3 perlguts - Introduction to the Perl API
7 This document attempts to describe how to use the Perl API, as well as
8 to provide some info on the basic workings of the Perl core. It is far
9 from complete and probably contains many errors. Please refer any
10 questions or comments to the author below.
16 Perl has three typedefs that handle Perl's three main data types:
22 Each typedef has specific routines that manipulate the various data types.
24 =head2 What is an "IV"?
26 Perl uses a special typedef IV which is a simple signed integer type that is
27 guaranteed to be large enough to hold a pointer (as well as an integer).
28 Additionally, there is the UV, which is simply an unsigned IV.
30 Perl also uses two special typedefs, I32 and I16, which will always be at
31 least 32-bits and 16-bits long, respectively. (Again, there are U32 and U16,
32 as well.) They will usually be exactly 32 and 16 bits long, but on Crays
33 they will both be 64 bits.
35 =head2 Working with SVs
37 An SV can be created and loaded with one command. There are five types of
38 values that can be loaded: an integer value (IV), an unsigned integer
39 value (UV), a double (NV), a string (PV), and another scalar (SV).
40 ("PV" stands for "Pointer Value". You might think that it is misnamed
41 because it is described as pointing only to strings. However, it is
42 possible to have it point to other things For example, it could point
43 to an array of UVs. But,
44 using it for non-strings requires care, as the underlying assumption of
45 much of the internals is that PVs are just for strings. Often, for
46 example, a trailing NUL is tacked on automatically. The non-string use
47 is documented only in this paragraph.)
49 The seven routines are:
54 SV* newSVpv(const char*, STRLEN);
55 SV* newSVpvn(const char*, STRLEN);
56 SV* newSVpvf(const char*, ...);
59 C<STRLEN> is an integer type (Size_t, usually defined as size_t in
60 F<config.h>) guaranteed to be large enough to represent the size of
61 any string that perl can handle.
63 In the unlikely case of a SV requiring more complex initialization, you
64 can create an empty SV with newSV(len). If C<len> is 0 an empty SV of
65 type NULL is returned, else an SV of type PV is returned with len + 1 (for
66 the NUL) bytes of storage allocated, accessible via SvPVX. In both cases
67 the SV has the undef value.
69 SV *sv = newSV(0); /* no storage allocated */
70 SV *sv = newSV(10); /* 10 (+1) bytes of uninitialised storage
73 To change the value of an I<already-existing> SV, there are eight routines:
75 void sv_setiv(SV*, IV);
76 void sv_setuv(SV*, UV);
77 void sv_setnv(SV*, double);
78 void sv_setpv(SV*, const char*);
79 void sv_setpvn(SV*, const char*, STRLEN)
80 void sv_setpvf(SV*, const char*, ...);
81 void sv_vsetpvfn(SV*, const char*, STRLEN, va_list *,
83 void sv_setsv(SV*, SV*);
85 Notice that you can choose to specify the length of the string to be
86 assigned by using C<sv_setpvn>, C<newSVpvn>, or C<newSVpv>, or you may
87 allow Perl to calculate the length by using C<sv_setpv> or by specifying
88 0 as the second argument to C<newSVpv>. Be warned, though, that Perl will
89 determine the string's length by using C<strlen>, which depends on the
90 string terminating with a NUL character, and not otherwise containing
93 The arguments of C<sv_setpvf> are processed like C<sprintf>, and the
94 formatted output becomes the value.
96 C<sv_vsetpvfn> is an analogue of C<vsprintf>, but it allows you to specify
97 either a pointer to a variable argument list or the address and length of
98 an array of SVs. The last argument points to a boolean; on return, if that
99 boolean is true, then locale-specific information has been used to format
100 the string, and the string's contents are therefore untrustworthy (see
101 L<perlsec>). This pointer may be NULL if that information is not
102 important. Note that this function requires you to specify the length of
105 The C<sv_set*()> functions are not generic enough to operate on values
106 that have "magic". See L<Magic Virtual Tables> later in this document.
108 All SVs that contain strings should be terminated with a NUL character.
109 If it is not NUL-terminated there is a risk of
110 core dumps and corruptions from code which passes the string to C
111 functions or system calls which expect a NUL-terminated string.
112 Perl's own functions typically add a trailing NUL for this reason.
113 Nevertheless, you should be very careful when you pass a string stored
114 in an SV to a C function or system call.
116 To access the actual value that an SV points to, you can use the macros:
121 SvPV(SV*, STRLEN len)
124 which will automatically coerce the actual scalar type into an IV, UV, double,
127 In the C<SvPV> macro, the length of the string returned is placed into the
128 variable C<len> (this is a macro, so you do I<not> use C<&len>). If you do
129 not care what the length of the data is, use the C<SvPV_nolen> macro.
130 Historically the C<SvPV> macro with the global variable C<PL_na> has been
131 used in this case. But that can be quite inefficient because C<PL_na> must
132 be accessed in thread-local storage in threaded Perl. In any case, remember
133 that Perl allows arbitrary strings of data that may both contain NULs and
134 might not be terminated by a NUL.
136 Also remember that C doesn't allow you to safely say C<foo(SvPV(s, len),
137 len);>. It might work with your
138 compiler, but it won't work for everyone.
139 Break this sort of statement up into separate assignments:
147 If you want to know if the scalar value is TRUE, you can use:
151 Although Perl will automatically grow strings for you, if you need to force
152 Perl to allocate more memory for your SV, you can use the macro
154 SvGROW(SV*, STRLEN newlen)
156 which will determine if more memory needs to be allocated. If so, it will
157 call the function C<sv_grow>. Note that C<SvGROW> can only increase, not
158 decrease, the allocated memory of an SV and that it does not automatically
159 add space for the trailing NUL byte (perl's own string functions typically do
160 C<SvGROW(sv, len + 1)>).
162 If you have an SV and want to know what kind of data Perl thinks is stored
163 in it, you can use the following macros to check the type of SV you have.
169 You can get and set the current length of the string stored in an SV with
170 the following macros:
173 SvCUR_set(SV*, I32 val)
175 You can also get a pointer to the end of the string stored in the SV
180 But note that these last three macros are valid only if C<SvPOK()> is true.
182 If you want to append something to the end of string stored in an C<SV*>,
183 you can use the following functions:
185 void sv_catpv(SV*, const char*);
186 void sv_catpvn(SV*, const char*, STRLEN);
187 void sv_catpvf(SV*, const char*, ...);
188 void sv_vcatpvfn(SV*, const char*, STRLEN, va_list *, SV **,
190 void sv_catsv(SV*, SV*);
192 The first function calculates the length of the string to be appended by
193 using C<strlen>. In the second, you specify the length of the string
194 yourself. The third function processes its arguments like C<sprintf> and
195 appends the formatted output. The fourth function works like C<vsprintf>.
196 You can specify the address and length of an array of SVs instead of the
197 va_list argument. The fifth function
198 extends the string stored in the first
199 SV with the string stored in the second SV. It also forces the second SV
200 to be interpreted as a string.
202 The C<sv_cat*()> functions are not generic enough to operate on values that
203 have "magic". See L<Magic Virtual Tables> later in this document.
205 If you know the name of a scalar variable, you can get a pointer to its SV
206 by using the following:
208 SV* get_sv("package::varname", 0);
210 This returns NULL if the variable does not exist.
212 If you want to know if this variable (or any other SV) is actually C<defined>,
217 The scalar C<undef> value is stored in an SV instance called C<PL_sv_undef>.
219 Its address can be used whenever an C<SV*> is needed. Make sure that
220 you don't try to compare a random sv with C<&PL_sv_undef>. For example
221 when interfacing Perl code, it'll work correctly for:
225 But won't work when called as:
230 So to repeat always use SvOK() to check whether an sv is defined.
232 Also you have to be careful when using C<&PL_sv_undef> as a value in
233 AVs or HVs (see L<AVs, HVs and undefined values>).
235 There are also the two values C<PL_sv_yes> and C<PL_sv_no>, which contain
236 boolean TRUE and FALSE values, respectively. Like C<PL_sv_undef>, their
237 addresses can be used whenever an C<SV*> is needed.
239 Do not be fooled into thinking that C<(SV *) 0> is the same as C<&PL_sv_undef>.
243 if (I-am-to-return-a-real-value) {
244 sv = sv_2mortal(newSViv(42));
248 This code tries to return a new SV (which contains the value 42) if it should
249 return a real value, or undef otherwise. Instead it has returned a NULL
250 pointer which, somewhere down the line, will cause a segmentation violation,
251 bus error, or just weird results. Change the zero to C<&PL_sv_undef> in the
252 first line and all will be well.
254 To free an SV that you've created, call C<SvREFCNT_dec(SV*)>. Normally this
255 call is not necessary (see L<Reference Counts and Mortality>).
259 Perl provides the function C<sv_chop> to efficiently remove characters
260 from the beginning of a string; you give it an SV and a pointer to
261 somewhere inside the PV, and it discards everything before the
262 pointer. The efficiency comes by means of a little hack: instead of
263 actually removing the characters, C<sv_chop> sets the flag C<OOK>
264 (offset OK) to signal to other functions that the offset hack is in
265 effect, and it moves the PV pointer (called C<SvPVX>) forward
266 by the number of bytes chopped off, and adjusts C<SvCUR> and C<SvLEN>
267 accordingly. (A portion of the space between the old and new PV
268 pointers is used to store the count of chopped bytes.)
270 Hence, at this point, the start of the buffer that we allocated lives
271 at C<SvPVX(sv) - SvIV(sv)> in memory and the PV pointer is pointing
272 into the middle of this allocated storage.
