV* sv, const char* name);
9abd00ed 663
ddd2cc91 664The following function tests whether the SV is a reference to a blessed object.
9abd00ed
GS
665
666 int sv_isobject(SV* sv);
667
ddd2cc91
DM
668The following function tests whether the SV is derived from the specified
669class. SV can be either a reference to a blessed object or a string
670containing a class name. This is the function implementing the
671C<UNIVERSAL::isa> functionality.
9abd00ed 672
08105a92 673 bool sv_derived_from(SV* sv, const char* name);
9abd00ed 674
00aadd71 675To check if you've got an object derived from a specific class you have
9abd00ed
GS
676to write:
677
678 if (sv_isobject(sv) && sv_derived_from(sv, class)) { ... }
cb1a09d0 679
5f05dabc 680=head2 Creating New Variables
cb1a09d0 681
5f05dabc
PP
682To create a new Perl variable with an undef value which can be accessed from
683your Perl script, use the following routines, depending on the variable type.
cb1a09d0 684
64ace3f8 685 SV* get_sv("package::varname", GV_ADD);
cbfd0a87 686 AV* get_av("package::varname", GV_ADD);
6673a63c 687 HV* get_hv("package::varname", GV_ADD);
cb1a09d0 688
058a5f6c 689Notice the use of GV_ADD as the second parameter. The new variable can now
cb1a09d0
AD
690be set, using the routines appropriate to the data type.
691
5f05dabc 692There are additional macros whose values may be bitwise OR'ed with the
058a5f6c 693C<GV_ADD> argument to enable certain extra features. Those bits are:
cb1a09d0 694
9a68f1db
SB
695=over
696
697=item GV_ADDMULTI
698
699Marks the variable as multiply defined, thus preventing the:
700
701 Name <varname> used only once: possible typo
702
703warning.
704
9a68f1db
SB
705=item GV_ADDWARN
706
707Issues the warning:
708
709 Had to create <varname> unexpectedly
710
711if the variable did not exist before the function was called.
712
713=back
cb1a09d0 714
07fa94a1
JO
715If you do not specify a package name, the variable is created in the current
716package.
cb1a09d0 717
5f05dabc 718=head2 Reference Counts and Mortality
a0d0e21e 719
d1be9408 720Perl uses a reference count-driven garbage collection mechanism. SVs,
54310121 721AVs, or HVs (xV for short in the following) start their life with a
55497cff 722reference count of 1. If the reference count of an xV ever drops to 0,
07fa94a1 723then it will be destroyed and its memory made available for reuse.
55497cff
PP
724
725This normally doesn't happen at the Perl level unless a variable is
5f05dabc
PP
726undef'ed or the last variable holding a reference to it is changed or
727overwritten. At the internal level, however, reference counts can be
55497cff
PP
728manipulated with the following macros:
729
730 int SvREFCNT(SV* sv);
5f05dabc 731 SV* SvREFCNT_inc(SV* sv);
55497cff
PP
732 void SvREFCNT_dec(SV* sv);
733
734However, there is one other function which manipulates the reference
07fa94a1
JO
735count of its argument. The C<newRV_inc> function, you will recall,
736creates a reference to the specified argument. As a side effect,
737it increments the argument's reference count. If this is not what
738you want, use C<newRV_noinc> instead.
739
740For example, imagine you want to return a reference from an XSUB function.
741Inside the XSUB routine, you create an SV which initially has a reference
742count of one. Then you call C<newRV_inc>, passing it the just-created SV.
5f05dabc
PP
743This returns the reference as a new SV, but the reference count of the
744SV you passed to C<newRV_inc> has been incremented to two. Now you
07fa94a1
JO
745return the reference from the XSUB routine and forget about the SV.
746But Perl hasn't! Whenever the returned reference is destroyed, the
747reference count of the original SV is decreased to one and nothing happens.
748The SV will hang around without any way to access it until Perl itself
749terminates. This is a memory leak.
5f05dabc
PP
750
751The correct procedure, then, is to use C<newRV_noinc> instead of
faed5253
JO
752C<newRV_inc>. Then, if and when the last reference is destroyed,
753the reference count of the SV will go to zero and it will be destroyed,
07fa94a1 754stopping any memory leak.
55497cff 755
5f05dabc 756There are some convenience functions available that can help with the
54310121 757destruction of xVs. These functions introduce the concept of "mortality".
07fa94a1
JO
758An xV that is mortal has had its reference count marked to be decremented,
759but not actually decremented, until "a short time later". Generally the
760term "short time later" means a single Perl statement, such as a call to
54310121 761an XSUB function. The actual determinant for when mortal xVs have their
07fa94a1
JO
762reference count decremented depends on two macros, SAVETMPS and FREETMPS.
763See L<perlcall> and L<perlxs> for more details on these macros.
55497cff
PP
764
765"Mortalization" then is at its simplest a deferred C<SvREFCNT_dec>.
766However, if you mortalize a variable twice, the reference count will
767later be decremented twice.
768
00aadd71
NIS
769"Mortal" SVs are mainly used for SVs that are placed on perl's stack.
770For example an SV which is created just to pass a number to a called sub
06f6df17
RGS
771is made mortal to have it cleaned up automatically when it's popped off
772the stack. Similarly, results returned by XSUBs (which are pushed on the
773stack) are often made mortal.
a0d0e21e
LW
774
775To create a mortal variable, use the functions:
776
777 SV* sv_newmortal()
778 SV* sv_2mortal(SV*)
779 SV* sv_mortalcopy(SV*)
780
00aadd71 781The first call creates a mortal SV (with no value), the second converts an existing
5f05dabc
PP
782SV to a mortal SV (and thus defers a call to C<SvREFCNT_dec>), and the
783third creates a mortal copy of an existing SV.
da8c5729 784Because C<sv_newmortal> gives the new SV no value, it must normally be given one
9a68f1db 785via C<sv_setpv>, C<sv_setiv>, etc. :
00aadd71
NIS
786
787 SV *tmp = sv_newmortal();
788 sv_setiv(tmp, an_integer);
789
790As that is multiple C statements it is quite common so see this idiom instead:
791
792 SV *tmp = sv_2mortal(newSViv(an_integer));
793
794
795You should be careful about creating mortal variables. Strange things
796can happen if you make the same value mortal within multiple contexts,
797or if you make a variable mortal multiple times. Thinking of "Mortalization"
798as deferred C<SvREFCNT_dec> should help to minimize such problems.
da8c5729 799For example if you are passing an SV which you I<know> has a high enough REFCNT
00aadd71
NIS
800to survive its use on the stack you need not do any mortalization.
801If you are not sure then doing an C<SvREFCNT_inc> and C<sv_2mortal>, or
802making a C<sv_mortalcopy> is safer.
a0d0e21e 803
ac036724 804The mortal routines are not just for SVs; AVs and HVs can be
faed5253 805made mortal by passing their address (type-casted to C<SV*>) to the
07fa94a1 806C<sv_2mortal> or C<sv_mortalcopy> routines.
a0d0e21e 807
5f05dabc 808=head2 Stashes and Globs
a0d0e21e 809
06f6df17
RGS
810A B<stash> is a hash that contains all variables that are defined
811within a package. Each key of the stash is a symbol
aa689395
PP
812name (shared by all the different types of objects that have the same
813name), and each value in the hash table is a GV (Glob Value). This GV
814in turn contains references to the various objects of that name,
815including (but not limited to) the following:
cb1a09d0 816
a0d0e21e
LW
817 Scalar Value
818 Array Value
819 Hash Value
a3cb178b 820 I/O Handle
a0d0e21e
LW
821 Format
822 Subroutine
823
06f6df17
RGS
824There is a single stash called C<PL_defstash> that holds the items that exist
825in the C<main> package. To get at the items in other packages, append the
826string "::" to the package name. The items in the C<Foo> package are in
827the stash C<Foo::> in PL_defstash. The items in the C<Bar::Baz> package are
828in the stash C<Baz::> in C<Bar::>'s stash.
a0d0e21e 829
d1b91892 830To get the stash pointer for a particular package, use the function:
a0d0e21e 831
da51bb9b
NC
832 HV* gv_stashpv(const char* name, I32 flags)
833 HV* gv_stashsv(SV*, I32 flags)
a0d0e21e
LW
834
835The first function takes a literal string, the second uses the string stored
d1b91892 836in the SV. Remember that a stash is just a hash table, so you get back an
da51bb9b 837C<HV*>. The C<flags> flag will create a new package if it is set to GV_ADD.
a0d0e21e
LW
838
839The name that C<gv_stash*v> wants is the name of the package whose symbol table
840you want. The default package is called C<main>. If you have multiply nested
d1b91892
AD
841packages, pass their names to C<gv_stash*v>, separated by C<::> as in the Perl
842language itself.
a0d0e21e
LW
843
844Alternately, if you have an SV that is a blessed reference, you can find
845out the stash pointer by using:
846
847 HV* SvSTASH(SvRV(SV*));
848
849then use the following to get the package name itself:
850
851 char* HvNAME(HV* stash);
852
5f05dabc
PP
853If you need to bless or re-bless an object you can use the following
854function:
a0d0e21e
LW
855
856 SV* sv_bless(SV*, HV* stash)
857
858where the first argument, an C<SV*>, must be a reference, and the second
859argument is a stash. The returned C<SV*> can now be used in the same way
860as any other SV.
861
d1b91892
AD
862For more information on references and blessings, consult L<perlref>.
863
54310121 864=head2 Double-Typed SVs
0a753a76
PP
865
866Scalar variables normally contain only one type of value, an integer,
867double, pointer, or reference. Perl will automatically convert the
868actual scalar data from the stored type into the requested type.
869
870Some scalar variables contain more than one type of scalar data. For
871example, the variable C<$!> contains either the numeric value of C<errno>
872or its string equivalent from either C<strerror> or C<sys_errlist[]>.
873
874To force multiple data values into an SV, you must do two things: use the
875C<sv_set*v> routines to add the additional scalar type, then set a flag
876so that Perl will believe it contains more than one type of data. The
877four macros to set the flags are:
878
879 SvIOK_on
880 SvNOK_on
881 SvPOK_on
882 SvROK_on
883
884The particular macro you must use depends on which C<sv_set*v> routine
885you called first. This is because every C<sv_set*v> routine turns on
886only the bit for the particular type of data being set, and turns off
887all the rest.
888
889For example, to create a new Perl variable called "dberror" that contains
890both the numeric and descriptive string error values, you could use the
891following code:
892
893 extern int dberror;
894 extern char *dberror_list;
895
64ace3f8 896 SV* sv = get_sv("dberror", GV_ADD);
0a753a76
PP
897 sv_setiv(sv, (IV) dberror);
898 sv_setpv(sv, dberror_list[dberror]);
899 SvIOK_on(sv);
900
901If the order of C<sv_setiv> and C<sv_setpv> had been reversed, then the
902macro C<SvPOK_on> would need to be called instead of C<SvIOK_on>.
903
904=head2 Magic Variables
a0d0e21e 905
d1b91892
AD
906[This section still under construction. Ignore everything here. Post no
907bills. Everything not permitted is forbidden.]
908
d1b91892
AD
909Any SV may be magical, that is, it has special features that a normal
910SV does not have. These features are stored in the SV structure in a
5f05dabc 911linked list of C<struct magic>'s, typedef'ed to C<MAGIC>.
d1b91892
AD
912
913 struct magic {
914 MAGIC* mg_moremagic;
915 MGVTBL* mg_virtual;
916 U16 mg_private;
917 char mg_type;
918 U8 mg_flags;
b205eb13 919 I32 mg_len;
d1b91892
AD
920 SV* mg_obj;
921 char* mg_ptr;
d1b91892
AD
922 };
923
924Note this is current as of patchlevel 0, and could change at any time.
925
926=head2 Assigning Magic
927
928Perl adds magic to an SV using the sv_magic function:
929
a9b0660e 930 void sv_magic(SV* sv, SV* obj, int how, const char* name, I32 namlen);
d1b91892
AD
931
932The C<sv> argument is a pointer to the SV that is to acquire a new magical
933feature.
934
935If C<sv> is not already magical, Perl uses the C<SvUPGRADE> macro to
645c22ef
DM
936convert C<sv> to type C<SVt_PVMG>. Perl then continues by adding new magic
937to the beginning of the linked list of magical features. Any prior entry
938of the same type of magic is deleted. Note that this can be overridden,
939and multiple instances of the same type of magic can be associated with an
940SV.
d1b91892 941
54310121
PP
942The C<name> and C<namlen> arguments are used to associate a string with
943the magic, typically the name of a variable. C<namlen> is stored in the
2d8d5d5a
SH
944C<mg_len> field and if C<name> is non-null then either a C<savepvn> copy of
945C<name> or C<name> itself is stored in the C<mg_ptr> field, depending on
946whether C<namlen> is greater than zero or equal to zero respectively. As a
947special case, if C<(name && namlen == HEf_SVKEY)> then C<name> is assumed
948to contain an C<SV*> and is stored as-is with its REFCNT incremented.
d1b91892
AD
949
950The sv_magic function uses C<how> to determine which, if any, predefined
951"Magic Virtual Table" should be assigned to the C<mg_virtual> field.