274 This is best demonstrated by example:
276 % ./perl -Ilib -MDevel::Peek -le '$a="12345"; $a=~s/.//; Dump($a)'
277 SV = PVIV(0x8128450) at 0x81340f0
279 FLAGS = (POK,OOK,pPOK)
281 PV = 0x8135781 ( "1" . ) "2345"\0
285 Here the number of bytes chopped off (1) is put into IV, and
286 C<Devel::Peek::Dump> helpfully reminds us that this is an offset. The
287 portion of the string between the "real" and the "fake" beginnings is
288 shown in parentheses, and the values of C<SvCUR> and C<SvLEN> reflect
289 the fake beginning, not the real one.
291 Something similar to the offset hack is performed on AVs to enable
292 efficient shifting and splicing off the beginning of the array; while
293 C<AvARRAY> points to the first element in the array that is visible from
294 Perl, C<AvALLOC> points to the real start of the C array. These are
295 usually the same, but a C<shift> operation can be carried out by
296 increasing C<AvARRAY> by one and decreasing C<AvFILL> and C<AvMAX>.
297 Again, the location of the real start of the C array only comes into
298 play when freeing the array. See C<av_shift> in F<av.c>.
300 =head2 What's Really Stored in an SV?
302 Recall that the usual method of determining the type of scalar you have is
303 to use C<Sv*OK> macros. Because a scalar can be both a number and a string,
304 usually these macros will always return TRUE and calling the C<Sv*V>
305 macros will do the appropriate conversion of string to integer/double or
306 integer/double to string.
308 If you I<really> need to know if you have an integer, double, or string
309 pointer in an SV, you can use the following three macros instead:
315 These will tell you if you truly have an integer, double, or string pointer
316 stored in your SV. The "p" stands for private.
318 There are various ways in which the private and public flags may differ.
319 For example, in perl 5.16 and earlier a tied SV may have a valid
320 underlying value in the IV slot (so SvIOKp is true), but the data
321 should be accessed via the FETCH routine rather than directly,
322 so SvIOK is false. (In perl 5.18 onwards, tied scalars use
323 the flags the same way as untied scalars.) Another is when
324 numeric conversion has occurred and precision has been lost: only the
325 private flag is set on 'lossy' values. So when an NV is converted to an
326 IV with loss, SvIOKp, SvNOKp and SvNOK will be set, while SvIOK wont be.
328 In general, though, it's best to use the C<Sv*V> macros.
330 =head2 Working with AVs
332 There are two ways to create and load an AV. The first method creates an
337 The second method both creates the AV and initially populates it with SVs:
339 AV* av_make(SSize_t num, SV **ptr);
341 The second argument points to an array containing C<num> C<SV*>'s. Once the
342 AV has been created, the SVs can be destroyed, if so desired.
344 Once the AV has been created, the following operations are possible on it:
346 void av_push(AV*, SV*);
349 void av_unshift(AV*, SSize_t num);
351 These should be familiar operations, with the exception of C<av_unshift>.
352 This routine adds C<num> elements at the front of the array with the C<undef>
353 value. You must then use C<av_store> (described below) to assign values
354 to these new elements.
356 Here are some other functions:
358 SSize_t av_top_index(AV*);
359 SV** av_fetch(AV*, SSize_t key, I32 lval);
360 SV** av_store(AV*, SSize_t key, SV* val);
362 The C<av_top_index> function returns the highest index value in an array (just
363 like $#array in Perl). If the array is empty, -1 is returned. The
364 C<av_fetch> function returns the value at index C<key>, but if C<lval>
365 is non-zero, then C<av_fetch> will store an undef value at that index.
366 The C<av_store> function stores the value C<val> at index C<key>, and does
367 not increment the reference count of C<val>. Thus the caller is responsible
368 for taking care of that, and if C<av_store> returns NULL, the caller will
369 have to decrement the reference count to avoid a memory leak. Note that
370 C<av_fetch> and C<av_store> both return C<SV**>'s, not C<SV*>'s as their
377 void av_extend(AV*, SSize_t key);
379 The C<av_clear> function deletes all the elements in the AV* array, but
380 does not actually delete the array itself. The C<av_undef> function will
381 delete all the elements in the array plus the array itself. The
382 C<av_extend> function extends the array so that it contains at least C<key+1>
383 elements. If C<key+1> is less than the currently allocated length of the array,
384 then nothing is done.
386 If you know the name of an array variable, you can get a pointer to its AV
387 by using the following:
389 AV* get_av("package::varname", 0);
391 This returns NULL if the variable does not exist.
393 See L<Understanding the Magic of Tied Hashes and Arrays> for more
394 information on how to use the array access functions on tied arrays.
396 =head2 Working with HVs
398 To create an HV, you use the following routine:
402 Once the HV has been created, the following operations are possible on it:
404 SV** hv_store(HV*, const char* key, U32 klen, SV* val, U32 hash);
405 SV** hv_fetch(HV*, const char* key, U32 klen, I32 lval);
407 The C<klen> parameter is the length of the key being passed in (Note that
408 you cannot pass 0 in as a value of C<klen> to tell Perl to measure the
409 length of the key). The C<val> argument contains the SV pointer to the
410 scalar being stored, and C<hash> is the precomputed hash value (zero if
411 you want C<hv_store> to calculate it for you). The C<lval> parameter
412 indicates whether this fetch is actually a part of a store operation, in
413 which case a new undefined value will be added to the HV with the supplied
414 key and C<hv_fetch> will return as if the value had already existed.
416 Remember that C<hv_store> and C<hv_fetch> return C<SV**>'s and not just
417 C<SV*>. To access the scalar value, you must first dereference the return
418 value. However, you should check to make sure that the return value is
419 not NULL before dereferencing it.
421 The first of these two functions checks if a hash table entry exists, and the
424 bool hv_exists(HV*, const char* key, U32 klen);
425 SV* hv_delete(HV*, const char* key, U32 klen, I32 flags);
427 If C<flags> does not include the C<G_DISCARD> flag then C<hv_delete> will
428 create and return a mortal copy of the deleted value.
430 And more miscellaneous functions:
435 Like their AV counterparts, C<hv_clear> deletes all the entries in the hash
436 table but does not actually delete the hash table. The C<hv_undef> deletes
437 both the entries and the hash table itself.
439 Perl keeps the actual data in a linked list of structures with a typedef of HE.
440 These contain the actual key and value pointers (plus extra administrative
441 overhead). The key is a string pointer; the value is an C<SV*>. However,
442 once you have an C<HE*>, to get the actual key and value, use the routines
445 I32 hv_iterinit(HV*);
446 /* Prepares starting point to traverse hash table */
447 HE* hv_iternext(HV*);
448 /* Get the next entry, and return a pointer to a
449 structure that has both the key and value */
450 char* hv_iterkey(HE* entry, I32* retlen);
451 /* Get the key from an HE structure and also return
452 the length of the key string */
453 SV* hv_iterval(HV*, HE* entry);
454 /* Return an SV pointer to the value of the HE
456 SV* hv_iternextsv(HV*, char** key, I32* retlen);
457 /* This convenience routine combines hv_iternext,
458 hv_iterkey, and hv_iterval. The key and retlen
459 arguments are return values for the key and its
460 length. The value is returned in the SV* argument */
462 If you know the name of a hash variable, you can get a pointer to its HV
463 by using the following:
465 HV* get_hv("package::varname", 0);
467 This returns NULL if the variable does not exist.
469 The hash algorithm is defined in the C<PERL_HASH> macro:
471 PERL_HASH(hash, key, klen)
473 The exact implementation of this macro varies by architecture and version
474 of perl, and the return value may change per invocation, so the value
475 is only valid for the duration of a single perl process.
477 See L<Understanding the Magic of Tied Hashes and Arrays> for more
478 information on how to use the hash access functions on tied hashes.
480 =head2 Hash API Extensions
482 Beginning with version 5.004, the following functions are also supported:
484 HE* hv_fetch_ent (HV* tb, SV* key, I32 lval, U32 hash);
485 HE* hv_store_ent (HV* tb, SV* key, SV* val, U32 hash);
487 bool hv_exists_ent (HV* tb, SV* key, U32 hash);
488 SV* hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash);
490 SV* hv_iterkeysv (HE* entry);
492 Note that these functions take C<SV*> keys, which simplifies writing
493 of extension code that deals with hash structures. These functions
494 also allow passing of C<SV*> keys to C<tie> functions without forcing
495 you to stringify the keys (unlike the previous set of functions).
497 They also return and accept whole hash entries (C<HE*>), making their
498 use more efficient (since the hash number for a particular string
499 doesn't have to be recomputed every time). See L<perlapi> for detailed
502 The following macros must always be used to access the contents of hash
503 entries. Note that the arguments to these macros must be simple
504 variables, since they may get evaluated more than once. See
505 L<perlapi> for detailed descriptions of these macros.
507 HePV(HE* he, STRLEN len)
511 HeSVKEY_force(HE* he)
512 HeSVKEY_set(HE* he, SV* sv)
514 These two lower level macros are defined, but must only be used when
515 dealing with keys that are not C<SV*>s:
520 Note that both C<hv_store> and C<hv_store_ent> do not increment the
521 reference count of the stored C<val>, which is the caller's responsibility.
522 If these functions return a NULL value, the caller will usually have to
523 decrement the reference count of C<val> to avoid a memory leak.
525 =head2 AVs, HVs and undefined values
527 Sometimes you have to store undefined values in AVs or HVs. Although
528 this may be a rare case, it can be tricky. That's because you're
529 used to using C<&PL_sv_undef> if you need an undefined SV.
531 For example, intuition tells you that this XS code:
534 av_store( av, 0, &PL_sv_undef );
536 is equivalent to this Perl code:
541 Unfortunately, this isn't true. In perl 5.18 and earlier, AVs use C<&PL_sv_undef> as a marker
542 for indicating that an array element has not yet been initialized.
543 Thus, C<exists $av[0]> would be true for the above Perl code, but
544 false for the array generated by the XS code. In perl 5.20, storing
545 &PL_sv_undef will create a read-only element, because the scalar
546 &PL_sv_undef itself is stored, not a copy.
548 Similar problems can occur when storing C<&PL_sv_undef> in HVs:
550 hv_store( hv, "key", 3, &PL_sv_undef, 0 );
552 This will indeed make the value C<undef>, but if you try to modify
553 the value of C<key>, you'll get the following error:
555 Modification of non-creatable hash value attempted
557 In perl 5.8.0, C<&PL_sv_undef> was also used to mark placeholders
558 in restricted hashes. This caused such hash entries not to appear
559 when iterating over the hash or when checking for the keys
560 with the C<hv_exists> function.
562 You can run into similar problems when you store C<&PL_sv_yes> or
563 C<&PL_sv_no> into AVs or HVs. Trying to modify such elements
564 will give you the following error:
566 Modification of a read-only value attempted
568 To make a long story short, you can use the special variables
569 C<&PL_sv_undef>, C<&PL_sv_yes> and C<&PL_sv_no> with AVs and
570 HVs, but you have to make sure you know what you're doing.
572 Generally, if you want to store an undefined value in an AV
573 or HV, you should not use C<&PL_sv_undef>, but rather create a
574 new undefined value using the C<newSV> function, for example:
576 av_store( av, 42, newSV(0) );
577 hv_store( hv, "foo", 3, newSV(0), 0 );
581 References are a special type of scalar that point to other data types
582 (including other references).
584 To create a reference, use either of the following functions:
586 SV* newRV_inc((SV*) thing);
587 SV* newRV_noinc((SV*) thing);
589 The C<thing> argument can be any of an C<SV*>, C<AV*>, or C<HV*>. The
590 functions are identical except that C<newRV_inc> increments the reference
591 count of the C<thing>, while C<newRV_noinc> does not. For historical
592 reasons, C<newRV> is a synonym for C<newRV_inc>.
594 Once you have a reference, you can use the following macro to dereference
599 then call the appropriate routines, casting the returned C<SV*> to either an
600 C<AV*> or C<HV*>, if required.
602 To determine if an SV is a reference, you can use the following macro:
606 To discover what type of value the reference refers to, use the following
607 macro and then check the return value.
611 The most useful types that will be returned are:
617 SVt_PVGV Glob (possibly a file handle)
619 See L<perlapi/svtype> for more details.
621 =head2 Blessed References and Class Objects
623 References are also used to support object-oriented programming. In perl's
624 OO lexicon, an object is simply a reference that has been blessed into a
625 package (or class). Once blessed, the programmer may now use the reference
626 to access the various methods in the class.
628 A reference can be blessed into a package with the following function:
630 SV* sv_bless(SV* sv, HV* stash);
632 The C<sv> argument must be a reference value. The C<stash> argument
633 specifies which class the reference will belong to. See
634 L<Stashes and Globs> for information on converting class names into stashes.
636 /* Still under construction */
638 The following function upgrades rv to reference if not already one.
639 Creates a new SV for rv to point to. If C<classname> is non-null, the SV
640 is blessed into the specified class. SV is returned.
642 SV* newSVrv(SV* rv, const char* classname);
644 The following three functions copy integer, unsigned integer or double
645 into an SV whose reference is C<rv>. SV is blessed if C<classname> is
648 SV* sv_setref_iv(SV* rv, const char* classname, IV iv);
649 SV* sv_setref_uv(SV* rv, const char* classname, UV uv);
650 SV* sv_setref_nv(SV* rv, const char* classname, NV iv);
652 The following function copies the pointer value (I<the address, not the
653 string!>) into an SV whose reference is rv. SV is blessed if C<classname>
656 SV* sv_setref_pv(SV* rv, const char* classname, void* pv);
658 The following function copies a string into an SV whose reference is C<rv>.
659 Set length to 0 to let Perl calculate the string length. SV is blessed if
660 C<classname> is non-null.
662 SV* sv_setref_pvn(SV* rv, const char* classname, char* pv,
665 The following function tests whether the SV is blessed into the specified
666 class. It does not check inheritance relationships.
668 int sv_isa(SV* sv, const char* name);
670 The following function tests whether the SV is a reference to a blessed object.
672 int sv_isobject(SV* sv);
674 The following function tests whether the SV is derived from the specified
675 class. SV can be either a reference to a blessed object or a string
676 containing a class name. This is the function implementing the
677 C<UNIVERSAL::isa> functionality.
679 bool sv_derived_from(SV* sv, const char* name);
681 To check if you've got an object derived from a specific class you have
684 if (sv_isobject(sv) && sv_derived_from(sv, class)) { ... }
686 =head2 Creating New Variables
688 To create a new Perl variable with an undef value which can be accessed from
689 your Perl script, use the following routines, depending on the variable type.
691 SV* get_sv("package::varname", GV_ADD);
692 AV* get_av("package::varname", GV_ADD);
693 HV* get_hv("package::varname", GV_ADD);
695 Notice the use of GV_ADD as the second parameter. The new variable can now
696 be set, using the routines appropriate to the data type.
698 There are additional macros whose values may be bitwise OR'ed with the
699 C<GV_ADD> argument to enable certain extra features. Those bits are:
705 Marks the variable as multiply defined, thus preventing the:
707 Name <varname> used only once: possible typo
715 Had to create <varname> unexpectedly
717 if the variable did not exist before the function was called.
721 If you do not specify a package name, the variable is created in the current
724 =head2 Reference Counts and Mortality
726 Perl uses a reference count-driven garbage collection mechanism. SVs,
727 AVs, or HVs (xV for short in the following) start their life with a
728 reference count of 1. If the reference count of an xV ever drops to 0,
729 then it will be destroyed and its memory made available for reuse.
731 This normally doesn't happen at the Perl level unless a variable is
732 undef'ed or the last variable holding a reference to it is changed or
733 overwritten. At the internal level, however, reference counts can be
734 manipulated with the following macros:
736 int SvREFCNT(SV* sv);
737 SV* SvREFCNT_inc(SV* sv);
738 void SvREFCNT_dec(SV* sv);
740 However, there is one other function which manipulates the reference
741 count of its argument. The C<newRV_inc> function, you will recall,
742 creates a reference to the specified argument. As a side effect,
743 it increments the argument's reference count. If this is not what
744 you want, use C<newRV_noinc> instead.
746 For example, imagine you want to return a reference from an XSUB function.
747 Inside the XSUB routine, you create an SV which initially has a reference
748 count of one. Then you call C<newRV_inc>, passing it the just-created SV.
749 This returns the reference as a new SV, but the reference count of the
750 SV you passed to C<newRV_inc> has been incremented to two. Now you
751 return the reference from the XSUB routine and forget about the SV.
752 But Perl hasn't! Whenever the returned reference is destroyed, the
753 reference count of the original SV is decreased to one and nothing happens.
754 The SV will hang around without any way to access it until Perl itself
755 terminates. This is a memory leak.
757 The correct procedure, then, is to use C<newRV_noinc> instead of
758 C<newRV_inc>. Then, if and when the last reference is destroyed,
759 the reference count of the SV will go to zero and it will be destroyed,
760 stopping any memory leak.
762 There are some convenience functions available that can help with the
763 destruction of xVs. These functions introduce the concept of "mortality".
764 An xV that is mortal has had its reference count marked to be decremented,
765 but not actually decremented, until "a short time later". Generally the
766 term "short time later" means a single Perl statement, such as a call to
767 an XSUB function. The actual determinant for when mortal xVs have their
768 reference count decremented depends on two macros, SAVETMPS and FREETMPS.
769 See L<perlcall> and L<perlxs> for more details on these macros.
771 "Mortalization" then is at its simplest a deferred C<SvREFCNT_dec>.
772 However, if you mortalize a variable twice, the reference count will
773 later be decremented twice.
775 "Mortal" SVs are mainly used for SVs that are placed on perl's stack.
776 For example an SV which is created just to pass a number to a called sub
777 is made mortal to have it cleaned up automatically when it's popped off
778 the stack. Similarly, results returned by XSUBs (which are pushed on the
779 stack) are often made mortal.
781 To create a mortal variable, use the functions:
785 SV* sv_mortalcopy(SV*)
787 The first call creates a mortal SV (with no value), the second converts an existing
788 SV to a mortal SV (and thus defers a call to C<SvREFCNT_dec>), and the
789 third creates a mortal copy of an existing SV.
790 Because C<sv_newmortal> gives the new SV no value, it must normally be given one
791 via C<sv_setpv>, C<sv_setiv>, etc. :
793 SV *tmp = sv_newmortal();
794 sv_setiv(tmp, an_integer);
796 As that is multiple C statements it is quite common so see this idiom instead:
798 SV *tmp = sv_2mortal(newSViv(an_integer));
801 You should be careful about creating mortal variables. Strange things
802 can happen if you make the same value mortal within multiple contexts,
803 or if you make a variable mortal multiple
804 times. Thinking of "Mortalization"
805 as deferred C<SvREFCNT_dec> should help to minimize such problems.