06f6df17 952See the L<Magic Virtual Tables> section below. The C<how> argument is also
14befaf4 953stored in the C<mg_type> field. The value of C<how> should be chosen
06f6df17 954from the set of macros C<PERL_MAGIC_foo> found in F<perl.h>. Note that before
645c22ef 955these macros were added, Perl internals used to directly use character
14befaf4 956literals, so you may occasionally come across old code or documentation
75d0f26d 957referring to 'U' magic rather than C<PERL_MAGIC_uvar> for example.
d1b91892
AD
958
959The C<obj> argument is stored in the C<mg_obj> field of the C<MAGIC>
960structure. If it is not the same as the C<sv> argument, the reference
961count of the C<obj> object is incremented. If it is the same, or if
645c22ef 962the C<how> argument is C<PERL_MAGIC_arylen>, or if it is a NULL pointer,
14befaf4 963then C<obj> is merely stored, without the reference count being incremented.
d1b91892 964
2d8d5d5a
SH
965See also C<sv_magicext> in L<perlapi> for a more flexible way to add magic
966to an SV.
967
cb1a09d0
AD
968There is also a function to add magic to an C<HV>:
969
970 void hv_magic(HV *hv, GV *gv, int how);
971
972This simply calls C<sv_magic> and coerces the C<gv> argument into an C<SV>.
973
974To remove the magic from an SV, call the function sv_unmagic:
975
70a53b35 976 int sv_unmagic(SV *sv, int type);
cb1a09d0
AD
977
978The C<type> argument should be equal to the C<how> value when the C<SV>
979was initially made magical.
980
f6ee7b17
FR
981However, note that C<sv_unmagic> removes all magic of a certain C<type> from the
982C<SV>. If you want to remove only certain magic of a C<type> based on the magic
983virtual table, use C<sv_unmagicext> instead:
984
985 int sv_unmagicext(SV *sv, int type, MGVTBL *vtbl);
986
d1b91892
AD
987=head2 Magic Virtual Tables
988
d1be9408 989The C<mg_virtual> field in the C<MAGIC> structure is a pointer to an
d1b91892
AD
990C<MGVTBL>, which is a structure of function pointers and stands for
991"Magic Virtual Table" to handle the various operations that might be
992applied to that variable.
993
301cb7e8
DM
994The C<MGVTBL> has five (or sometimes eight) pointers to the following
995routine types:
d1b91892
AD
996
997 int (*svt_get)(SV* sv, MAGIC* mg);
998 int (*svt_set)(SV* sv, MAGIC* mg);
999 U32 (*svt_len)(SV* sv, MAGIC* mg);
1000 int (*svt_clear)(SV* sv, MAGIC* mg);
1001 int (*svt_free)(SV* sv, MAGIC* mg);
1002
a9b0660e
KW
1003 int (*svt_copy)(SV *sv, MAGIC* mg, SV *nsv,
1004 const char *name, I32 namlen);
301cb7e8
DM
1005 int (*svt_dup)(MAGIC *mg, CLONE_PARAMS *param);
1006 int (*svt_local)(SV *nsv, MAGIC *mg);
1007
1008
06f6df17 1009This MGVTBL structure is set at compile-time in F<perl.h> and there are
b7a0f54c
SM
1010currently 32 types. These different structures contain pointers to various
1011routines that perform additional actions depending on which function is
1012being called.
d1b91892 1013
a9b0660e
KW
1014 Function pointer Action taken
1015 ---------------- ------------
1016 svt_get Do something before the value of the SV is
1017 retrieved.
1018 svt_set Do something after the SV is assigned a value.
1019 svt_len Report on the SV's length.
1020 svt_clear Clear something the SV represents.
1021 svt_free Free any extra storage associated with the SV.
d1b91892 1022
a9b0660e
KW
1023 svt_copy copy tied variable magic to a tied element
1024 svt_dup duplicate a magic structure during thread cloning
1025 svt_local copy magic to local value during 'local'
301cb7e8 1026
d1b91892 1027For instance, the MGVTBL structure called C<vtbl_sv> (which corresponds
14befaf4 1028to an C<mg_type> of C<PERL_MAGIC_sv>) contains:
d1b91892
AD
1029
1030 { magic_get, magic_set, magic_len, 0, 0 }
1031
14befaf4
DM
1032Thus, when an SV is determined to be magical and of type C<PERL_MAGIC_sv>,
1033if a get operation is being performed, the routine C<magic_get> is
1034called. All the various routines for the various magical types begin
1035with C<magic_>. NOTE: the magic routines are not considered part of
1036the Perl API, and may not be exported by the Perl library.
d1b91892 1037
301cb7e8
DM
1038The last three slots are a recent addition, and for source code
1039compatibility they are only checked for if one of the three flags
1040MGf_COPY, MGf_DUP or MGf_LOCAL is set in mg_flags. This means that most
1041code can continue declaring a vtable as a 5-element value. These three are
1042currently used exclusively by the threading code, and are highly subject
1043to change.
1044
d1b91892
AD
1045The current kinds of Magic Virtual Tables are:
1046
f1f5ddd7
FC
1047=for comment
1048This table is generated by regen/mg_vtable.pl. Any changes made here
1049will be lost.
1050
1051=for mg_vtable.pl begin
1052
a9b0660e 1053 mg_type
bd6e6c12
FC
1054 (old-style char and macro) MGVTBL Type of magic
1055 -------------------------- ------ -------------
1056 \0 PERL_MAGIC_sv vtbl_sv Special scalar variable
1057 # PERL_MAGIC_arylen vtbl_arylen Array length ($#ary)
1058 % PERL_MAGIC_rhash (none) extra data for restricted
1059 hashes
2a388207 1060 & PERL_MAGIC_proto (none) my sub prototype CV
bd6e6c12
FC
1061 . PERL_MAGIC_pos vtbl_pos pos() lvalue
1062 : PERL_MAGIC_symtab (none) extra data for symbol
1063 tables
1064 < PERL_MAGIC_backref vtbl_backref for weak ref data
1065 @ PERL_MAGIC_arylen_p (none) to move arylen out of XPVAV
1066 B PERL_MAGIC_bm vtbl_regexp Boyer-Moore
1067 (fast string search)
1068 c PERL_MAGIC_overload_table vtbl_ovrld Holds overload table
1069 (AMT) on stash
1070 D PERL_MAGIC_regdata vtbl_regdata Regex match position data
1071 (@+ and @- vars)
1072 d PERL_MAGIC_regdatum vtbl_regdatum Regex match position data
1073 element
1074 E PERL_MAGIC_env vtbl_env %ENV hash
1075 e PERL_MAGIC_envelem vtbl_envelem %ENV hash element
eccba044 1076 f PERL_MAGIC_fm vtbl_regexp Formline
bd6e6c12 1077 ('compiled' format)
bd6e6c12
FC
1078 g PERL_MAGIC_regex_global vtbl_mglob m//g target
1079 H PERL_MAGIC_hints vtbl_hints %^H hash
1080 h PERL_MAGIC_hintselem vtbl_hintselem %^H hash element
1081 I PERL_MAGIC_isa vtbl_isa @ISA array
1082 i PERL_MAGIC_isaelem vtbl_isaelem @ISA array element
1083 k PERL_MAGIC_nkeys vtbl_nkeys scalar(keys()) lvalue
1084 L PERL_MAGIC_dbfile (none) Debugger %_<filename
1085 l PERL_MAGIC_dbline vtbl_dbline Debugger %_<filename
1086 element
1087 N PERL_MAGIC_shared (none) Shared between threads
1088 n PERL_MAGIC_shared_scalar (none) Shared between threads
1089 o PERL_MAGIC_collxfrm vtbl_collxfrm Locale transformation
1090 P PERL_MAGIC_tied vtbl_pack Tied array or hash
1091 p PERL_MAGIC_tiedelem vtbl_packelem Tied array or hash element
1092 q PERL_MAGIC_tiedscalar vtbl_packelem Tied scalar or handle
1093 r PERL_MAGIC_qr vtbl_regexp precompiled qr// regex
1094 S PERL_MAGIC_sig (none) %SIG hash
1095 s PERL_MAGIC_sigelem vtbl_sigelem %SIG hash element
1096 t PERL_MAGIC_taint vtbl_taint Taintedness
1097 U PERL_MAGIC_uvar vtbl_uvar Available for use by
1098 extensions
1099 u PERL_MAGIC_uvar_elem (none) Reserved for use by
1100 extensions
4499db73 1101 V PERL_MAGIC_vstring (none) SV was vstring literal
bd6e6c12
FC
1102 v PERL_MAGIC_vec vtbl_vec vec() lvalue
1103 w PERL_MAGIC_utf8 vtbl_utf8 Cached UTF-8 information
1104 x PERL_MAGIC_substr vtbl_substr substr() lvalue
1105 y PERL_MAGIC_defelem vtbl_defelem Shadow "foreach" iterator
1106 variable / smart parameter
1107 vivification
1108 ] PERL_MAGIC_checkcall vtbl_checkcall inlining/mutation of call
1109 to this CV
1110 ~ PERL_MAGIC_ext (none) Available for use by
1111 extensions
0cbee0a4 1112
f1f5ddd7 1113=for mg_vtable.pl end
d1b91892 1114
68dc0745 1115When an uppercase and lowercase letter both exist in the table, then the
92f0c265
JP
1116uppercase letter is typically used to represent some kind of composite type
1117(a list or a hash), and the lowercase letter is used to represent an element
1118of that composite type. Some internals code makes use of this case
1119relationship. However, 'v' and 'V' (vec and v-string) are in no way related.
14befaf4
DM
1120
1121The C<PERL_MAGIC_ext> and C<PERL_MAGIC_uvar> magic types are defined
1122specifically for use by extensions and will not be used by perl itself.
1123Extensions can use C<PERL_MAGIC_ext> magic to 'attach' private information
1124to variables (typically objects). This is especially useful because
1125there is no way for normal perl code to corrupt this private information
1126(unlike using extra elements of a hash object).
1127
1128Similarly, C<PERL_MAGIC_uvar> magic can be used much like tie() to call a
1129C function any time a scalar's value is used or changed. The C<MAGIC>'s
bdbeb323
SM
1130C<mg_ptr> field points to a C<ufuncs> structure:
1131
1132 struct ufuncs {
a9402793
AB
1133 I32 (*uf_val)(pTHX_ IV, SV*);
1134 I32 (*uf_set)(pTHX_ IV, SV*);
bdbeb323
SM
1135 IV uf_index;
1136 };
1137
1138When the SV is read from or written to, the C<uf_val> or C<uf_set>
14befaf4
DM
1139function will be called with C<uf_index> as the first arg and a pointer to
1140the SV as the second. A simple example of how to add C<PERL_MAGIC_uvar>
1526ead6
AB
1141magic is shown below. Note that the ufuncs structure is copied by
1142sv_magic, so you can safely allocate it on the stack.
1143
1144 void
1145 Umagic(sv)
1146 SV *sv;
1147 PREINIT:
1148 struct ufuncs uf;
1149 CODE:
1150 uf.uf_val = &my_get_fn;
1151 uf.uf_set = &my_set_fn;
1152 uf.uf_index = 0;
14befaf4 1153 sv_magic(sv, 0, PERL_MAGIC_uvar, (char*)&uf, sizeof(uf));
5f05dabc 1154
1e73acc8
AS
1155Attaching C<PERL_MAGIC_uvar> to arrays is permissible but has no effect.
1156
1157For hashes there is a specialized hook that gives control over hash
1158keys (but not values). This hook calls C<PERL_MAGIC_uvar> 'get' magic
1159if the "set" function in the C<ufuncs> structure is NULL. The hook
1160is activated whenever the hash is accessed with a key specified as
1161an C<SV> through the functions C<hv_store_ent>, C<hv_fetch_ent>,
1162C<hv_delete_ent>, and C<hv_exists_ent>. Accessing the key as a string
1163through the functions without the C<..._ent> suffix circumvents the
4509d391 1164hook. See L<Hash::Util::FieldHash/GUTS> for a detailed description.
1e73acc8 1165
14befaf4
DM
1166Note that because multiple extensions may be using C<PERL_MAGIC_ext>
1167or C<PERL_MAGIC_uvar> magic, it is important for extensions to take
1168extra care to avoid conflict. Typically only using the magic on
1169objects blessed into the same class as the extension is sufficient.
2f07f21a
FR
1170For C<PERL_MAGIC_ext> magic, it is usually a good idea to define an
1171C<MGVTBL>, even if all its fields will be C<0>, so that individual
1172C<MAGIC> pointers can be identified as a particular kind of magic
f6ee7b17
FR
1173using their magic virtual table. C<mg_findext> provides an easy way
1174to do that:
2f07f21a
FR
1175
1176 STATIC MGVTBL my_vtbl = { 0, 0, 0, 0, 0, 0, 0, 0 };
1177
1178 MAGIC *mg;
f6ee7b17
FR
1179 if ((mg = mg_findext(sv, PERL_MAGIC_ext, &my_vtbl))) {
1180 /* this is really ours, not another module's PERL_MAGIC_ext */
1181 my_priv_data_t *priv = (my_priv_data_t *)mg->mg_ptr;
1182 ...
2f07f21a 1183 }
5f05dabc 1184
ef50df4b
GS
1185Also note that the C<sv_set*()> and C<sv_cat*()> functions described
1186earlier do B<not> invoke 'set' magic on their targets. This must
1187be done by the user either by calling the C<SvSETMAGIC()> macro after
1188calling these functions, or by using one of the C<sv_set*_mg()> or
1189C<sv_cat*_mg()> functions. Similarly, generic C code must call the
1190C<SvGETMAGIC()> macro to invoke any 'get' magic if they use an SV
1191obtained from external sources in functions that don't handle magic.