806 For example if you are passing an SV which you I<know> has a high enough REFCNT
807 to survive its use on the stack you need not do any mortalization.
808 If you are not sure then doing an C<SvREFCNT_inc> and C<sv_2mortal>, or
809 making a C<sv_mortalcopy> is safer.
811 The mortal routines are not just for SVs; AVs and HVs can be
812 made mortal by passing their address (type-casted to C<SV*>) to the
813 C<sv_2mortal> or C<sv_mortalcopy> routines.
815 =head2 Stashes and Globs
817 A B<stash> is a hash that contains all variables that are defined
818 within a package. Each key of the stash is a symbol
819 name (shared by all the different types of objects that have the same
820 name), and each value in the hash table is a GV (Glob Value). This GV
821 in turn contains references to the various objects of that name,
822 including (but not limited to) the following:
831 There is a single stash called C<PL_defstash> that holds the items that exist
832 in the C<main> package. To get at the items in other packages, append the
833 string "::" to the package name. The items in the C<Foo> package are in
834 the stash C<Foo::> in PL_defstash. The items in the C<Bar::Baz> package are
835 in the stash C<Baz::> in C<Bar::>'s stash.
837 To get the stash pointer for a particular package, use the function:
839 HV* gv_stashpv(const char* name, I32 flags)
840 HV* gv_stashsv(SV*, I32 flags)
842 The first function takes a literal string, the second uses the string stored
843 in the SV. Remember that a stash is just a hash table, so you get back an
844 C<HV*>. The C<flags> flag will create a new package if it is set to GV_ADD.
846 The name that C<gv_stash*v> wants is the name of the package whose symbol table
847 you want. The default package is called C<main>. If you have multiply nested
848 packages, pass their names to C<gv_stash*v>, separated by C<::> as in the Perl
851 Alternately, if you have an SV that is a blessed reference, you can find
852 out the stash pointer by using:
854 HV* SvSTASH(SvRV(SV*));
856 then use the following to get the package name itself:
858 char* HvNAME(HV* stash);
860 If you need to bless or re-bless an object you can use the following
863 SV* sv_bless(SV*, HV* stash)
865 where the first argument, an C<SV*>, must be a reference, and the second
866 argument is a stash. The returned C<SV*> can now be used in the same way
869 For more information on references and blessings, consult L<perlref>.
871 =head2 Double-Typed SVs
873 Scalar variables normally contain only one type of value, an integer,
874 double, pointer, or reference. Perl will automatically convert the
875 actual scalar data from the stored type into the requested type.
877 Some scalar variables contain more than one type of scalar data. For
878 example, the variable C<$!> contains either the numeric value of C<errno>
879 or its string equivalent from either C<strerror> or C<sys_errlist[]>.
881 To force multiple data values into an SV, you must do two things: use the
882 C<sv_set*v> routines to add the additional scalar type, then set a flag
883 so that Perl will believe it contains more than one type of data. The
884 four macros to set the flags are:
891 The particular macro you must use depends on which C<sv_set*v> routine
892 you called first. This is because every C<sv_set*v> routine turns on
893 only the bit for the particular type of data being set, and turns off
896 For example, to create a new Perl variable called "dberror" that contains
897 both the numeric and descriptive string error values, you could use the
901 extern char *dberror_list;
903 SV* sv = get_sv("dberror", GV_ADD);
904 sv_setiv(sv, (IV) dberror);
905 sv_setpv(sv, dberror_list[dberror]);
908 If the order of C<sv_setiv> and C<sv_setpv> had been reversed, then the
909 macro C<SvPOK_on> would need to be called instead of C<SvIOK_on>.
911 =head2 Read-Only Values
913 In Perl 5.16 and earlier, copy-on-write (see the next section) shared a
914 flag bit with read-only scalars. So the only way to test whether
915 C<sv_setsv>, etc., will raise a "Modification of a read-only value" error
916 in those versions is:
918 SvREADONLY(sv) && !SvIsCOW(sv)
920 Under Perl 5.18 and later, SvREADONLY only applies to read-only variables,
921 and, under 5.20, copy-on-write scalars can also be read-only, so the above
922 check is incorrect. You just want:
926 If you need to do this check often, define your own macro like this:
928 #if PERL_VERSION >= 18
929 # define SvTRULYREADONLY(sv) SvREADONLY(sv)
931 # define SvTRULYREADONLY(sv) (SvREADONLY(sv) && !SvIsCOW(sv))
936 Perl implements a copy-on-write (COW) mechanism for scalars, in which
937 string copies are not immediately made when requested, but are deferred
938 until made necessary by one or the other scalar changing. This is mostly
939 transparent, but one must take care not to modify string buffers that are
940 shared by multiple SVs.
942 You can test whether an SV is using copy-on-write with C<SvIsCOW(sv)>.
944 You can force an SV to make its own copy of its string buffer by calling C<sv_force_normal(sv)> or SvPV_force_nolen(sv).
946 If you want to make the SV drop its string buffer, use
947 C<sv_force_normal_flags(sv, SV_COW_DROP_PV)> or simply
948 C<sv_setsv(sv, NULL)>.
950 All of these functions will croak on read-only scalars (see the previous
951 section for more on those).
953 To test that your code is behaving correctly and not modifying COW buffers,
954 on systems that support L<mmap(2)> (i.e., Unix) you can configure perl with
955 C<-Accflags=-DPERL_DEBUG_READONLY_COW> and it will turn buffer violations
956 into crashes. You will find it to be marvellously slow, so you may want to
957 skip perl's own tests.
959 =head2 Magic Variables
961 [This section still under construction. Ignore everything here. Post no
962 bills. Everything not permitted is forbidden.]
964 Any SV may be magical, that is, it has special features that a normal
965 SV does not have. These features are stored in the SV structure in a
966 linked list of C<struct magic>'s, typedef'ed to C<MAGIC>.
979 Note this is current as of patchlevel 0, and could change at any time.
981 =head2 Assigning Magic
983 Perl adds magic to an SV using the sv_magic function:
985 void sv_magic(SV* sv, SV* obj, int how, const char* name, I32 namlen);
987 The C<sv> argument is a pointer to the SV that is to acquire a new magical
990 If C<sv> is not already magical, Perl uses the C<SvUPGRADE> macro to
991 convert C<sv> to type C<SVt_PVMG>.
992 Perl then continues by adding new magic
993 to the beginning of the linked list of magical features. Any prior entry
994 of the same type of magic is deleted. Note that this can be overridden,
995 and multiple instances of the same type of magic can be associated with an
998 The C<name> and C<namlen> arguments are used to associate a string with
999 the magic, typically the name of a variable. C<namlen> is stored in the
1000 C<mg_len> field and if C<name> is non-null then either a C<savepvn> copy of
1001 C<name> or C<name> itself is stored in the C<mg_ptr> field, depending on
1002 whether C<namlen> is greater than zero or equal to zero respectively. As a
1003 special case, if C<(name && namlen == HEf_SVKEY)> then C<name> is assumed
1004 to contain an C<SV*> and is stored as-is with its REFCNT incremented.
1006 The sv_magic function uses C<how> to determine which, if any, predefined
1007 "Magic Virtual Table" should be assigned to the C<mg_virtual> field.
1008 See the L<Magic Virtual Tables> section below. The C<how> argument is also
1009 stored in the C<mg_type> field. The value of
1010 C<how> should be chosen from the set of macros
1011 C<PERL_MAGIC_foo> found in F<perl.h>. Note that before
1012 these macros were added, Perl internals used to directly use character
1013 literals, so you may occasionally come across old code or documentation
1014 referring to 'U' magic rather than C<PERL_MAGIC_uvar> for example.
1016 The C<obj> argument is stored in the C<mg_obj> field of the C<MAGIC>
1017 structure. If it is not the same as the C<sv> argument, the reference
1018 count of the C<obj> object is incremented. If it is the same, or if
1019 the C<how> argument is C<PERL_MAGIC_arylen>, or if it is a NULL pointer,
1020 then C<obj> is merely stored, without the reference count being incremented.
1022 See also C<sv_magicext> in L<perlapi> for a more flexible way to add magic
1025 There is also a function to add magic to an C<HV>:
1027 void hv_magic(HV *hv, GV *gv, int how);
1029 This simply calls C<sv_magic> and coerces the C<gv> argument into an C<SV>.
1031 To remove the magic from an SV, call the function sv_unmagic:
1033 int sv_unmagic(SV *sv, int type);
1035 The C<type> argument should be equal to the C<how> value when the C<SV>
1036 was initially made magical.
1038 However, note that C<sv_unmagic> removes all magic of a certain C<type> from the
1039 C<SV>. If you want to remove only certain
1040 magic of a C<type> based on the magic
1041 virtual table, use C<sv_unmagicext> instead:
1043 int sv_unmagicext(SV *sv, int type, MGVTBL *vtbl);
1045 =head2 Magic Virtual Tables
1047 The C<mg_virtual> field in the C<MAGIC> structure is a pointer to an
1048 C<MGVTBL>, which is a structure of function pointers and stands for
1049 "Magic Virtual Table" to handle the various operations that might be
1050 applied to that variable.
1052 The C<MGVTBL> has five (or sometimes eight) pointers to the following
1055 int (*svt_get)(SV* sv, MAGIC* mg);
1056 int (*svt_set)(SV* sv, MAGIC* mg);
1057 U32 (*svt_len)(SV* sv, MAGIC* mg);
1058 int (*svt_clear)(SV* sv, MAGIC* mg);
1059 int (*svt_free)(SV* sv, MAGIC* mg);
1061 int (*svt_copy)(SV *sv, MAGIC* mg, SV *nsv,
1062 const char *name, I32 namlen);
1063 int (*svt_dup)(MAGIC *mg, CLONE_PARAMS *param);
1064 int (*svt_local)(SV *nsv, MAGIC *mg);
1067 This MGVTBL structure is set at compile-time in F<perl.h> and there are
1068 currently 32 types. These different structures contain pointers to various
1069 routines that perform additional actions depending on which function is
1072 Function pointer Action taken
1073 ---------------- ------------
1074 svt_get Do something before the value of the SV is
1076 svt_set Do something after the SV is assigned a value.
1077 svt_len Report on the SV's length.
1078 svt_clear Clear something the SV represents.
1079 svt_free Free any extra storage associated with the SV.
1081 svt_copy copy tied variable magic to a tied element
1082 svt_dup duplicate a magic structure during thread cloning
1083 svt_local copy magic to local value during 'local'
1085 For instance, the MGVTBL structure called C<vtbl_sv> (which corresponds
1086 to an C<mg_type> of C<PERL_MAGIC_sv>) contains:
1088 { magic_get, magic_set, magic_len, 0, 0 }
1090 Thus, when an SV is determined to be magical and of type C<PERL_MAGIC_sv>,
1091 if a get operation is being performed, the routine C<magic_get> is
1092 called. All the various routines for the various magical types begin
1093 with C<magic_>. NOTE: the magic routines are not considered part of
1094 the Perl API, and may not be exported by the Perl library.
1096 The last three slots are a recent addition, and for source code
1097 compatibility they are only checked for if one of the three flags
1098 MGf_COPY, MGf_DUP or MGf_LOCAL is set in mg_flags.
1099 This means that most code can continue declaring
1100 a vtable as a 5-element value. These three are
1101 currently used exclusively by the threading code, and are highly subject
1104 The current kinds of Magic Virtual Tables are:
1107 This table is generated by regen/mg_vtable.pl. Any changes made here
1110 =for mg_vtable.pl begin
1113 (old-style char and macro) MGVTBL Type of magic
1114 -------------------------- ------ -------------
1115 \0 PERL_MAGIC_sv vtbl_sv Special scalar variable
1116 # PERL_MAGIC_arylen vtbl_arylen Array length ($#ary)
1117 % PERL_MAGIC_rhash (none) extra data for restricted
1119 & PERL_MAGIC_proto (none) my sub prototype CV
1120 . PERL_MAGIC_pos vtbl_pos pos() lvalue
1121 : PERL_MAGIC_symtab (none) extra data for symbol
1123 < PERL_MAGIC_backref vtbl_backref for weak ref data
1124 @ PERL_MAGIC_arylen_p (none) to move arylen out of XPVAV
1125 B PERL_MAGIC_bm vtbl_regexp Boyer-Moore
1126 (fast string search)
1127 c PERL_MAGIC_overload_table vtbl_ovrld Holds overload table
1129 D PERL_MAGIC_regdata vtbl_regdata Regex match position data
1131 d PERL_MAGIC_regdatum vtbl_regdatum Regex match position data
1133 E PERL_MAGIC_env vtbl_env %ENV hash
1134 e PERL_MAGIC_envelem vtbl_envelem %ENV hash element
1135 f PERL_MAGIC_fm vtbl_regexp Formline
1137 g PERL_MAGIC_regex_global vtbl_mglob m//g target
1138 H PERL_MAGIC_hints vtbl_hints %^H hash
1139 h PERL_MAGIC_hintselem vtbl_hintselem %^H hash element
1140 I PERL_MAGIC_isa vtbl_isa @ISA array
1141 i PERL_MAGIC_isaelem vtbl_isaelem @ISA array element
1142 k PERL_MAGIC_nkeys vtbl_nkeys scalar(keys()) lvalue
1143 L PERL_MAGIC_dbfile (none) Debugger %_<filename
1144 l PERL_MAGIC_dbline vtbl_dbline Debugger %_<filename
1146 N PERL_MAGIC_shared (none) Shared between threads
1147 n PERL_MAGIC_shared_scalar (none) Shared between threads
1148 o PERL_MAGIC_collxfrm vtbl_collxfrm Locale transformation
1149 P PERL_MAGIC_tied vtbl_pack Tied array or hash
1150 p PERL_MAGIC_tiedelem vtbl_packelem Tied array or hash element
1151 q PERL_MAGIC_tiedscalar vtbl_packelem Tied scalar or handle
1152 r PERL_MAGIC_qr vtbl_regexp precompiled qr// regex
1153 S PERL_MAGIC_sig (none) %SIG hash
1154 s PERL_MAGIC_sigelem vtbl_sigelem %SIG hash element
1155 t PERL_MAGIC_taint vtbl_taint Taintedness
1156 U PERL_MAGIC_uvar vtbl_uvar Available for use by
1158 u PERL_MAGIC_uvar_elem (none) Reserved for use by
1160 V PERL_MAGIC_vstring (none) SV was vstring literal
1161 v PERL_MAGIC_vec vtbl_vec vec() lvalue
1162 w PERL_MAGIC_utf8 vtbl_utf8 Cached UTF-8 information
1163 x PERL_MAGIC_substr vtbl_substr substr() lvalue
1164 y PERL_MAGIC_defelem vtbl_defelem Shadow "foreach" iterator
1165 variable / smart parameter
1167 ] PERL_MAGIC_checkcall vtbl_checkcall inlining/mutation of call
1169 ~ PERL_MAGIC_ext (none) Available for use by
1172 =for mg_vtable.pl end
1174 When an uppercase and lowercase letter both exist in the table, then the
1175 uppercase letter is typically used to represent some kind of composite type
1176 (a list or a hash), and the lowercase letter is used to represent an element
1177 of that composite type. Some internals code makes use of this case
1178 relationship. However, 'v' and 'V' (vec and v-string) are in no way related.
1180 The C<PERL_MAGIC_ext> and C<PERL_MAGIC_uvar> magic types are defined
1181 specifically for use by extensions and will not be used by perl itself.
1182 Extensions can use C<PERL_MAGIC_ext> magic to 'attach' private information
1183 to variables (typically objects). This is especially useful because
1184 there is no way for normal perl code to corrupt this private information
1185 (unlike using extra elements of a hash object).
1187 Similarly, C<PERL_MAGIC_uvar> magic can be used much like tie() to call a
1188 C function any time a scalar's value is used or changed. The C<MAGIC>'s
1189 C<mg_ptr> field points to a C<ufuncs> structure:
1192 I32 (*uf_val)(pTHX_ IV, SV*);
1193 I32 (*uf_set)(pTHX_ IV, SV*);
1197 When the SV is read from or written to, the C<uf_val> or C<uf_set>
1198 function will be called with C<uf_index> as the first arg and a pointer to
1199 the SV as the second. A simple example of how to add C<PERL_MAGIC_uvar>
1200 magic is shown below. Note that the ufuncs structure is copied by
1201 sv_magic, so you can safely allocate it on the stack.
1209 uf.uf_val = &my_get_fn;
1210 uf.uf_set = &my_set_fn;
1212 sv_magic(sv, 0, PERL_MAGIC_uvar, (char*)&uf, sizeof(uf));
1214 Attaching C<PERL_MAGIC_uvar> to arrays is permissible but has no effect.
1216 For hashes there is a specialized hook that gives control over hash
1217 keys (but not values). This hook calls C<PERL_MAGIC_uvar> 'get' magic
1218 if the "set" function in the C<ufuncs> structure is NULL. The hook
1219 is activated whenever the hash is accessed with a key specified as
1220 an C<SV> through the functions C<hv_store_ent>, C<hv_fetch_ent>,
1221 C<hv_delete_ent>, and C<hv_exists_ent>. Accessing the key as a string
1222 through the functions without the C<..._ent> suffix circumvents the
1223 hook. See L<Hash::Util::FieldHash/GUTS> for a detailed description.
1225 Note that because multiple extensions may be using C<PERL_MAGIC_ext>
1226 or C<PERL_MAGIC_uvar> magic, it is important for extensions to take
1227 extra care to avoid conflict. Typically only using the magic on
1228 objects blessed into the same class as the extension is sufficient.
1229 For C<PERL_MAGIC_ext> magic, it is usually a good idea to define an
1230 C<MGVTBL>, even if all its fields will be C<0>, so that individual
1231 C<MAGIC> pointers can be identified as a particular kind of magic
1232 using their magic virtual table. C<mg_findext> provides an easy way
1235 STATIC MGVTBL my_vtbl = { 0, 0, 0, 0, 0, 0, 0, 0 };
1238 if ((mg = mg_findext(sv, PERL_MAGIC_ext, &my_vtbl))) {
1239 /* this is really ours, not another module's PERL_MAGIC_ext */
1240 my_priv_data_t *priv = (my_priv_data_t *)mg->mg_ptr;
1244 Also note that the C<sv_set*()> and C<sv_cat*()> functions described
1245 earlier do B<not> invoke 'set' magic on their targets. This must
1246 be done by the user either by calling the C<SvSETMAGIC()> macro after
1247 calling these functions, or by using one of the C<sv_set*_mg()> or
1248 C<sv_cat*_mg()> functions. Similarly, generic C code must call the
1249 C<SvGETMAGIC()> macro to invoke any 'get' magic if they use an SV
1250 obtained from external sources in functions that don't handle magic.
1251 See L<perlapi> for a description of these functions.
1252 For example, calls to the C<sv_cat*()> functions typically need to be
1253 followed by C<SvSETMAGIC()>, but they don't need a prior C<SvGETMAGIC()>
1254 since their implementation handles 'get' magic.