4a4eefd0 1192See L<perlapi> for a description of these functions.
189b2af5
GS
1193For example, calls to the C<sv_cat*()> functions typically need to be
1194followed by C<SvSETMAGIC()>, but they don't need a prior C<SvGETMAGIC()>
1195since their implementation handles 'get' magic.
1196
d1b91892
AD
1197=head2 Finding Magic
1198
a9b0660e
KW
1199 MAGIC *mg_find(SV *sv, int type); /* Finds the magic pointer of that
1200 * type */
f6ee7b17
FR
1201
1202This routine returns a pointer to a C<MAGIC> structure stored in the SV.
1203If the SV does not have that magical feature, C<NULL> is returned. If the
1204SV has multiple instances of that magical feature, the first one will be
1205returned. C<mg_findext> can be used to find a C<MAGIC> structure of an SV
da8c5729 1206based on both its magic type and its magic virtual table:
f6ee7b17
FR
1207
1208 MAGIC *mg_findext(SV *sv, int type, MGVTBL *vtbl);
d1b91892 1209
f6ee7b17
FR
1210Also, if the SV passed to C<mg_find> or C<mg_findext> is not of type
1211SVt_PVMG, Perl may core dump.
d1b91892 1212
08105a92 1213 int mg_copy(SV* sv, SV* nsv, const char* key, STRLEN klen);
d1b91892
AD
1214
1215This routine checks to see what types of magic C<sv> has. If the mg_type
68dc0745
PP
1216field is an uppercase letter, then the mg_obj is copied to C<nsv>, but
1217the mg_type field is changed to be the lowercase letter.
a0d0e21e 1218
04343c6d
GS
1219=head2 Understanding the Magic of Tied Hashes and Arrays
1220
14befaf4
DM
1221Tied hashes and arrays are magical beasts of the C<PERL_MAGIC_tied>
1222magic type.
9edb2b46
GS
1223
1224WARNING: As of the 5.004 release, proper usage of the array and hash
1225access functions requires understanding a few caveats. Some
1226of these caveats are actually considered bugs in the API, to be fixed
1227in later releases, and are bracketed with [MAYCHANGE] below. If
1228you find yourself actually applying such information in this section, be
1229aware that the behavior may change in the future, umm, without warning.
04343c6d 1230
1526ead6 1231The perl tie function associates a variable with an object that implements
9a68f1db 1232the various GET, SET, etc methods. To perform the equivalent of the perl
1526ead6
AB
1233tie function from an XSUB, you must mimic this behaviour. The code below
1234carries out the necessary steps - firstly it creates a new hash, and then
1235creates a second hash which it blesses into the class which will implement
1236the tie methods. Lastly it ties the two hashes together, and returns a
1237reference to the new tied hash. Note that the code below does NOT call the
1238TIEHASH method in the MyTie class -
1239see L<Calling Perl Routines from within C Programs> for details on how
1240to do this.
1241
1242 SV*
1243 mytie()
1244 PREINIT:
1245 HV *hash;
1246 HV *stash;
1247 SV *tie;
1248 CODE:
1249 hash = newHV();
1250 tie = newRV_noinc((SV*)newHV());
da51bb9b 1251 stash = gv_stashpv("MyTie", GV_ADD);
1526ead6 1252 sv_bless(tie, stash);
899e16d0 1253 hv_magic(hash, (GV*)tie, PERL_MAGIC_tied);
1526ead6
AB
1254 RETVAL = newRV_noinc(hash);
1255 OUTPUT:
1256 RETVAL
1257
04343c6d
GS
1258The C<av_store> function, when given a tied array argument, merely
1259copies the magic of the array onto the value to be "stored", using
1260C<mg_copy>. It may also return NULL, indicating that the value did not
9edb2b46
GS
1261actually need to be stored in the array. [MAYCHANGE] After a call to
1262C<av_store> on a tied array, the caller will usually need to call
1263C<mg_set(val)> to actually invoke the perl level "STORE" method on the
1264TIEARRAY object. If C<av_store> did return NULL, a call to
1265C<SvREFCNT_dec(val)> will also be usually necessary to avoid a memory
1266leak. [/MAYCHANGE]
04343c6d
GS
1267
1268The previous paragraph is applicable verbatim to tied hash access using the
1269C<hv_store> and C<hv_store_ent> functions as well.
1270
1271C<av_fetch> and the corresponding hash functions C<hv_fetch> and
1272C<hv_fetch_ent> actually return an undefined mortal value whose magic
1273has been initialized using C<mg_copy>. Note the value so returned does not
9edb2b46
GS
1274need to be deallocated, as it is already mortal. [MAYCHANGE] But you will
1275need to call C<mg_get()> on the returned value in order to actually invoke
1276the perl level "FETCH" method on the underlying TIE object. Similarly,
04343c6d
GS
1277you may also call C<mg_set()> on the return value after possibly assigning
1278a suitable value to it using C<sv_setsv>, which will invoke the "STORE"
9edb2b46 1279method on the TIE object. [/MAYCHANGE]
04343c6d 1280
9edb2b46 1281[MAYCHANGE]
04343c6d
GS
1282In other words, the array or hash fetch/store functions don't really
1283fetch and store actual values in the case of tied arrays and hashes. They
1284merely call C<mg_copy> to attach magic to the values that were meant to be
1285"stored" or "fetched". Later calls to C<mg_get> and C<mg_set> actually
1286do the job of invoking the TIE methods on the underlying objects. Thus
9edb2b46 1287the magic mechanism currently implements a kind of lazy access to arrays
04343c6d
GS
1288and hashes.
1289
1290Currently (as of perl version 5.004), use of the hash and array access
1291functions requires the user to be aware of whether they are operating on
9edb2b46
GS
1292"normal" hashes and arrays, or on their tied variants. The API may be
1293changed to provide more transparent access to both tied and normal data
1294types in future versions.
1295[/MAYCHANGE]
04343c6d
GS
1296
1297You would do well to understand that the TIEARRAY and TIEHASH interfaces
1298are mere sugar to invoke some perl method calls while using the uniform hash
1299and array syntax. The use of this sugar imposes some overhead (typically
1300about two to four extra opcodes per FETCH/STORE operation, in addition to
1301the creation of all the mortal variables required to invoke the methods).
1302This overhead will be comparatively small if the TIE methods are themselves
1303substantial, but if they are only a few statements long, the overhead
1304will not be insignificant.
1305
d1c897a1
IZ
1306=head2 Localizing changes
1307
1308Perl has a very handy construction
1309
1310 {
1311 local $var = 2;
1312 ...
1313 }
1314
1315This construction is I<approximately> equivalent to
1316
1317 {
1318 my $oldvar = $var;
1319 $var = 2;
1320 ...
1321 $var = $oldvar;
1322 }
1323
1324The biggest difference is that the first construction would
1325reinstate the initial value of $var, irrespective of how control exits
9a68f1db 1326the block: C<goto>, C<return>, C<die>/C<eval>, etc. It is a little bit
d1c897a1
IZ
1327more efficient as well.
1328
1329There is a way to achieve a similar task from C via Perl API: create a
1330I<pseudo-block>, and arrange for some changes to be automatically
1331undone at the end of it, either explicit, or via a non-local exit (via
1332die()). A I<block>-like construct is created by a pair of
b687b08b
TC
1333C<ENTER>/C<LEAVE> macros (see L<perlcall/"Returning a Scalar">).
1334Such a construct may be created specially for some important localized
1335task, or an existing one (like boundaries of enclosing Perl
1336subroutine/block, or an existing pair for freeing TMPs) may be
1337used. (In the second case the overhead of additional localization must
1338be almost negligible.) Note that any XSUB is automatically enclosed in
1339an C<ENTER>/C<LEAVE> pair.
d1c897a1
IZ
1340
1341Inside such a I<pseudo-block> the following service is available:
1342
13a2d996 1343=over 4
d1c897a1
IZ
1344
1345=item C<SAVEINT(int i)>
1346
1347=item C<SAVEIV(IV i)>
1348
1349=item C<SAVEI32(I32 i)>
1350
1351=item C<SAVELONG(long i)>
1352
1353These macros arrange things to restore the value of integer variable
1354C<i> at the end of enclosing I<pseudo-block>.
1355
1356=item C<SAVESPTR(s)>
1357
1358=item C<SAVEPPTR(p)>
1359
1360These macros arrange things to restore the value of pointers C<s> and
1361C<p>. C<s> must be a pointer of a type which survives conversion to
1362C<SV*> and back, C<p> should be able to survive conversion to C<char*>
1363and back.
1364
1365=item C<SAVEFREESV(SV *sv)>
1366
1367The refcount of C<sv> would be decremented at the end of
26d9b02f
JH
1368I<pseudo-block>. This is similar to C<sv_2mortal> in that it is also a
1369mechanism for doing a delayed C<SvREFCNT_dec>. However, while C<sv_2mortal>
1370extends the lifetime of C<sv> until the beginning of the next statement,
1371C<SAVEFREESV> extends it until the end of the enclosing scope. These
1372lifetimes can be wildly different.
1373
1374Also compare C<SAVEMORTALIZESV>.
1375
1376=item C<SAVEMORTALIZESV(SV *sv)>
1377
1378Just like C<SAVEFREESV>, but mortalizes C<sv> at the end of the current
1379scope instead of decrementing its reference count. This usually has the
1380effect of keeping C<sv> alive until the statement that called the currently
1381live scope has finished executing.
d1c897a1
IZ
1382
1383=item C<SAVEFREEOP(OP *op)>
1384
1385The C<OP *> is op_free()ed at the end of I<pseudo-block>.
1386
1387=item C<SAVEFREEPV(p)>
1388
1389The chunk of memory which is pointed to by C<p> is Safefree()ed at the
1390end of I<pseudo-block>.
1391
1392=item C<SAVECLEARSV(SV *sv)>
1393
1394Clears a slot in the current scratchpad which corresponds to C<sv> at
1395the end of I<pseudo-block>.
1396
1397=item C<SAVEDELETE(HV *hv, char *key, I32 length)>
1398
1399The key C<key> of C<hv> is deleted at the end of I<pseudo-block>. The
1400string pointed to by C<key> is Safefree()ed. If one has a I<key> in
1401short-lived storage, the corresponding string may be reallocated like
1402this:
1403
9cde0e7f 1404 SAVEDELETE(PL_defstash, savepv(tmpbuf), strlen(tmpbuf));
d1c897a1 1405
c76ac1ee 1406=item C<SAVEDESTRUCTOR(DESTRUCTORFUNC_NOCONTEXT_t f, void *p)>
d1c897a1
IZ
1407
1408At the end of I<pseudo-block> the function C<f> is called with the
c76ac1ee
GS
1409only argument C<p>.
1410
1411=item C<SAVEDESTRUCTOR_X(DESTRUCTORFUNC_t f, void *p)>
1412
1413At the end of I<pseudo-block> the function C<f> is called with the
1414implicit context argument (if any), and C<p>.
d1c897a1
IZ
1415
1416=item C<SAVESTACK_POS()>
1417
1418The current offset on the Perl internal stack (cf. C<SP>) is restored
1419at the end of I<pseudo-block>.
1420
1421=back
1422
1423The following API list contains functions, thus one needs to
1424provide pointers to the modifiable data explicitly (either C pointers,
00aadd71 1425or Perlish C<GV *>s). Where the above macros take C<int>, a similar
d1c897a1
IZ
1426function takes C<int *>.
1427
13a2d996 1428=over 4
d1c897a1
IZ
1429
1430=item C<SV* save_scalar(GV *gv)>
1431
1432Equivalent to Perl code C<local $gv>.
1433
1434=item C<AV* save_ary(GV *gv)>
1435
1436=item C<HV* save_hash(GV *gv)>
1437
1438Similar to C<save_scalar>, but localize C<@gv> and C<%gv>.
1439
1440=item C<void save_item(SV *item)>
1441
1442Duplicates the current value of C<SV>, on the exit from the current
1443C<ENTER>/C<LEAVE> I<pseudo-block> will restore the value of C<SV>
038fcae3
SB
1444using the stored value. It doesn't handle magic. Use C<save_scalar> if
1445magic is affected.
d1c897a1
IZ
1446
1447=item C<void save_list(SV **sarg, I32 maxsarg)>
1448
1449A variant of C<save_item> which takes multiple arguments via an array
1450C<sarg> of C<SV*> of length C<maxsarg>.
1451
1452=item C<SV* save_svref(SV **sptr)>
1453
d1be9408 1454Similar to C<save_scalar>, but will reinstate an C<SV *>.
d1c897a1
IZ
1455
1456=item C<void save_aptr(AV **aptr)>
1457
1458=item C<void save_hptr(HV **hptr)>
1459
1460Similar to C<save_svref>, but localize C<AV *> and C<HV *>.
1461
1462=back
1463
1464The C<Alias> module implements localization of the basic types within the
1465I<caller's scope>. People who are interested in how to localize things in
1466the containing scope should take a look there too.
1467
0a753a76 1468=head1 Subroutines
a0d0e21e 1469
68dc0745 1470=head2 XSUBs and the Argument Stack
5f05dabc
PP
1471
1472The XSUB mechanism is a simple way for Perl programs to access C subroutines.