1256 =head2 Finding Magic
1258 MAGIC *mg_find(SV *sv, int type); /* Finds the magic pointer of that
1261 This routine returns a pointer to a C<MAGIC> structure stored in the SV.
1262 If the SV does not have that magical
1263 feature, C<NULL> is returned. If the
1264 SV has multiple instances of that magical feature, the first one will be
1265 returned. C<mg_findext> can be used
1266 to find a C<MAGIC> structure of an SV
1267 based on both its magic type and its magic virtual table:
1269 MAGIC *mg_findext(SV *sv, int type, MGVTBL *vtbl);
1271 Also, if the SV passed to C<mg_find> or C<mg_findext> is not of type
1272 SVt_PVMG, Perl may core dump.
1274 int mg_copy(SV* sv, SV* nsv, const char* key, STRLEN klen);
1276 This routine checks to see what types of magic C<sv> has. If the mg_type
1277 field is an uppercase letter, then the mg_obj is copied to C<nsv>, but
1278 the mg_type field is changed to be the lowercase letter.
1280 =head2 Understanding the Magic of Tied Hashes and Arrays
1282 Tied hashes and arrays are magical beasts of the C<PERL_MAGIC_tied>
1285 WARNING: As of the 5.004 release, proper usage of the array and hash
1286 access functions requires understanding a few caveats. Some
1287 of these caveats are actually considered bugs in the API, to be fixed
1288 in later releases, and are bracketed with [MAYCHANGE] below. If
1289 you find yourself actually applying such information in this section, be
1290 aware that the behavior may change in the future, umm, without warning.
1292 The perl tie function associates a variable with an object that implements
1293 the various GET, SET, etc methods. To perform the equivalent of the perl
1294 tie function from an XSUB, you must mimic this behaviour. The code below
1295 carries out the necessary steps - firstly it creates a new hash, and then
1296 creates a second hash which it blesses into the class which will implement
1297 the tie methods. Lastly it ties the two hashes together, and returns a
1298 reference to the new tied hash. Note that the code below does NOT call the
1299 TIEHASH method in the MyTie class -
1300 see L<Calling Perl Routines from within C Programs> for details on how
1311 tie = newRV_noinc((SV*)newHV());
1312 stash = gv_stashpv("MyTie", GV_ADD);
1313 sv_bless(tie, stash);
1314 hv_magic(hash, (GV*)tie, PERL_MAGIC_tied);
1315 RETVAL = newRV_noinc(hash);
1319 The C<av_store> function, when given a tied array argument, merely
1320 copies the magic of the array onto the value to be "stored", using
1321 C<mg_copy>. It may also return NULL, indicating that the value did not
1322 actually need to be stored in the array. [MAYCHANGE] After a call to
1323 C<av_store> on a tied array, the caller will usually need to call
1324 C<mg_set(val)> to actually invoke the perl level "STORE" method on the
1325 TIEARRAY object. If C<av_store> did return NULL, a call to
1326 C<SvREFCNT_dec(val)> will also be usually necessary to avoid a memory
1329 The previous paragraph is applicable verbatim to tied hash access using the
1330 C<hv_store> and C<hv_store_ent> functions as well.
1332 C<av_fetch> and the corresponding hash functions C<hv_fetch> and
1333 C<hv_fetch_ent> actually return an undefined mortal value whose magic
1334 has been initialized using C<mg_copy>. Note the value so returned does not
1335 need to be deallocated, as it is already mortal. [MAYCHANGE] But you will
1336 need to call C<mg_get()> on the returned value in order to actually invoke
1337 the perl level "FETCH" method on the underlying TIE object. Similarly,
1338 you may also call C<mg_set()> on the return value after possibly assigning
1339 a suitable value to it using C<sv_setsv>, which will invoke the "STORE"
1340 method on the TIE object. [/MAYCHANGE]
1343 In other words, the array or hash fetch/store functions don't really
1344 fetch and store actual values in the case of tied arrays and hashes. They
1345 merely call C<mg_copy> to attach magic to the values that were meant to be
1346 "stored" or "fetched". Later calls to C<mg_get> and C<mg_set> actually
1347 do the job of invoking the TIE methods on the underlying objects. Thus
1348 the magic mechanism currently implements a kind of lazy access to arrays
1351 Currently (as of perl version 5.004), use of the hash and array access
1352 functions requires the user to be aware of whether they are operating on
1353 "normal" hashes and arrays, or on their tied variants. The API may be
1354 changed to provide more transparent access to both tied and normal data
1355 types in future versions.
1358 You would do well to understand that the TIEARRAY and TIEHASH interfaces
1359 are mere sugar to invoke some perl method calls while using the uniform hash
1360 and array syntax. The use of this sugar imposes some overhead (typically
1361 about two to four extra opcodes per FETCH/STORE operation, in addition to
1362 the creation of all the mortal variables required to invoke the methods).
1363 This overhead will be comparatively small if the TIE methods are themselves
1364 substantial, but if they are only a few statements long, the overhead
1365 will not be insignificant.
1367 =head2 Localizing changes
1369 Perl has a very handy construction
1376 This construction is I<approximately> equivalent to
1385 The biggest difference is that the first construction would
1386 reinstate the initial value of $var, irrespective of how control exits
1387 the block: C<goto>, C<return>, C<die>/C<eval>, etc. It is a little bit
1388 more efficient as well.
1390 There is a way to achieve a similar task from C via Perl API: create a
1391 I<pseudo-block>, and arrange for some changes to be automatically
1392 undone at the end of it, either explicit, or via a non-local exit (via
1393 die()). A I<block>-like construct is created by a pair of
1394 C<ENTER>/C<LEAVE> macros (see L<perlcall/"Returning a Scalar">).
1395 Such a construct may be created specially for some important localized
1396 task, or an existing one (like boundaries of enclosing Perl
1397 subroutine/block, or an existing pair for freeing TMPs) may be
1398 used. (In the second case the overhead of additional localization must
1399 be almost negligible.) Note that any XSUB is automatically enclosed in
1400 an C<ENTER>/C<LEAVE> pair.
1402 Inside such a I<pseudo-block> the following service is available:
1406 =item C<SAVEINT(int i)>
1408 =item C<SAVEIV(IV i)>
1410 =item C<SAVEI32(I32 i)>
1412 =item C<SAVELONG(long i)>
1414 These macros arrange things to restore the value of integer variable
1415 C<i> at the end of enclosing I<pseudo-block>.
1417 =item C<SAVESPTR(s)>
1419 =item C<SAVEPPTR(p)>
1421 These macros arrange things to restore the value of pointers C<s> and
1422 C<p>. C<s> must be a pointer of a type which survives conversion to
1423 C<SV*> and back, C<p> should be able to survive conversion to C<char*>
1426 =item C<SAVEFREESV(SV *sv)>
1428 The refcount of C<sv> would be decremented at the end of
1429 I<pseudo-block>. This is similar to C<sv_2mortal> in that it is also a
1430 mechanism for doing a delayed C<SvREFCNT_dec>. However, while C<sv_2mortal>
1431 extends the lifetime of C<sv> until the beginning of the next statement,
1432 C<SAVEFREESV> extends it until the end of the enclosing scope. These
1433 lifetimes can be wildly different.
1435 Also compare C<SAVEMORTALIZESV>.
1437 =item C<SAVEMORTALIZESV(SV *sv)>
1439 Just like C<SAVEFREESV>, but mortalizes C<sv> at the end of the current
1440 scope instead of decrementing its reference count. This usually has the
1441 effect of keeping C<sv> alive until the statement that called the currently
1442 live scope has finished executing.
1444 =item C<SAVEFREEOP(OP *op)>
1446 The C<OP *> is op_free()ed at the end of I<pseudo-block>.
1448 =item C<SAVEFREEPV(p)>
1450 The chunk of memory which is pointed to by C<p> is Safefree()ed at the
1451 end of I<pseudo-block>.
1453 =item C<SAVECLEARSV(SV *sv)>
1455 Clears a slot in the current scratchpad which corresponds to C<sv> at
1456 the end of I<pseudo-block>.
1458 =item C<SAVEDELETE(HV *hv, char *key, I32 length)>
1460 The key C<key> of C<hv> is deleted at the end of I<pseudo-block>. The
1461 string pointed to by C<key> is Safefree()ed. If one has a I<key> in
1462 short-lived storage, the corresponding string may be reallocated like
1465 SAVEDELETE(PL_defstash, savepv(tmpbuf), strlen(tmpbuf));
1467 =item C<SAVEDESTRUCTOR(DESTRUCTORFUNC_NOCONTEXT_t f, void *p)>
1469 At the end of I<pseudo-block> the function C<f> is called with the
1472 =item C<SAVEDESTRUCTOR_X(DESTRUCTORFUNC_t f, void *p)>
1474 At the end of I<pseudo-block> the function C<f> is called with the
1475 implicit context argument (if any), and C<p>.
1477 =item C<SAVESTACK_POS()>
1479 The current offset on the Perl internal stack (cf. C<SP>) is restored
1480 at the end of I<pseudo-block>.
1484 The following API list contains functions, thus one needs to
1485 provide pointers to the modifiable data explicitly (either C pointers,
1486 or Perlish C<GV *>s). Where the above macros take C<int>, a similar
1487 function takes C<int *>.