1473An XSUB routine will have a stack that contains the arguments from the Perl
1474program, and a way to map from the Perl data structures to a C equivalent.
1475
1476The stack arguments are accessible through the C<ST(n)> macro, which returns
1477the C<n>'th stack argument. Argument 0 is the first argument passed in the
1478Perl subroutine call. These arguments are C<SV*>, and can be used anywhere
1479an C<SV*> is used.
1480
1481Most of the time, output from the C routine can be handled through use of
1482the RETVAL and OUTPUT directives. However, there are some cases where the
1483argument stack is not already long enough to handle all the return values.
1484An example is the POSIX tzname() call, which takes no arguments, but returns
1485two, the local time zone's standard and summer time abbreviations.
1486
1487To handle this situation, the PPCODE directive is used and the stack is
1488extended using the macro:
1489
924508f0 1490 EXTEND(SP, num);
5f05dabc 1491
924508f0
GS
1492where C<SP> is the macro that represents the local copy of the stack pointer,
1493and C<num> is the number of elements the stack should be extended by.
5f05dabc 1494
00aadd71 1495Now that there is room on the stack, values can be pushed on it using C<PUSHs>
06f6df17 1496macro. The pushed values will often need to be "mortal" (See
d82b684c 1497L</Reference Counts and Mortality>):
5f05dabc 1498
00aadd71 1499 PUSHs(sv_2mortal(newSViv(an_integer)))
d82b684c
SH
1500 PUSHs(sv_2mortal(newSVuv(an_unsigned_integer)))
1501 PUSHs(sv_2mortal(newSVnv(a_double)))
00aadd71 1502 PUSHs(sv_2mortal(newSVpv("Some String",0)))
a9b0660e
KW
1503 /* Although the last example is better written as the more
1504 * efficient: */
a3179684 1505 PUSHs(newSVpvs_flags("Some String", SVs_TEMP))
5f05dabc
PP
1506
1507And now the Perl program calling C<tzname>, the two values will be assigned
1508as in:
1509
1510 ($standard_abbrev, $summer_abbrev) = POSIX::tzname;
1511
1512An alternate (and possibly simpler) method to pushing values on the stack is
00aadd71 1513to use the macro:
5f05dabc 1514
5f05dabc
PP
1515 XPUSHs(SV*)
1516
da8c5729 1517This macro automatically adjusts the stack for you, if needed. Thus, you
5f05dabc 1518do not need to call C<EXTEND> to extend the stack.
00aadd71
NIS
1519
1520Despite their suggestions in earlier versions of this document the macros
d82b684c
SH
1521C<(X)PUSH[iunp]> are I<not> suited to XSUBs which return multiple results.
1522For that, either stick to the C<(X)PUSHs> macros shown above, or use the new
1523C<m(X)PUSH[iunp]> macros instead; see L</Putting a C value on Perl stack>.
5f05dabc
PP
1524
1525For more information, consult L<perlxs> and L<perlxstut>.
1526
5b36e945
FC
1527=head2 Autoloading with XSUBs
1528
1529If an AUTOLOAD routine is an XSUB, as with Perl subroutines, Perl puts the
1530fully-qualified name of the autoloaded subroutine in the $AUTOLOAD variable
1531of the XSUB's package.
1532
1533But it also puts the same information in certain fields of the XSUB itself:
1534
1535 HV *stash = CvSTASH(cv);
1536 const char *subname = SvPVX(cv);
1537 STRLEN name_length = SvCUR(cv); /* in bytes */
1538 U32 is_utf8 = SvUTF8(cv);
1539
1540C<SvPVX(cv)> contains just the sub name itself, not including the package.
d8893903
FC
1541For an AUTOLOAD routine in UNIVERSAL or one of its superclasses,
1542C<CvSTASH(cv)> returns NULL during a method call on a nonexistent package.
5b36e945
FC
1543
1544B<Note>: Setting $AUTOLOAD stopped working in 5.6.1, which did not support
1545XS AUTOLOAD subs at all. Perl 5.8.0 introduced the use of fields in the
1546XSUB itself. Perl 5.16.0 restored the setting of $AUTOLOAD. If you need
1547to support 5.8-5.14, use the XSUB's fields.
1548
5f05dabc 1549=head2 Calling Perl Routines from within C Programs
a0d0e21e
LW
1550
1551There are four routines that can be used to call a Perl subroutine from
1552within a C program. These four are:
1553
954c1994
GS
1554 I32 call_sv(SV*, I32);
1555 I32 call_pv(const char*, I32);
1556 I32 call_method(const char*, I32);
1557 I32 call_argv(const char*, I32, register char**);
a0d0e21e 1558
954c1994 1559The routine most often used is C<call_sv>. The C<SV*> argument
d1b91892
AD
1560contains either the name of the Perl subroutine to be called, or a
1561reference to the subroutine. The second argument consists of flags
1562that control the context in which the subroutine is called, whether
1563or not the subroutine is being passed arguments, how errors should be
1564trapped, and how to treat return values.
a0d0e21e
LW
1565
1566All four routines return the number of arguments that the subroutine returned
1567on the Perl stack.
1568
9a68f1db 1569These routines used to be called C<perl_call_sv>, etc., before Perl v5.6.0,
954c1994
GS
1570but those names are now deprecated; macros of the same name are provided for
1571compatibility.
1572
1573When using any of these routines (except C<call_argv>), the programmer
d1b91892
AD
1574must manipulate the Perl stack. These include the following macros and
1575functions:
a0d0e21e
LW
1576
1577 dSP
924508f0 1578 SP
a0d0e21e
LW
1579 PUSHMARK()
1580 PUTBACK
1581 SPAGAIN
1582 ENTER
1583 SAVETMPS
1584 FREETMPS
1585 LEAVE
1586 XPUSH*()
cb1a09d0 1587 POP*()
a0d0e21e 1588
5f05dabc
PP
1589For a detailed description of calling conventions from C to Perl,
1590consult L<perlcall>.
a0d0e21e 1591
5f05dabc 1592=head2 Memory Allocation
a0d0e21e 1593
06f6df17
RGS
1594=head3 Allocation
1595
86058a2d
GS
1596All memory meant to be used with the Perl API functions should be manipulated
1597using the macros described in this section. The macros provide the necessary
1598transparency between differences in the actual malloc implementation that is
1599used within perl.
1600
1601It is suggested that you enable the version of malloc that is distributed
5f05dabc 1602with Perl. It keeps pools of various sizes of unallocated memory in
07fa94a1
JO
1603order to satisfy allocation requests more quickly. However, on some
1604platforms, it may cause spurious malloc or free errors.
d1b91892 1605
06f6df17
RGS
1606The following three macros are used to initially allocate memory :
1607
9f653bb5
SH
1608 Newx(pointer, number, type);
1609 Newxc(pointer, number, type, cast);
1610 Newxz(pointer, number, type);
d1b91892 1611
9f653bb5 1612The first argument C<pointer> should be the name of a variable that will
5f05dabc 1613point to the newly allocated memory.
d1b91892 1614
9f653bb5 1615The second and third arguments C<number> and C<type> specify how many of
d1b91892 1616the specified type of data structure should be allocated. The argument
9f653bb5 1617C<type> is passed to C<sizeof>. The final argument to C<Newxc>, C<cast>,
d1b91892
AD
1618should be used if the C<pointer> argument is different from the C<type>
1619argument.
1620
9f653bb5 1621Unlike the C<Newx> and C<Newxc> macros, the C<Newxz> macro calls C<memzero>
d1b91892
AD
1622to zero out all the newly allocated memory.
1623
06f6df17
RGS
1624=head3 Reallocation
1625
d1b91892
AD
1626 Renew(pointer, number, type);
1627 Renewc(pointer, number, type, cast);
1628 Safefree(pointer)
1629
1630These three macros are used to change a memory buffer size or to free a
1631piece of memory no longer needed. The arguments to C<Renew> and C<Renewc>
1632match those of C<New> and C<Newc> with the exception of not needing the
1633"magic cookie" argument.
1634
06f6df17
RGS
1635=head3 Moving
1636
d1b91892
AD
1637 Move(source, dest, number, type);
1638 Copy(source, dest, number, type);
1639 Zero(dest, number, type);
1640
1641These three macros are used to move, copy, or zero out previously allocated
1642memory. The C<source> and C<dest> arguments point to the source and
1643destination starting points. Perl will move, copy, or zero out C<number>
1644instances of the size of the C<type> data structure (using the C<sizeof>
1645function).
a0d0e21e 1646
5f05dabc 1647=head2 PerlIO
ce3d39e2 1648
da8c5729 1649The most recent development releases of Perl have been experimenting with
5f05dabc
PP
1650removing Perl's dependency on the "normal" standard I/O suite and allowing
1651other stdio implementations to be used. This involves creating a new
1652abstraction layer that then calls whichever implementation of stdio Perl
68dc0745 1653was compiled with. All XSUBs should now use the functions in the PerlIO
5f05dabc
PP
1654abstraction layer and not make any assumptions about what kind of stdio
1655is being used.
1656
1657For a complete description of the PerlIO abstraction, consult L<perlapio>.
1658
8ebc5c01 1659=head2 Putting a C value on Perl stack
ce3d39e2
IZ
1660
1661A lot of opcodes (this is an elementary operation in the internal perl
1662stack machine) put an SV* on the stack. However, as an optimization
1663the corresponding SV is (usually) not recreated each time. The opcodes
1664reuse specially assigned SVs (I<target>s) which are (as a corollary)
1665not constantly freed/created.
1666
0a753a76 1667Each of the targets is created only once (but see
ce3d39e2
IZ
1668L<Scratchpads and recursion> below), and when an opcode needs to put
1669an integer, a double, or a string on stack, it just sets the
1670corresponding parts of its I<target> and puts the I<target> on stack.
1671
1672The macro to put this target on stack is C<PUSHTARG>, and it is
1673directly used in some opcodes, as well as indirectly in zillions of
d82b684c 1674others, which use it via C<(X)PUSH[iunp]>.
ce3d39e2 1675
1bd1c0d5
SC
1676Because the target is reused, you must be careful when pushing multiple
1677values on the stack. The following code will not do what you think:
1678
1679 XPUSHi(10);
1680 XPUSHi(20);
1681
1682This translates as "set C<TARG> to 10, push a pointer to C<TARG> onto
1683the stack; set C<TARG> to 20, push a pointer to C<TARG> onto the stack".
1684At the end of the operation, the stack does not contain the values 10
1685and 20, but actually contains two pointers to C<TARG>, which we have set
d82b684c 1686to 20.
1bd1c0d5 1687
d82b684c
SH
1688If you need to push multiple different values then you should either use
1689the C<(X)PUSHs> macros, or else use the new C<m(X)PUSH[iunp]> macros,
1690none of which make use of C<TARG>. The C<(X)PUSHs> macros simply push an
1691SV* on the stack, which, as noted under L</XSUBs and the Argument Stack>,
1692will often need to be "mortal". The new C<m(X)PUSH[iunp]> macros make
1693this a little easier to achieve by creating a new mortal for you (via
1694C<(X)PUSHmortal>), pushing that onto the stack (extending it if necessary
1695in the case of the C<mXPUSH[iunp]> macros), and then setting its value.
1696Thus, instead of writing this to "fix" the example above:
1697
1698 XPUSHs(sv_2mortal(newSViv(10)))
1699 XPUSHs(sv_2mortal(newSViv(20)))
1700
1701you can simply write:
1702
1703 mXPUSHi(10)
1704 mXPUSHi(20)
1705
1706On a related note, if you do use C<(X)PUSH[iunp]>, then you're going to
1bd1c0d5 1707need a C<dTARG> in your variable declarations so that the C<*PUSH*>
d82b684c
SH
1708macros can make use of the local variable C<TARG>. See also C<dTARGET>
1709and C<dXSTARG>.
1bd1c0d5 1710
8ebc5c01 1711=head2 Scratchpads
ce3d39e2 1712
54310121 1713The question remains on when the SVs which are I<target>s for opcodes
ac036724 1714are created. The answer is that they are created when the current
1715unit--a subroutine or a file (for opcodes for statements outside of
1716subroutines)--is compiled. During this time a special anonymous Perl
1717array is created, which is called a scratchpad for the current unit.
ce3d39e2 1718
54310121 1719A scratchpad keeps SVs which are lexicals for the current unit and are
ce3d39e2
IZ
1720targets for opcodes. One can deduce that an SV lives on a scratchpad
1721by looking on its flags: lexicals have C<SVs_PADMY> set, and
1722I<target>s have C<SVs_PADTMP> set.
1723
54310121
PP
1724The correspondence between OPs and I<target>s is not 1-to-1. Different
1725OPs in the compile tree of the unit can use the same target, if this
ce3d39e2
IZ
1726would not conflict with the expected life of the temporary.
1727
2ae324a7 1728=head2 Scratchpads and recursion
ce3d39e2
IZ
1729
1730In fact it is not 100% true that a compiled unit contains a pointer to
1731the scratchpad AV. In fact it contains a pointer to an AV of
1732(initially) one element, and this element is the scratchpad AV. Why do
1733we need an extra level of indirection?
1734
9a68f1db 1735The answer is B<recursion>, and maybe B<threads>. Both
ce3d39e2
IZ
1736these can create several execution pointers going into the same
1737subroutine. For the subroutine-child not write over the temporaries
1738for the subroutine-parent (lifespan of which covers the call to the
1739child), the parent and the child should have different
1740scratchpads. (I<And> the lexicals should be separate anyway!)
1741
5f05dabc
PP
1742So each subroutine is born with an array of scratchpads (of length 1).
1743On each entry to the subroutine it is checked that the current
ce3d39e2
IZ
1744depth of the recursion is not more than the length of this array, and
1745if it is, new scratchpad is created and pushed into the array.
1746
1747The I<target>s on this scratchpad are C<undef>s, but they are already
1748marked with correct flags.
1749
0a753a76
PP
1750=head1 Compiled code
1751
1752=head2 Code tree
1753
1754Here we describe the internal form your code is converted to by
1755Perl. Start with a simple example:
1756
1757 $a = $b + $c;
1758
1759This is converted to a tree similar to this one:
1760
1761 assign-to
1762 / \
1763 + $a
1764 / \
1765 $b $c
1766
7b8d334a 1767(but slightly more complicated). This tree reflects the way Perl
0a753a76
PP
1768parsed your code, but has nothing to do with the execution order.
1769There is an additional "thread" going through the nodes of the tree
1770which shows the order of execution of the nodes. In our simplified
1771example above it looks like:
1772
1773 $b ---> $c ---> + ---> $a ---> assign-to
1774
1775But with the actual compile tree for C<$a = $b + $c> it is different:
1776some nodes I<optimized away>. As a corollary, though the actual tree
1777contains more nodes than our simplified example, the execution order
1778is the same as in our example.
1779
1780=head2 Examining the tree
1781
06f6df17
RGS
1782If you have your perl compiled for debugging (usually done with
1783C<-DDEBUGGING> on the C<Configure> command line), you may examine the
0a753a76
PP
1784compiled tree by specifying C<-Dx> on the Perl command line. The
1785output takes several lines per node, and for C<$b+$c> it looks like
1786this:
1787
1788 5 TYPE = add ===> 6
1789 TARG = 1
1790 FLAGS = (SCALAR,KIDS)
1791 {
1792 TYPE = null ===> (4)
1793 (was rv2sv)
1794 FLAGS = (SCALAR,KIDS)
1795 {
1796 3 TYPE = gvsv ===> 4
1797 FLAGS = (SCALAR)
1798 GV = main::b
1799 }
1800 }
1801 {
1802 TYPE = null ===> (5)
1803 (was rv2sv)
1804 FLAGS = (SCALAR,KIDS)
1805 {
1806 4 TYPE = gvsv ===> 5
1807 FLAGS = (SCALAR)
1808 GV = main::c
1809 }
1810 }
1811
1812This tree has 5 nodes (one per C<TYPE> specifier), only 3 of them are
1813not optimized away (one per number in the left column). The immediate
1814children of the given node correspond to C<{}> pairs on the same level
1815of indentation, thus this listing corresponds to the tree:
1816
1817 add
1818 / \
1819 null null
1820 | |
1821 gvsv gvsv
1822
1823The execution order is indicated by C<===E<gt>> marks, thus it is C<3
18244 5 6> (node C<6> is not included into above listing), i.e.,
1825C<gvsv gvsv add whatever>.
1826
9afa14e3
SC
1827Each of these nodes represents an op, a fundamental operation inside the
1828Perl core. The code which implements each operation can be found in the
1829F<pp*.c> files; the function which implements the op with type C<gvsv>
1830is C<pp_gvsv>, and so on. As the tree above shows, different ops have
1831different numbers of children: C<add> is a binary operator, as one would
1832expect, and so has two children. To accommodate the various different
1833numbers of children, there are various types of op data structure, and
1834they link together in different ways.
1835
1836The simplest type of op structure is C<OP>: this has no children. Unary
1837operators, C<UNOP>s, have one child, and this is pointed to by the
1838C<op_first> field. Binary operators (C<BINOP>s) have not only an
1839C<op_first> field but also an C<op_last> field. The most complex type of
1840op is a C<LISTOP>, which has any number of children. In this case, the
1841first child is pointed to by C<op_first> and the last child by
1842C<op_last>. The children in between can be found by iteratively
1843following the C<op_sibling> pointer from the first child to the last.
1844
1845There are also two other op types: a C<PMOP> holds a regular expression,
1846and has no children, and a C<LOOP> may or may not have children. If the
1847C<op_children> field is non-zero, it behaves like a C<LISTOP>. To
1848complicate matters, if a C<UNOP> is actually a C<null> op after
1849optimization (see L</Compile pass 2: context propagation>) it will still
1850have children in accordance with its former type.
1851
06f6df17
RGS
1852Another way to examine the tree is to use a compiler back-end module, such
1853as L<B::Concise>.
1854
0a753a76
PP
1855=head2 Compile pass 1: check routines
1856
8870b5c7
GS
1857The tree is created by the compiler while I<yacc> code feeds it
1858the constructions it recognizes. Since I<yacc> works bottom-up, so does
0a753a76
PP
1859the first pass of perl compilation.
1860
1861What makes this pass interesting for perl developers is that some
1862optimization may be performed on this pass. This is optimization by
8870b5c7 1863so-called "check routines". The correspondence between node names
0a753a76
PP
1864and corresponding check routines is described in F<opcode.pl> (do not
1865forget to run C<make regen_headers> if you modify this file).
1866
1867A check routine is called when the node is fully constructed except
7b8d334a 1868for the execution-order thread. Since at this time there are no
0a753a76
PP
1869back-links to the currently constructed node, one can do most any
1870operation to the top-level node, including freeing it and/or creating
1871new nodes above/below it.
1872
1873The check routine returns the node which should be inserted into the
1874tree (if the top-level node was not modified, check routine returns
1875its argument).
1876
1877By convention, check routines have names C<ck_*>. They are usually
1878called from C<new*OP> subroutines (or C<convert>) (which in turn are
1879called from F<perly.y>).
1880
1881=head2 Compile pass 1a: constant folding
1882
1883Immediately after the check routine is called the returned node is
1884checked for being compile-time executable. If it is (the value is
1885judged to be constant) it is immediately executed, and a I<constant>
1886node with the "return value" of the corresponding subtree is
1887substituted instead. The subtree is deleted.
1888
1889If constant folding was not performed, the execution-order thread is
1890created.
1891
1892=head2 Compile pass 2: context propagation
1893
1894When a context for a part of compile tree is known, it is propagated
a3cb178b 1895down through the tree. At this time the context can have 5 values
0a753a76
PP
1896(instead of 2 for runtime context): void, boolean, scalar, list, and
1897lvalue. In contrast with the pass 1 this pass is processed from top
1898to bottom: a node's context determines the context for its children.
1899
1900Additional context-dependent optimizations are performed at this time.
1901Since at this moment the compile tree contains back-references (via
1902"thread" pointers), nodes cannot be free()d now. To allow
1903optimized-away nodes at this stage, such nodes are null()ified instead
1904of free()ing (i.e. their type is changed to OP_NULL).
1905
1906=head2 Compile pass 3: peephole optimization
1907
1908After the compile tree for a subroutine (or for an C<eval> or a file)
1909is created, an additional pass over the code is performed. This pass
1910is neither top-down or bottom-up, but in the execution order (with
9ea12537
Z
1911additional complications for conditionals). Optimizations performed
1912at this stage are subject to the same restrictions as in the pass 2.
1913
1914Peephole optimizations are done by calling the function pointed to
1915by the global variable C<PL_peepp>. By default, C<PL_peepp> just
1916calls the function pointed to by the global variable C<PL_rpeepp>.
1917By default, that performs some basic op fixups and optimisations along
1918the execution-order op chain, and recursively calls C<PL_rpeepp> for
1919each side chain of ops (resulting from conditionals). Extensions may
1920provide additional optimisations or fixups, hooking into either the
1921per-subroutine or recursive stage, like this:
1922
1923 static peep_t prev_peepp;
1924 static void my_peep(pTHX_ OP *o)
1925 {
1926 /* custom per-subroutine optimisation goes here */
1927 prev_peepp(o);
1928 /* custom per-subroutine optimisation may also go here */
1929 }
1930 BOOT:
1931 prev_peepp = PL_peepp;
1932 PL_peepp = my_peep;
1933
1934 static peep_t prev_rpeepp;
1935 static void my_rpeep(pTHX_ OP *o)
1936 {
1937 OP *orig_o = o;
1938 for(; o; o = o->op_next) {
1939 /* custom per-op optimisation goes here */
1940 }
1941 prev_rpeepp(orig_o);
1942 }
1943 BOOT:
1944 prev_rpeepp = PL_rpeepp;
1945 PL_rpeepp = my_rpeep;
0a753a76 1946
1ba7f851
PJ
1947=head2 Pluggable runops
1948
1949The compile tree is executed in a runops function. There are two runops
1388f78e
RGS
1950functions, in F<run.c> and in F<dump.c>. C<Perl_runops_debug> is used
1951with DEBUGGING and C<Perl_runops_standard> is used otherwise. For fine
1952control over the execution of the compile tree it is possible to provide
1953your own runops function.
1ba7f851
PJ
1954
1955It's probably best to copy one of the existing runops functions and
1956change it to suit your needs. Then, in the BOOT section of your XS
1957file, add the line:
1958
1959 PL_runops = my_runops;
1960
1961This function should be as efficient as possible to keep your programs
1962running as fast as possible.
1963
fd85fad2
BM
1964=head2 Compile-time scope hooks
1965
1966As of perl 5.14 it is possible to hook into the compile-time lexical
1967scope mechanism using C<Perl_blockhook_register>. This is used like
1968this:
1969
1970 STATIC void my_start_hook(pTHX_ int full);
1971 STATIC BHK my_hooks;
1972
1973 BOOT:
a88d97bf 1974 BhkENTRY_set(&my_hooks, bhk_start, my_start_hook);
fd85fad2
BM
1975 Perl_blockhook_register(aTHX_ &my_hooks);
1976
1977This will arrange to have C<my_start_hook> called at the start of
1978compiling every lexical scope. The available hooks are:
1979
1980=over 4
1981
a88d97bf 1982=item C<void bhk_start(pTHX_ int full)>
fd85fad2
BM
1983
1984This is called just after starting a new lexical scope. Note that Perl
1985code like
1986
1987 if ($x) { ... }
1988
1989creates two scopes: the first starts at the C<(> and has C<full == 1>,
1990the second starts at the C<{> and has C<full == 0>. Both end at the
1991C<}>, so calls to C<start> and C<pre/post_end> will match. Anything
1992pushed onto the save stack by this hook will be popped just before the
1993scope ends (between the C<pre_> and C<post_end> hooks, in fact).
1994
a88d97bf 1995=item C<void bhk_pre_end(pTHX_ OP **o)>
fd85fad2
BM
1996
1997This is called at the end of a lexical scope, just before unwinding the
1998stack. I<o> is the root of the optree representing the scope; it is a
1999double pointer so you can replace the OP if you need to.
2000
a88d97bf 2001=item C<void bhk_post_end(pTHX_ OP **o)>
fd85fad2
BM
2002
2003This is called at the end of a lexical scope, just after unwinding the
2004stack. I<o> is as above. Note that it is possible for calls to C<pre_>
2005and C<post_end> to nest, if there is something on the save stack that
2006calls string eval.
2007
a88d97bf 2008=item C<void bhk_eval(pTHX_ OP *const o)>
fd85fad2
BM
2009
2010This is called just before starting to compile an C<eval STRING>, C<do
2011FILE>, C<require> or C<use>, after the eval has been set up. I<o> is the
2012OP that requested the eval, and will normally be an C<OP_ENTEREVAL>,
2013C<OP_DOFILE> or C<OP_REQUIRE>.
2014
2015=back
2016
2017Once you have your hook functions, you need a C<BHK> structure to put
2018them in. It's best to allocate it statically, since there is no way to
2019free it once it's registered. The function pointers should be inserted
2020into this structure using the C<BhkENTRY_set> macro, which will also set
2021flags indicating which entries are valid. If you do need to allocate
2022your C<BHK> dynamically for some reason, be sure to zero it before you
2023start.
2024
2025Once registered, there is no mechanism to switch these hooks off, so if
2026that is necessary you will need to do this yourself. An entry in C<%^H>
a3e07c87
BM
2027is probably the best way, so the effect is lexically scoped; however it
2028is also possible to use the C<BhkDISABLE> and C<BhkENABLE> macros to
2029temporarily switch entries on and off. You should also be aware that
2030generally speaking at least one scope will have opened before your
2031extension is loaded, so you will see some C<pre/post_end> pairs that
2032didn't have a matching C<start>.
fd85fad2 2033
9afa14e3
SC
2034=head1 Examining internal data structures with the C<dump> functions
2035
2036To aid debugging, the source file F<dump.c> contains a number of
2037functions which produce formatted output of internal data structures.
2038
2039The most commonly used of these functions is C<Perl_sv_dump>; it's used
2040for dumping SVs, AVs, HVs, and CVs. The C<Devel::Peek> module calls
2041C<sv_dump> to produce debugging output from Perl-space, so users of that
00aadd71 2042module should already be familiar with its format.
9afa14e3
SC
2043
2044C<Perl_op_dump> can be used to dump an C<OP> structure or any of its
210b36aa 2045derivatives, and produces output similar to C<perl -Dx>; in fact,
9afa14e3
SC
2046C<Perl_dump_eval> will dump the main root of the code being evaluated,
2047exactly like C<-Dx>.