1491 =item C<SV* save_scalar(GV *gv)>
1493 Equivalent to Perl code C<local $gv>.
1495 =item C<AV* save_ary(GV *gv)>
1497 =item C<HV* save_hash(GV *gv)>
1499 Similar to C<save_scalar>, but localize C<@gv> and C<%gv>.
1501 =item C<void save_item(SV *item)>
1503 Duplicates the current value of C<SV>, on the exit from the current
1504 C<ENTER>/C<LEAVE> I<pseudo-block> will restore the value of C<SV>
1505 using the stored value. It doesn't handle magic. Use C<save_scalar> if
1508 =item C<void save_list(SV **sarg, I32 maxsarg)>
1510 A variant of C<save_item> which takes multiple arguments via an array
1511 C<sarg> of C<SV*> of length C<maxsarg>.
1513 =item C<SV* save_svref(SV **sptr)>
1515 Similar to C<save_scalar>, but will reinstate an C<SV *>.
1517 =item C<void save_aptr(AV **aptr)>
1519 =item C<void save_hptr(HV **hptr)>
1521 Similar to C<save_svref>, but localize C<AV *> and C<HV *>.
1525 The C<Alias> module implements localization of the basic types within the
1526 I<caller's scope>. People who are interested in how to localize things in
1527 the containing scope should take a look there too.
1531 =head2 XSUBs and the Argument Stack
1533 The XSUB mechanism is a simple way for Perl programs to access C subroutines.
1534 An XSUB routine will have a stack that contains the arguments from the Perl
1535 program, and a way to map from the Perl data structures to a C equivalent.
1537 The stack arguments are accessible through the C<ST(n)> macro, which returns
1538 the C<n>'th stack argument. Argument 0 is the first argument passed in the
1539 Perl subroutine call. These arguments are C<SV*>, and can be used anywhere
1542 Most of the time, output from the C routine can be handled through use of
1543 the RETVAL and OUTPUT directives. However, there are some cases where the
1544 argument stack is not already long enough to handle all the return values.
1545 An example is the POSIX tzname() call, which takes no arguments, but returns
1546 two, the local time zone's standard and summer time abbreviations.
1548 To handle this situation, the PPCODE directive is used and the stack is
1549 extended using the macro:
1553 where C<SP> is the macro that represents the local copy of the stack pointer,
1554 and C<num> is the number of elements the stack should be extended by.
1556 Now that there is room on the stack, values can be pushed on it using C<PUSHs>
1557 macro. The pushed values will often need to be "mortal" (See
1558 L</Reference Counts and Mortality>):
1560 PUSHs(sv_2mortal(newSViv(an_integer)))
1561 PUSHs(sv_2mortal(newSVuv(an_unsigned_integer)))
1562 PUSHs(sv_2mortal(newSVnv(a_double)))
1563 PUSHs(sv_2mortal(newSVpv("Some String",0)))
1564 /* Although the last example is better written as the more
1566 PUSHs(newSVpvs_flags("Some String", SVs_TEMP))
1568 And now the Perl program calling C<tzname>, the two values will be assigned
1571 ($standard_abbrev, $summer_abbrev) = POSIX::tzname;
1573 An alternate (and possibly simpler) method to pushing values on the stack is
1578 This macro automatically adjusts the stack for you, if needed. Thus, you
1579 do not need to call C<EXTEND> to extend the stack.
1581 Despite their suggestions in earlier versions of this document the macros
1582 C<(X)PUSH[iunp]> are I<not> suited to XSUBs which return multiple results.
1583 For that, either stick to the C<(X)PUSHs> macros shown above, or use the new
1584 C<m(X)PUSH[iunp]> macros instead; see L</Putting a C value on Perl stack>.
1586 For more information, consult L<perlxs> and L<perlxstut>.
1588 =head2 Autoloading with XSUBs
1590 If an AUTOLOAD routine is an XSUB, as with Perl subroutines, Perl puts the
1591 fully-qualified name of the autoloaded subroutine in the $AUTOLOAD variable
1592 of the XSUB's package.
1594 But it also puts the same information in certain fields of the XSUB itself:
1596 HV *stash = CvSTASH(cv);
1597 const char *subname = SvPVX(cv);
1598 STRLEN name_length = SvCUR(cv); /* in bytes */
1599 U32 is_utf8 = SvUTF8(cv);
1601 C<SvPVX(cv)> contains just the sub name itself, not including the package.
1602 For an AUTOLOAD routine in UNIVERSAL or one of its superclasses,
1603 C<CvSTASH(cv)> returns NULL during a method call on a nonexistent package.
1605 B<Note>: Setting $AUTOLOAD stopped working in 5.6.1, which did not support
1606 XS AUTOLOAD subs at all. Perl 5.8.0 introduced the use of fields in the
1607 XSUB itself. Perl 5.16.0 restored the setting of $AUTOLOAD. If you need
1608 to support 5.8-5.14, use the XSUB's fields.
1610 =head2 Calling Perl Routines from within C Programs
1612 There are four routines that can be used to call a Perl subroutine from
1613 within a C program. These four are:
1615 I32 call_sv(SV*, I32);
1616 I32 call_pv(const char*, I32);
1617 I32 call_method(const char*, I32);
1618 I32 call_argv(const char*, I32, char**);
1620 The routine most often used is C<call_sv>. The C<SV*> argument
1621 contains either the name of the Perl subroutine to be called, or a
1622 reference to the subroutine. The second argument consists of flags
1623 that control the context in which the subroutine is called, whether
1624 or not the subroutine is being passed arguments, how errors should be
1625 trapped, and how to treat return values.
1627 All four routines return the number of arguments that the subroutine returned
1630 These routines used to be called C<perl_call_sv>, etc., before Perl v5.6.0,
1631 but those names are now deprecated; macros of the same name are provided for
1634 When using any of these routines (except C<call_argv>), the programmer
1635 must manipulate the Perl stack. These include the following macros and
1650 For a detailed description of calling conventions from C to Perl,
1651 consult L<perlcall>.
1653 =head2 Putting a C value on Perl stack
1655 A lot of opcodes (this is an elementary operation in the internal perl
1656 stack machine) put an SV* on the stack. However, as an optimization
1657 the corresponding SV is (usually) not recreated each time. The opcodes
1658 reuse specially assigned SVs (I<target>s) which are (as a corollary)
1659 not constantly freed/created.
1661 Each of the targets is created only once (but see
1662 L<Scratchpads and recursion> below), and when an opcode needs to put
1663 an integer, a double, or a string on stack, it just sets the
1664 corresponding parts of its I<target> and puts the I<target> on stack.
1666 The macro to put this target on stack is C<PUSHTARG>, and it is
1667 directly used in some opcodes, as well as indirectly in zillions of
1668 others, which use it via C<(X)PUSH[iunp]>.
1670 Because the target is reused, you must be careful when pushing multiple
1671 values on the stack. The following code will not do what you think:
1676 This translates as "set C<TARG> to 10, push a pointer to C<TARG> onto
1677 the stack; set C<TARG> to 20, push a pointer to C<TARG> onto the stack".
1678 At the end of the operation, the stack does not contain the values 10
1679 and 20, but actually contains two pointers to C<TARG>, which we have set
1682 If you need to push multiple different values then you should either use
1683 the C<(X)PUSHs> macros, or else use the new C<m(X)PUSH[iunp]> macros,
1684 none of which make use of C<TARG>. The C<(X)PUSHs> macros simply push an
1685 SV* on the stack, which, as noted under L</XSUBs and the Argument Stack>,
1686 will often need to be "mortal". The new C<m(X)PUSH[iunp]> macros make
1687 this a little easier to achieve by creating a new mortal for you (via
1688 C<(X)PUSHmortal>), pushing that onto the stack (extending it if necessary
1689 in the case of the C<mXPUSH[iunp]> macros), and then setting its value.
1690 Thus, instead of writing this to "fix" the example above:
1692 XPUSHs(sv_2mortal(newSViv(10)))
1693 XPUSHs(sv_2mortal(newSViv(20)))
1695 you can simply write:
1700 On a related note, if you do use C<(X)PUSH[iunp]>, then you're going to
1701 need a C<dTARG> in your variable declarations so that the C<*PUSH*>
1702 macros can make use of the local variable C<TARG>. See also C<dTARGET>
1707 The question remains on when the SVs which are I<target>s for opcodes
1708 are created. The answer is that they are created when the current
1709 unit--a subroutine or a file (for opcodes for statements outside of
1710 subroutines)--is compiled. During this time a special anonymous Perl
1711 array is created, which is called a scratchpad for the current unit.
1713 A scratchpad keeps SVs which are lexicals for the current unit and are
1714 targets for opcodes. A previous version of this document
1715 stated that one can deduce that an SV lives on a scratchpad
1716 by looking on its flags: lexicals have C<SVs_PADMY> set, and
1717 I<target>s have C<SVs_PADTMP> set. But this have never been fully true.
1718 C<SVs_PADMY> could be set on a variable that no longer resides in any pad.
1719 While I<target>s do have C<SVs_PADTMP> set, it can also be set on variables
1720 that have never resided in a pad, but nonetheless act like I<target>s.
1722 The correspondence between OPs and I<target>s is not 1-to-1. Different
1723 OPs in the compile tree of the unit can use the same target, if this
1724 would not conflict with the expected life of the temporary.
1726 =head2 Scratchpads and recursion
1728 In fact it is not 100% true that a compiled unit contains a pointer to
1729 the scratchpad AV. In fact it contains a pointer to an AV of
1730 (initially) one element, and this element is the scratchpad AV. Why do
1731 we need an extra level of indirection?
1733 The answer is B<recursion>, and maybe B<threads>. Both
1734 these can create several execution pointers going into the same
1735 subroutine. For the subroutine-child not write over the temporaries
1736 for the subroutine-parent (lifespan of which covers the call to the
1737 child), the parent and the child should have different
1738 scratchpads. (I<And> the lexicals should be separate anyway!)