2048
2049Other useful functions are C<Perl_dump_sub>, which turns a C<GV> into an
2050op tree, C<Perl_dump_packsubs> which calls C<Perl_dump_sub> on all the
2051subroutines in a package like so: (Thankfully, these are all xsubs, so
2052there is no op tree)
2053
2054 (gdb) print Perl_dump_packsubs(PL_defstash)
2055
2056 SUB attributes::bootstrap = (xsub 0x811fedc 0)
2057
2058 SUB UNIVERSAL::can = (xsub 0x811f50c 0)
2059
2060 SUB UNIVERSAL::isa = (xsub 0x811f304 0)
2061
2062 SUB UNIVERSAL::VERSION = (xsub 0x811f7ac 0)
2063
2064 SUB DynaLoader::boot_DynaLoader = (xsub 0x805b188 0)
2065
2066and C<Perl_dump_all>, which dumps all the subroutines in the stash and
2067the op tree of the main root.
2068
954c1994 2069=head1 How multiple interpreters and concurrency are supported
ee072b34 2070
ee072b34
GS
2071=head2 Background and PERL_IMPLICIT_CONTEXT
2072
2073The Perl interpreter can be regarded as a closed box: it has an API
2074for feeding it code or otherwise making it do things, but it also has
2075functions for its own use. This smells a lot like an object, and
2076there are ways for you to build Perl so that you can have multiple
acfe0abc
GS
2077interpreters, with one interpreter represented either as a C structure,
2078or inside a thread-specific structure. These structures contain all
2079the context, the state of that interpreter.
2080
7b52221d
RGS
2081One macro controls the major Perl build flavor: MULTIPLICITY. The
2082MULTIPLICITY build has a C structure that packages all the interpreter
2083state. With multiplicity-enabled perls, PERL_IMPLICIT_CONTEXT is also
2084normally defined, and enables the support for passing in a "hidden" first
2085argument that represents all three data structures. MULTIPLICITY makes
1a64a5e6 2086multi-threaded perls possible (with the ithreads threading model, related
7b52221d 2087to the macro USE_ITHREADS.)
54aff467 2088
27da23d5
JH
2089Two other "encapsulation" macros are the PERL_GLOBAL_STRUCT and
2090PERL_GLOBAL_STRUCT_PRIVATE (the latter turns on the former, and the
2091former turns on MULTIPLICITY.) The PERL_GLOBAL_STRUCT causes all the
2092internal variables of Perl to be wrapped inside a single global struct,
2093struct perl_vars, accessible as (globals) &PL_Vars or PL_VarsPtr or
2094the function Perl_GetVars(). The PERL_GLOBAL_STRUCT_PRIVATE goes
2095one step further, there is still a single struct (allocated in main()
2096either from heap or from stack) but there are no global data symbols
2097pointing to it. In either case the global struct should be initialised
2098as the very first thing in main() using Perl_init_global_struct() and
2099correspondingly tear it down after perl_free() using Perl_free_global_struct(),
2100please see F<miniperlmain.c> for usage details. You may also need
2101to use C<dVAR> in your coding to "declare the global variables"
2102when you are using them. dTHX does this for you automatically.
2103
bc028b6b
JH
2104To see whether you have non-const data you can use a BSD-compatible C<nm>:
2105
2106 nm libperl.a | grep -v ' [TURtr] '
2107
2108If this displays any C<D> or C<d> symbols, you have non-const data.
2109
27da23d5
JH
2110For backward compatibility reasons defining just PERL_GLOBAL_STRUCT
2111doesn't actually hide all symbols inside a big global struct: some
2112PerlIO_xxx vtables are left visible. The PERL_GLOBAL_STRUCT_PRIVATE
2113then hides everything (see how the PERLIO_FUNCS_DECL is used).
2114
54aff467 2115All this obviously requires a way for the Perl internal functions to be
acfe0abc 2116either subroutines taking some kind of structure as the first
ee072b34 2117argument, or subroutines taking nothing as the first argument. To
acfe0abc 2118enable these two very different ways of building the interpreter,
ee072b34
GS
2119the Perl source (as it does in so many other situations) makes heavy
2120use of macros and subroutine naming conventions.
2121
54aff467 2122First problem: deciding which functions will be public API functions and
00aadd71 2123which will be private. All functions whose names begin C<S_> are private
954c1994
GS
2124(think "S" for "secret" or "static"). All other functions begin with
2125"Perl_", but just because a function begins with "Perl_" does not mean it is
00aadd71
NIS
2126part of the API. (See L</Internal Functions>.) The easiest way to be B<sure> a
2127function is part of the API is to find its entry in L<perlapi>.
2128If it exists in L<perlapi>, it's part of the API. If it doesn't, and you
2129think it should be (i.e., you need it for your extension), send mail via
a422fd2d 2130L<perlbug> explaining why you think it should be.
ee072b34
GS
2131
2132Second problem: there must be a syntax so that the same subroutine
2133declarations and calls can pass a structure as their first argument,
2134or pass nothing. To solve this, the subroutines are named and
2135declared in a particular way. Here's a typical start of a static
2136function used within the Perl guts:
2137
2138 STATIC void
2139 S_incline(pTHX_ char *s)
2140
acfe0abc 2141STATIC becomes "static" in C, and may be #define'd to nothing in some
da8c5729 2142configurations in the future.
ee072b34 2143
651a3225
GS
2144A public function (i.e. part of the internal API, but not necessarily
2145sanctioned for use in extensions) begins like this:
ee072b34
GS
2146
2147 void
2307c6d0 2148 Perl_sv_setiv(pTHX_ SV* dsv, IV num)
ee072b34 2149
0147cd53 2150C<pTHX_> is one of a number of macros (in F<perl.h>) that hide the
ee072b34
GS
2151details of the interpreter's context. THX stands for "thread", "this",
2152or "thingy", as the case may be. (And no, George Lucas is not involved. :-)
2153The first character could be 'p' for a B<p>rototype, 'a' for B<a>rgument,
a7486cbb
JH
2154or 'd' for B<d>eclaration, so we have C<pTHX>, C<aTHX> and C<dTHX>, and
2155their variants.
ee072b34 2156
a7486cbb
JH
2157When Perl is built without options that set PERL_IMPLICIT_CONTEXT, there is no
2158first argument containing the interpreter's context. The trailing underscore
ee072b34
GS
2159in the pTHX_ macro indicates that the macro expansion needs a comma
2160after the context argument because other arguments follow it. If
2161PERL_IMPLICIT_CONTEXT is not defined, pTHX_ will be ignored, and the
54aff467
GS
2162subroutine is not prototyped to take the extra argument. The form of the
2163macro without the trailing underscore is used when there are no additional
ee072b34
GS
2164explicit arguments.
2165
54aff467 2166When a core function calls another, it must pass the context. This
2307c6d0 2167is normally hidden via macros. Consider C<sv_setiv>. It expands into
ee072b34
GS
2168something like this:
2169
2307c6d0
SB
2170 #ifdef PERL_IMPLICIT_CONTEXT
2171 #define sv_setiv(a,b) Perl_sv_setiv(aTHX_ a, b)
ee072b34 2172 /* can't do this for vararg functions, see below */
2307c6d0
SB
2173 #else
2174 #define sv_setiv Perl_sv_setiv
2175 #endif
ee072b34
GS
2176
2177This works well, and means that XS authors can gleefully write:
2178
2307c6d0 2179 sv_setiv(foo, bar);
ee072b34
GS
2180
2181and still have it work under all the modes Perl could have been
2182compiled with.
2183
ee072b34
GS
2184This doesn't work so cleanly for varargs functions, though, as macros
2185imply that the number of arguments is known in advance. Instead we
2186either need to spell them out fully, passing C<aTHX_> as the first
2187argument (the Perl core tends to do this with functions like
2188Perl_warner), or use a context-free version.
2189
2190The context-free version of Perl_warner is called
2191Perl_warner_nocontext, and does not take the extra argument. Instead
2192it does dTHX; to get the context from thread-local storage. We
2193C<#define warner Perl_warner_nocontext> so that extensions get source
2194compatibility at the expense of performance. (Passing an arg is
2195cheaper than grabbing it from thread-local storage.)
2196
acfe0abc 2197You can ignore [pad]THXx when browsing the Perl headers/sources.
ee072b34
GS
2198Those are strictly for use within the core. Extensions and embedders
2199need only be aware of [pad]THX.
2200
a7486cbb
JH
2201=head2 So what happened to dTHR?
2202
2203C<dTHR> was introduced in perl 5.005 to support the older thread model.
2204The older thread model now uses the C<THX> mechanism to pass context
2205pointers around, so C<dTHR> is not useful any more. Perl 5.6.0 and
2206later still have it for backward source compatibility, but it is defined
2207to be a no-op.
2208
ee072b34
GS
2209=head2 How do I use all this in extensions?
2210
2211When Perl is built with PERL_IMPLICIT_CONTEXT, extensions that call
2212any functions in the Perl API will need to pass the initial context
2213argument somehow. The kicker is that you will need to write it in
2214such a way that the extension still compiles when Perl hasn't been
2215built with PERL_IMPLICIT_CONTEXT enabled.
2216
2217There are three ways to do this. First, the easy but inefficient way,
2218which is also the default, in order to maintain source compatibility
0147cd53 2219with extensions: whenever F<XSUB.h> is #included, it redefines the aTHX
ee072b34
GS
2220and aTHX_ macros to call a function that will return the context.
2221Thus, something like:
2222
2307c6d0 2223 sv_setiv(sv, num);
ee072b34 2224
4375e838 2225in your extension will translate to this when PERL_IMPLICIT_CONTEXT is
54aff467 2226in effect:
ee072b34 2227
2307c6d0 2228 Perl_sv_setiv(Perl_get_context(), sv, num);
ee072b34 2229
54aff467 2230or to this otherwise:
ee072b34 2231
2307c6d0 2232 Perl_sv_setiv(sv, num);
ee072b34 2233
da8c5729 2234You don't have to do anything new in your extension to get this; since
2fa86c13 2235the Perl library provides Perl_get_context(), it will all just
ee072b34
GS
2236work.
2237
2238The second, more efficient way is to use the following template for
2239your Foo.xs:
2240
c52f9dcd
JH
2241 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2242 #include "EXTERN.h"
2243 #include "perl.h"
2244 #include "XSUB.h"
ee072b34 2245
fd061412 2246 STATIC void my_private_function(int arg1, int arg2);
ee072b34 2247
fd061412 2248 STATIC void
c52f9dcd
JH
2249 my_private_function(int arg1, int arg2)
2250 {
2251 dTHX; /* fetch context */
2252 ... call many Perl API functions ...
2253 }
ee072b34
GS
2254
2255 [... etc ...]
2256
c52f9dcd 2257 MODULE = Foo PACKAGE = Foo
ee072b34 2258
c52f9dcd 2259 /* typical XSUB */
ee072b34 2260
c52f9dcd
JH
2261 void
2262 my_xsub(arg)
2263 int arg
2264 CODE:
2265 my_private_function(arg, 10);
ee072b34
GS
2266
2267Note that the only two changes from the normal way of writing an
2268extension is the addition of a C<#define PERL_NO_GET_CONTEXT> before
2269including the Perl headers, followed by a C<dTHX;> declaration at
2270the start of every function that will call the Perl API. (You'll
2271know which functions need this, because the C compiler will complain
2272that there's an undeclared identifier in those functions.) No changes
2273are needed for the XSUBs themselves, because the XS() macro is
2274correctly defined to pass in the implicit context if needed.
2275
2276The third, even more efficient way is to ape how it is done within
2277the Perl guts:
2278
2279
c52f9dcd
JH
2280 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2281 #include "EXTERN.h"
2282 #include "perl.h"
2283 #include "XSUB.h"
ee072b34
GS
2284
2285 /* pTHX_ only needed for functions that call Perl API */
fd061412 2286 STATIC void my_private_function(pTHX_ int arg1, int arg2);
ee072b34 2287
fd061412 2288 STATIC void
c52f9dcd
JH
2289 my_private_function(pTHX_ int arg1, int arg2)
2290 {
2291 /* dTHX; not needed here, because THX is an argument */
2292 ... call Perl API functions ...
2293 }
ee072b34
GS
2294
2295 [... etc ...]
2296
c52f9dcd 2297 MODULE = Foo PACKAGE = Foo
ee072b34 2298
c52f9dcd 2299 /* typical XSUB */
ee072b34 2300
c52f9dcd
JH
2301 void
2302 my_xsub(arg)
2303 int arg
2304 CODE:
2305 my_private_function(aTHX_ arg, 10);
ee072b34
GS
2306
2307This implementation never has to fetch the context using a function
2308call, since it is always passed as an extra argument. Depending on
2309your needs for simplicity or efficiency, you may mix the previous
2310two approaches freely.
2311
651a3225
GS
2312Never add a comma after C<pTHX> yourself--always use the form of the
2313macro with the underscore for functions that take explicit arguments,
2314or the form without the argument for functions with no explicit arguments.
ee072b34 2315
27da23d5
JH
2316If one is compiling Perl with the C<-DPERL_GLOBAL_STRUCT> the C<dVAR>
2317definition is needed if the Perl global variables (see F<perlvars.h>
2318or F<globvar.sym>) are accessed in the function and C<dTHX> is not
2319used (the C<dTHX> includes the C<dVAR> if necessary). One notices
2320the need for C<dVAR> only with the said compile-time define, because
2321otherwise the Perl global variables are visible as-is.
2322
a7486cbb
JH
2323=head2 Should I do anything special if I call perl from multiple threads?
2324
2325If you create interpreters in one thread and then proceed to call them in
2326another, you need to make sure perl's own Thread Local Storage (TLS) slot is
2327initialized correctly in each of those threads.
2328
2329The C<perl_alloc> and C<perl_clone> API functions will automatically set
2330the TLS slot to the interpreter they created, so that there is no need to do
2331anything special if the interpreter is always accessed in the same thread that
2332created it, and that thread did not create or call any other interpreters
2333afterwards. If that is not the case, you have to set the TLS slot of the
2334thread before calling any functions in the Perl API on that particular
2335interpreter. This is done by calling the C<PERL_SET_CONTEXT> macro in that
2336thread as the first thing you do:
2337
2338 /* do this before doing anything else with some_perl */
2339 PERL_SET_CONTEXT(some_perl);
2340
2341 ... other Perl API calls on some_perl go here ...
2342
ee072b34
GS
2343=head2 Future Plans and PERL_IMPLICIT_SYS
2344
2345Just as PERL_IMPLICIT_CONTEXT provides a way to bundle up everything
2346that the interpreter knows about itself and pass it around, so too are
2347there plans to allow the interpreter to bundle up everything it knows
2348about the environment it's running on. This is enabled with the
7b52221d
RGS
2349PERL_IMPLICIT_SYS macro. Currently it only works with USE_ITHREADS on
2350Windows.
ee072b34
GS
2351
2352This allows the ability to provide an extra pointer (called the "host"
2353environment) for all the system calls. This makes it possible for
2354all the system stuff to maintain their own state, broken down into
2355seven C structures. These are thin wrappers around the usual system
0147cd53 2356calls (see F<win32/perllib.c>) for the default perl executable, but for a
ee072b34
GS
2357more ambitious host (like the one that would do fork() emulation) all
2358the extra work needed to pretend that different interpreters are
2359actually different "processes", would be done here.
2360
2361The Perl engine/interpreter and the host are orthogonal entities.
2362There could be one or more interpreters in a process, and one or
2363more "hosts", with free association between them.
2364
a422fd2d
SC
2365=head1 Internal Functions
2366
2367All of Perl's internal functions which will be exposed to the outside
06f6df17 2368world are prefixed by C<Perl_> so that they will not conflict with XS
a422fd2d
SC
2369functions or functions used in a program in which Perl is embedded.
2370Similarly, all global variables begin with C<PL_>. (By convention,
06f6df17 2371static functions start with C<S_>.)
a422fd2d 2372
0972ecdf
DM
2373Inside the Perl core (C<PERL_CORE> defined), you can get at the functions
2374either with or without the C<Perl_> prefix, thanks to a bunch of defines
2375that live in F<embed.h>. Note that extension code should I<not> set
2376C<PERL_CORE>; this exposes the full perl internals, and is likely to cause
2377breakage of the XS in each new perl release.
2378
2379The file F<embed.h> is generated automatically from
dc9b1d22
MHM
2380F<embed.pl> and F<embed.fnc>. F<embed.pl> also creates the prototyping
2381header files for the internal functions, generates the documentation
2382and a lot of other bits and pieces. It's important that when you add
2383a new function to the core or change an existing one, you change the
2384data in the table in F<embed.fnc> as well. Here's a sample entry from
2385that table:
a422fd2d
SC
2386
2387 Apd |SV** |av_fetch |AV* ar|I32 key|I32 lval
2388
2389The second column is the return type, the third column the name. Columns
2390after that are the arguments. The first column is a set of flags:
2391
2392=over 3
2393
2394=item A
2395
1aa6ea50
JC
2396This function is a part of the public API. All such functions should also
2397have 'd', very few do not.
a422fd2d
SC
2398
2399=item p
2400
1aa6ea50
JC
2401This function has a C<Perl_> prefix; i.e. it is defined as
2402C<Perl_av_fetch>.
a422fd2d
SC
2403
2404=item d
2405
2406This function has documentation using the C<apidoc> feature which we'll
1aa6ea50 2407look at in a second. Some functions have 'd' but not 'A'; docs are good.
a422fd2d
SC
2408
2409=back
2410
2411Other available flags are:
2412
2413=over 3
2414
2415=item s
2416
1aa6ea50
JC
2417This is a static function and is defined as C<STATIC S_whatever>, and
2418usually called within the sources as C<whatever(...)>.
a422fd2d
SC
2419
2420=item n
2421
da8c5729 2422This does not need an interpreter context, so the definition has no
1aa6ea50 2423C<pTHX>, and it follows that callers don't use C<aTHX>. (See
d3a43cd8 2424L</Background and PERL_IMPLICIT_CONTEXT>.)
a422fd2d
SC
2425
2426=item r
2427
2428This function never returns; C<croak>, C<exit> and friends.
2429
2430=item f
2431
2432This function takes a variable number of arguments, C<printf> style.
2433The argument list should end with C<...>, like this:
2434
2435 Afprd |void |croak |const char* pat|...
2436
a7486cbb 2437=item M
a422fd2d 2438
00aadd71 2439This function is part of the experimental development API, and may change
a422fd2d
SC
2440or disappear without notice.
2441
2442=item o
2443
2444This function should not have a compatibility macro to define, say,
2445C<Perl_parse> to C<parse>. It must be called as C<Perl_parse>.
2446
a422fd2d
SC
2447=item x
2448
2449This function isn't exported out of the Perl core.
2450
dc9b1d22
MHM
2451=item m
2452
2453This is implemented as a macro.
2454
2455=item X
2456
2457This function is explicitly exported.
2458
2459=item E
2460
2461This function is visible to extensions included in the Perl core.
2462
2463=item b
2464
2465Binary backward compatibility; this function is a macro but also has
2466a C<Perl_> implementation (which is exported).
2467
1aa6ea50
JC
2468=item others
2469
2470See the comments at the top of C<embed.fnc> for others.
2471
a422fd2d
SC
2472=back
2473
dc9b1d22
MHM
2474If you edit F<embed.pl> or F<embed.fnc>, you will need to run
2475C<make regen_headers> to force a rebuild of F<embed.h> and other
2476auto-generated files.
a422fd2d 2477
6b4667fc 2478=head2 Formatted Printing of IVs, UVs, and NVs
9dd9db0b 2479
6b4667fc
A
2480If you are printing IVs, UVs, or NVS instead of the stdio(3) style
2481formatting codes like C<%d>, C<%ld>, C<%f>, you should use the
2482following macros for portability
9dd9db0b 2483
c52f9dcd
JH
2484 IVdf IV in decimal
2485 UVuf UV in decimal
2486 UVof UV in octal
2487 UVxf UV in hexadecimal
2488 NVef NV %e-like
2489 NVff NV %f-like
2490 NVgf NV %g-like
9dd9db0b 2491
6b4667fc
A
2492These will take care of 64-bit integers and long doubles.
2493For example:
2494
c52f9dcd 2495 printf("IV is %"IVdf"\n", iv);
6b4667fc
A
2496
2497The IVdf will expand to whatever is the correct format for the IVs.
9dd9db0b 2498
8908e76d
JH
2499If you are printing addresses of pointers, use UVxf combined
2500with PTR2UV(), do not use %lx or %p.
2501
2502=head2 Pointer-To-Integer and Integer-To-Pointer
2503
2504Because pointer size does not necessarily equal integer size,
2505use the follow macros to do it right.
2506
c52f9dcd
JH
2507 PTR2UV(pointer)
2508 PTR2IV(pointer)
2509 PTR2NV(pointer)
2510 INT2PTR(pointertotype, integer)
8908e76d
JH
2511
2512For example:
2513
c52f9dcd
JH
2514 IV iv = ...;
2515 SV *sv = INT2PTR(SV*, iv);
8908e76d
JH
2516
2517and
2518
c52f9dcd
JH
2519 AV *av = ...;
2520 UV uv = PTR2UV(av);
8908e76d 2521
0ca3a874
MHM
2522=head2 Exception Handling
2523
9b5c3821
MHM
2524There are a couple of macros to do very basic exception handling in XS
2525modules. You have to define C<NO_XSLOCKS> before including F<XSUB.h> to
2526be able to use these macros:
2527
2528 #define NO_XSLOCKS
2529 #include "XSUB.h"
2530
2531You can use these macros if you call code that may croak, but you need
2532to do some cleanup before giving control back to Perl. For example:
0ca3a874 2533
d7f8936a 2534 dXCPT; /* set up necessary variables */
0ca3a874
MHM
2535
2536 XCPT_TRY_START {
2537 code_that_may_croak();
2538 } XCPT_TRY_END
2539
2540 XCPT_CATCH
2541 {
2542 /* do cleanup here */
2543 XCPT_RETHROW;
2544 }
2545
2546Note that you always have to rethrow an exception that has been
2547caught. Using these macros, it is not possible to just catch the
2548exception and ignore it. If you have to ignore the exception, you
2549have to use the C<call_*> function.
2550
2551The advantage of using the above macros is that you don't have
2552to setup an extra function for C<call_*>, and that using these
2553macros is faster than using C<call_*>.
2554
a422fd2d
SC
2555=head2 Source Documentation
2556
2557There's an effort going on to document the internal functions and
2558automatically produce reference manuals from them - L<perlapi> is one
2559such manual which details all the functions which are available to XS
2560writers. L<perlintern> is the autogenerated manual for the functions
2561which are not part of the API and are supposedly for internal use only.
2562
2563Source documentation is created by putting POD comments into the C
2564source, like this:
2565
2566 /*
2567 =for apidoc sv_setiv
2568
2569 Copies an integer into the given SV. Does not handle 'set' magic. See
2570 C<sv_setiv_mg>.
2571
2572 =cut
2573 */
2574
2575Please try and supply some documentation if you add functions to the
2576Perl core.
2577
0d098d33
MHM
2578=head2 Backwards compatibility
2579
2580The Perl API changes over time. New functions are added or the interfaces
2581of existing functions are changed. The C<Devel::PPPort> module tries to
2582provide compatibility code for some of these changes, so XS writers don't
2583have to code it themselves when supporting multiple versions of Perl.
2584
2585C<Devel::PPPort> generates a C header file F<ppport.h> that can also
2586be run as a Perl script. To generate F<ppport.h>, run:
2587
2588 perl -MDevel::PPPort -eDevel::PPPort::WriteFile
2589
2590Besides checking existing XS code, the script can also be used to retrieve
2591compatibility information for various API calls using the C<--api-info>
2592command line switch. For example:
2593
2594 % perl ppport.h --api-info=sv_magicext
2595
2596For details, see C<perldoc ppport.h>.
2597
a422fd2d
SC
2598=head1 Unicode Support
2599
2600Perl 5.6.0 introduced Unicode support. It's important for porters and XS
2601writers to understand this support and make sure that the code they
2602write does not corrupt Unicode data.
2603
2604=head2 What B<is> Unicode, anyway?
2605
2606In the olden, less enlightened times, we all used to use ASCII. Most of
2607us did, anyway. The big problem with ASCII is that it's American. Well,
2608no, that's not actually the problem; the problem is that it's not
2609particularly useful for people who don't use the Roman alphabet. What
2610used to happen was that particular languages would stick their own
2611alphabet in the upper range of the sequence, between 128 and 255. Of
2612course, we then ended up with plenty of variants that weren't quite
2613ASCII, and the whole point of it being a standard was lost.
2614
2615Worse still, if you've got a language like Chinese or
2616Japanese that has hundreds or thousands of characters, then you really
2617can't fit them into a mere 256, so they had to forget about ASCII
2618altogether, and build their own systems using pairs of numbers to refer
2619to one character.
2620
2621To fix this, some people formed Unicode, Inc. and
2622produced a new character set containing all the characters you can
2623possibly think of and more. There are several ways of representing these
1e54db1a 2624characters, and the one Perl uses is called UTF-8. UTF-8 uses
2575c402
JW
2625a variable number of bytes to represent a character. You can learn more
2626about Unicode and Perl's Unicode model in L<perlunicode>.
a422fd2d 2627
1e54db1a 2628=head2 How can I recognise a UTF-8 string?
a422fd2d 2629
1e54db1a
JH
2630You can't. This is because UTF-8 data is stored in bytes just like
2631non-UTF-8 data. The Unicode character 200, (C<0xC8> for you hex types)
a422fd2d
SC
2632capital E with a grave accent, is represented by the two bytes
2633C<v196.172>. Unfortunately, the non-Unicode string C<chr(196).chr(172)>
2634has that byte sequence as well. So you can't tell just by looking - this
2635is what makes Unicode input an interesting problem.
2636
2575c402
JW
2637In general, you either have to know what you're dealing with, or you
2638have to guess. The API function C<is_utf8_string> can help; it'll tell
2639you if a string contains only valid UTF-8 characters. However, it can't
49f4c4e4
KW
2640do the work for you. On a character-by-character basis,
2641C<is_utf8_char_buf>
2575c402 2642will tell you whether the current character in a string is valid UTF-8.
a422fd2d 2643
1e54db1a 2644=head2 How does UTF-8 represent Unicode characters?
a422fd2d 2645
1e54db1a 2646As mentioned above, UTF-8 uses a variable number of bytes to store a
2575c402
JW
2647character. Characters with values 0...127 are stored in one byte, just
2648like good ol' ASCII. Character 128 is stored as C<v194.128>; this
a31a806a 2649continues up to character 191, which is C<v194.191>. Now we've run out of
a422fd2d
SC
2650bits (191 is binary C<10111111>) so we move on; 192 is C<v195.128>. And
2651so it goes on, moving to three bytes at character 2048.
2652
1e54db1a 2653Assuming you know you're dealing with a UTF-8 string, you can find out
a422fd2d
SC
2654how long the first character in it is with the C<UTF8SKIP> macro:
2655
2656 char *utf = "\305\233\340\240\201";
2657 I32 len;
2658
2659 len = UTF8SKIP(utf); /* len is 2 here */
2660 utf += len;
2661 len = UTF8SKIP(utf); /* len is 3 here */
2662
1e54db1a 2663Another way to skip over characters in a UTF-8 string is to use
a422fd2d
SC
2664C<utf8_hop>, which takes a string and a number of characters to skip
2665over. You're on your own about bounds checking, though, so don't use it
2666lightly.
2667
1e54db1a 2668All bytes in a multi-byte UTF-8 character will have the high bit set,
3a2263fe
RGS
2669so you can test if you need to do something special with this
2670character like this (the UTF8_IS_INVARIANT() is a macro that tests
2671whether the byte can be encoded as a single byte even in UTF-8):
a422fd2d 2672
3a2263fe 2673 U8 *utf;
4b88fb76 2674 U8 *utf_end; /* 1 beyond buffer pointed to by utf */
3a2263fe 2675 UV uv; /* Note: a UV, not a U8, not a char */
95701e00 2676 STRLEN len; /* length of character in bytes */
a422fd2d 2677
3a2263fe 2678 if (!UTF8_IS_INVARIANT(*utf))
1e54db1a 2679 /* Must treat this as UTF-8 */
4b88fb76 2680 uv = utf8_to_uvchr_buf(utf, utf_end, &len);
a422fd2d
SC
2681 else
2682 /* OK to treat this character as a byte */
2683 uv = *utf;
2684
4b88fb76 2685You can also see in that example that we use C<utf8_to_uvchr_buf> to get the
95701e00 2686value of the character; the inverse function C<uvchr_to_utf8> is available
1e54db1a 2687for putting a UV into UTF-8:
a422fd2d 2688
3a2263fe 2689 if (!UTF8_IS_INVARIANT(uv))
a422fd2d 2690 /* Must treat this as UTF8 */
95701e00 2691 utf8 = uvchr_to_utf8(utf8, uv);
a422fd2d
SC
2692 else
2693 /* OK to treat this character as a byte */
2694 *utf8++ = uv;
2695
2696You B<must> convert characters to UVs using the above functions if
1e54db1a
JH
2697you're ever in a situation where you have to match UTF-8 and non-UTF-8
2698characters. You may not skip over UTF-8 characters in this case. If you
2699do this, you'll lose the ability to match hi-bit non-UTF-8 characters;
2700for instance, if your UTF-8 string contains C<v196.172>, and you skip
2701that character, you can never match a C<chr(200)> in a non-UTF-8 string.
a422fd2d
SC
2702So don't do that!
2703
1e54db1a 2704=head2 How does Perl store UTF-8 strings?
a422fd2d
SC
2705
2706Currently, Perl deals with Unicode strings and non-Unicode strings
2575c402
JW
2707slightly differently. A flag in the SV, C<SVf_UTF8>, indicates that the
2708string is internally encoded as UTF-8. Without it, the byte value is the
2709codepoint number and vice versa (in other words, the string is encoded
e1b711da
KW
2710as iso-8859-1, but C<use feature 'unicode_strings'> is needed to get iso-8859-1
2711semantics). You can check and manipulate this flag with the
2575c402 2712following macros:
a422fd2d
SC
2713
2714 SvUTF8(sv)
2715 SvUTF8_on(sv)
2716 SvUTF8_off(sv)
2717
2718This flag has an important effect on Perl's treatment of the string: if
2719Unicode data is not properly distinguished, regular expressions,
2720C<length>, C<substr> and other string handling operations will have
2721undesirable results.
2722
2723The problem comes when you have, for instance, a string that isn't
2575c402 2724flagged as UTF-8, and contains a byte sequence that could be UTF-8 -
1e54db1a 2725especially when combining non-UTF-8 and UTF-8 strings.
a422fd2d
SC
2726
2727Never forget that the C<SVf_UTF8> flag is separate to the PV value; you
2728need be sure you don't accidentally knock it off while you're
2729manipulating SVs. More specifically, you cannot expect to do this:
2730
2731 SV *sv;
2732 SV *nsv;
2733 STRLEN len;
2734 char *p;
2735
2736 p = SvPV(sv, len);
2737 frobnicate(p);
2738 nsv = newSVpvn(p, len);
2739
2740The C<char*> string does not tell you the whole story, and you can't
2741copy or reconstruct an SV just by copying the string value. Check if the
2575c402 2742old SV has the UTF8 flag set, and act accordingly:
a422fd2d
SC
2743
2744 p = SvPV(sv, len);
2745 frobnicate(p);
2746 nsv = newSVpvn(p, len);
2747 if (SvUTF8(sv))
2748 SvUTF8_on(nsv);
2749
2750In fact, your C<frobnicate> function should be made aware of whether or
1e54db1a 2751not it's dealing with UTF-8 data, so that it can handle the string
a422fd2d
SC
2752appropriately.
2753
3a2263fe 2754Since just passing an SV to an XS function and copying the data of
2575c402 2755the SV is not enough to copy the UTF8 flags, even less right is just
3a2263fe
RGS
2756passing a C<char *> to an XS function.
2757
1e54db1a 2758=head2 How do I convert a string to UTF-8?
a422fd2d 2759
2575c402
JW
2760If you're mixing UTF-8 and non-UTF-8 strings, it is necessary to upgrade
2761one of the strings to UTF-8. If you've got an SV, the easiest way to do
2762this is:
a422fd2d
SC
2763
2764 sv_utf8_upgrade(sv);
2765
2766However, you must not do this, for example:
2767
2768 if (!SvUTF8(left))
2769 sv_utf8_upgrade(left);
2770
2771If you do this in a binary operator, you will actually change one of the
b1866b2d 2772strings that came into the operator, and, while it shouldn't be noticeable
2575c402 2773by the end user, it can cause problems in deficient code.
a422fd2d 2774
1e54db1a 2775Instead, C<bytes_to_utf8> will give you a UTF-8-encoded B<copy> of its
a422fd2d 2776string argument. This is useful for having the data available for
b1866b2d 2777comparisons and so on, without harming the original SV. There's also
a422fd2d
SC
2778C<utf8_to_bytes> to go the other way, but naturally, this will fail if
2779the string contains any characters above 255 that can't be represented
2780in a single byte.
2781
2782=head2 Is there anything else I need to know?
2783
2784Not really. Just remember these things:
2785
2786=over 3
2787
2788=item *
2789
1e54db1a 2790There's no way to tell if a string is UTF-8 or not. You can tell if an SV
da8c5729 2791is UTF-8 by looking at its C<SvUTF8> flag. Don't forget to set the flag if
1e54db1a 2792something should be UTF-8. Treat the flag as part of the PV, even though
a422fd2d
SC
2793it's not - if you pass on the PV to somewhere, pass on the flag too.
2794
2795=item *
2796
4b88fb76 2797If a string is UTF-8, B<always> use C<utf8_to_uvchr_buf> to get at the value,
3a2263fe 2798unless C<UTF8_IS_INVARIANT(*s)> in which case you can use C<*s>.
a422fd2d
SC
2799
2800=item *
2801
1e54db1a 2802When writing a character C<uv> to a UTF-8 string, B<always> use
95701e00 2803C<uvchr_to_utf8>, unless C<UTF8_IS_INVARIANT(uv))> in which case
3a2263fe 2804you can use C<*s = uv>.
a422fd2d
SC
2805
2806=item *
2807
1e54db1a 2808Mixing UTF-8 and non-UTF-8 strings is tricky. Use C<bytes_to_utf8> to get
2bbc8d55 2809a new string which is UTF-8 encoded, and then combine them.
a422fd2d
SC
2810
2811=back
2812
53e06cf0
SC
2813=head1 Custom Operators
2814
9a68f1db 2815Custom operator support is a new experimental feature that allows you to
53e06cf0
SC
2816define your own ops. This is primarily to allow the building of
2817interpreters for other languages in the Perl core, but it also allows
2818optimizations through the creation of "macro-ops" (ops which perform the
2819functions of multiple ops which are usually executed together, such as
1aa6ea50 2820C<gvsv, gvsv, add>.)
53e06cf0 2821
b455bf3f 2822This feature is implemented as a new op type, C<OP_CUSTOM>. The Perl
53e06cf0
SC
2823core does not "know" anything special about this op type, and so it will
2824not be involved in any optimizations. This also means that you can
2825define your custom ops to be any op structure - unary, binary, list and
2826so on - you like.
2827
2828It's important to know what custom operators won't do for you. They
2829won't let you add new syntax to Perl, directly. They won't even let you
2830add new keywords, directly. In fact, they won't change the way Perl
2831compiles a program at all. You have to do those changes yourself, after
2832Perl has compiled the program. You do this either by manipulating the op
2833tree using a C<CHECK> block and the C<B::Generate> module, or by adding
2834a custom peephole optimizer with the C<optimize> module.
2835
2836When you do this, you replace ordinary Perl ops with custom ops by
2837creating ops with the type C<OP_CUSTOM> and the C<pp_addr> of your own
2838PP function. This should be defined in XS code, and should look like
2839the PP ops in C<pp_*.c>. You are responsible for ensuring that your op
2840takes the appropriate number of values from the stack, and you are
2841responsible for adding stack marks if necessary.
2842
2843You should also "register" your op with the Perl interpreter so that it
2844can produce sensible error and warning messages. Since it is possible to
2845have multiple custom ops within the one "logical" op type C<OP_CUSTOM>,
9733086d
BM
2846Perl uses the value of C<< o->op_ppaddr >> to determine which custom op
2847it is dealing with. You should create an C<XOP> structure for each
2848ppaddr you use, set the properties of the custom op with
2849C<XopENTRY_set>, and register the structure against the ppaddr using
2850C<Perl_custom_op_register>. A trivial example might look like:
2851
2852 static XOP my_xop;
2853 static OP *my_pp(pTHX);
2854
2855 BOOT:
2856 XopENTRY_set(&my_xop, xop_name, "myxop");
2857 XopENTRY_set(&my_xop, xop_desc, "Useless custom op");
2858 Perl_custom_op_register(aTHX_ my_pp, &my_xop);
2859
2860The available fields in the structure are:
2861
2862=over 4
2863
2864=item xop_name
2865
2866A short name for your op. This will be included in some error messages,
2867and will also be returned as C<< $op->name >> by the L<B|B> module, so
2868it will appear in the output of module like L<B::Concise|B::Concise>.
2869
2870=item xop_desc
2871
2872A short description of the function of the op.
2873
2874=item xop_class
2875
2876Which of the various C<*OP> structures this op uses. This should be one of
2877the C<OA_*> constants from F<op.h>, namely
2878
2879=over 4
2880
2881=item OA_BASEOP
2882
2883=item OA_UNOP
2884
2885=item OA_BINOP
2886
2887=item OA_LOGOP
2888
2889=item OA_LISTOP
2890
2891=item OA_PMOP
2892
2893=item OA_SVOP
2894
2895=item OA_PADOP
2896
2897=item OA_PVOP_OR_SVOP
2898
2899This should be interpreted as 'C<PVOP>' only. The C<_OR_SVOP> is because
2900the only core C<PVOP>, C<OP_TRANS>, can sometimes be a C<SVOP> instead.
2901
2902=item OA_LOOP
2903
2904=item OA_COP
2905
2906=back
2907
2908The other C<OA_*> constants should not be used.
2909
2910=item xop_peep
2911
2912This member is of type C<Perl_cpeep_t>, which expands to C<void
2913(*Perl_cpeep_t)(aTHX_ OP *o, OP *oldop)>. If it is set, this function
2914will be called from C<Perl_rpeep> when ops of this type are encountered
2915by the peephole optimizer. I<o> is the OP that needs optimizing;
2916I<oldop> is the previous OP optimized, whose C<op_next> points to I<o>.
2917
2918=back
53e06cf0 2919
e7d4c058 2920C<B::Generate> directly supports the creation of custom ops by name.
53e06cf0 2921
954c1994 2922=head1 AUTHORS
e89caa19 2923
954c1994 2924Until May 1997, this document was maintained by Jeff Okamoto
9b5bb84f
SB
2925E<lt>okamoto@corp.hp.comE<gt>. It is now maintained as part of Perl
2926itself by the Perl 5 Porters E<lt>perl5-porters@perl.orgE<gt>.
cb1a09d0 2927
954c1994
GS
2928With lots of help and suggestions from Dean Roehrich, Malcolm Beattie,
2929Andreas Koenig, Paul Hudson, Ilya Zakharevich, Paul Marquess, Neil
2930Bowers, Matthew Green, Tim Bunce, Spider Boardman, Ulrich Pfeifer,
2931Stephen McCamant, and Gurusamy Sarathy.
cb1a09d0 2932
954c1994 2933=head1 SEE ALSO
cb1a09d0 2934
ba555bf5 2935L<perlapi>, L<perlintern>, L<perlxs>, L<perlembed>