1740 So each subroutine is born with an array of scratchpads (of length 1).
1741 On each entry to the subroutine it is checked that the current
1742 depth of the recursion is not more than the length of this array, and
1743 if it is, new scratchpad is created and pushed into the array.
1745 The I<target>s on this scratchpad are C<undef>s, but they are already
1746 marked with correct flags.
1748 =head1 Memory Allocation
1752 All memory meant to be used with the Perl API functions should be manipulated
1753 using the macros described in this section. The macros provide the necessary
1754 transparency between differences in the actual malloc implementation that is
1757 It is suggested that you enable the version of malloc that is distributed
1758 with Perl. It keeps pools of various sizes of unallocated memory in
1759 order to satisfy allocation requests more quickly. However, on some
1760 platforms, it may cause spurious malloc or free errors.
1762 The following three macros are used to initially allocate memory :
1764 Newx(pointer, number, type);
1765 Newxc(pointer, number, type, cast);
1766 Newxz(pointer, number, type);
1768 The first argument C<pointer> should be the name of a variable that will
1769 point to the newly allocated memory.
1771 The second and third arguments C<number> and C<type> specify how many of
1772 the specified type of data structure should be allocated. The argument
1773 C<type> is passed to C<sizeof>. The final argument to C<Newxc>, C<cast>,
1774 should be used if the C<pointer> argument is different from the C<type>
1777 Unlike the C<Newx> and C<Newxc> macros, the C<Newxz> macro calls C<memzero>
1778 to zero out all the newly allocated memory.
1782 Renew(pointer, number, type);
1783 Renewc(pointer, number, type, cast);
1786 These three macros are used to change a memory buffer size or to free a
1787 piece of memory no longer needed. The arguments to C<Renew> and C<Renewc>
1788 match those of C<New> and C<Newc> with the exception of not needing the
1789 "magic cookie" argument.
1793 Move(source, dest, number, type);
1794 Copy(source, dest, number, type);
1795 Zero(dest, number, type);
1797 These three macros are used to move, copy, or zero out previously allocated
1798 memory. The C<source> and C<dest> arguments point to the source and
1799 destination starting points. Perl will move, copy, or zero out C<number>
1800 instances of the size of the C<type> data structure (using the C<sizeof>
1805 The most recent development releases of Perl have been experimenting with
1806 removing Perl's dependency on the "normal" standard I/O suite and allowing
1807 other stdio implementations to be used. This involves creating a new
1808 abstraction layer that then calls whichever implementation of stdio Perl
1809 was compiled with. All XSUBs should now use the functions in the PerlIO
1810 abstraction layer and not make any assumptions about what kind of stdio
1813 For a complete description of the PerlIO abstraction, consult L<perlapio>.
1815 =head1 Compiled code
1819 Here we describe the internal form your code is converted to by
1820 Perl. Start with a simple example:
1824 This is converted to a tree similar to this one:
1832 (but slightly more complicated). This tree reflects the way Perl
1833 parsed your code, but has nothing to do with the execution order.
1834 There is an additional "thread" going through the nodes of the tree
1835 which shows the order of execution of the nodes. In our simplified
1836 example above it looks like:
1838 $b ---> $c ---> + ---> $a ---> assign-to
1840 But with the actual compile tree for C<$a = $b + $c> it is different:
1841 some nodes I<optimized away>. As a corollary, though the actual tree
1842 contains more nodes than our simplified example, the execution order
1843 is the same as in our example.
1845 =head2 Examining the tree
1847 If you have your perl compiled for debugging (usually done with
1848 C<-DDEBUGGING> on the C<Configure> command line), you may examine the
1849 compiled tree by specifying C<-Dx> on the Perl command line. The
1850 output takes several lines per node, and for C<$b+$c> it looks like
1855 FLAGS = (SCALAR,KIDS)
1857 TYPE = null ===> (4)
1859 FLAGS = (SCALAR,KIDS)
1861 3 TYPE = gvsv ===> 4
1867 TYPE = null ===> (5)
1869 FLAGS = (SCALAR,KIDS)
1871 4 TYPE = gvsv ===> 5
1877 This tree has 5 nodes (one per C<TYPE> specifier), only 3 of them are
1878 not optimized away (one per number in the left column). The immediate
1879 children of the given node correspond to C<{}> pairs on the same level
1880 of indentation, thus this listing corresponds to the tree:
1888 The execution order is indicated by C<===E<gt>> marks, thus it is C<3
1889 4 5 6> (node C<6> is not included into above listing), i.e.,
1890 C<gvsv gvsv add whatever>.
1892 Each of these nodes represents an op, a fundamental operation inside the
1893 Perl core. The code which implements each operation can be found in the
1894 F<pp*.c> files; the function which implements the op with type C<gvsv>
1895 is C<pp_gvsv>, and so on. As the tree above shows, different ops have
1896 different numbers of children: C<add> is a binary operator, as one would
1897 expect, and so has two children. To accommodate the various different
1898 numbers of children, there are various types of op data structure, and
1899 they link together in different ways.
1901 The simplest type of op structure is C<OP>: this has no children. Unary
1902 operators, C<UNOP>s, have one child, and this is pointed to by the
1903 C<op_first> field. Binary operators (C<BINOP>s) have not only an
1904 C<op_first> field but also an C<op_last> field. The most complex type of
1905 op is a C<LISTOP>, which has any number of children. In this case, the
1906 first child is pointed to by C<op_first> and the last child by
1907 C<op_last>. The children in between can be found by iteratively
1908 following the C<op_sibling> pointer from the first child to the last.
1910 There are also two other op types: a C<PMOP> holds a regular expression,
1911 and has no children, and a C<LOOP> may or may not have children. If the
1912 C<op_children> field is non-zero, it behaves like a C<LISTOP>. To
1913 complicate matters, if a C<UNOP> is actually a C<null> op after
1914 optimization (see L</Compile pass 2: context propagation>) it will still
1915 have children in accordance with its former type.
1917 Another way to examine the tree is to use a compiler back-end module, such
1920 =head2 Compile pass 1: check routines
1922 The tree is created by the compiler while I<yacc> code feeds it
1923 the constructions it recognizes. Since I<yacc> works bottom-up, so does
1924 the first pass of perl compilation.
1926 What makes this pass interesting for perl developers is that some
1927 optimization may be performed on this pass. This is optimization by
1928 so-called "check routines". The correspondence between node names
1929 and corresponding check routines is described in F<opcode.pl> (do not
1930 forget to run C<make regen_headers> if you modify this file).
1932 A check routine is called when the node is fully constructed except
1933 for the execution-order thread. Since at this time there are no
1934 back-links to the currently constructed node, one can do most any
1935 operation to the top-level node, including freeing it and/or creating
1936 new nodes above/below it.
1938 The check routine returns the node which should be inserted into the
1939 tree (if the top-level node was not modified, check routine returns
1942 By convention, check routines have names C<ck_*>. They are usually
1943 called from C<new*OP> subroutines (or C<convert>) (which in turn are
1944 called from F<perly.y>).
1946 =head2 Compile pass 1a: constant folding
1948 Immediately after the check routine is called the returned node is
1949 checked for being compile-time executable. If it is (the value is
1950 judged to be constant) it is immediately executed, and a I<constant>
1951 node with the "return value" of the corresponding subtree is
1952 substituted instead. The subtree is deleted.
1954 If constant folding was not performed, the execution-order thread is
1957 =head2 Compile pass 2: context propagation
1959 When a context for a part of compile tree is known, it is propagated
1960 down through the tree. At this time the context can have 5 values
1961 (instead of 2 for runtime context): void, boolean, scalar, list, and
1962 lvalue. In contrast with the pass 1 this pass is processed from top
1963 to bottom: a node's context determines the context for its children.
1965 Additional context-dependent optimizations are performed at this time.
1966 Since at this moment the compile tree contains back-references (via
1967 "thread" pointers), nodes cannot be free()d now. To allow
1968 optimized-away nodes at this stage, such nodes are null()ified instead
1969 of free()ing (i.e. their type is changed to OP_NULL).
1971 =head2 Compile pass 3: peephole optimization
1973 After the compile tree for a subroutine (or for an C<eval> or a file)
1974 is created, an additional pass over the code is performed. This pass
1975 is neither top-down or bottom-up, but in the execution order (with
1976 additional complications for conditionals). Optimizations performed
1977 at this stage are subject to the same restrictions as in the pass 2.
1979 Peephole optimizations are done by calling the function pointed to
1980 by the global variable C<PL_peepp>. By default, C<PL_peepp> just
1981 calls the function pointed to by the global variable C<PL_rpeepp>.
1982 By default, that performs some basic op fixups and optimisations along
1983 the execution-order op chain, and recursively calls C<PL_rpeepp> for
1984 each side chain of ops (resulting from conditionals). Extensions may
1985 provide additional optimisations or fixups, hooking into either the
1986 per-subroutine or recursive stage, like this:
1988 static peep_t prev_peepp;
1989 static void my_peep(pTHX_ OP *o)
1991 /* custom per-subroutine optimisation goes here */
1992 prev_peepp(aTHX_ o);
1993 /* custom per-subroutine optimisation may also go here */
1996 prev_peepp = PL_peepp;
1999 static peep_t prev_rpeepp;
2000 static void my_rpeep(pTHX_ OP *o)
2003 for(; o; o = o->op_next) {
2004 /* custom per-op optimisation goes here */
2006 prev_rpeepp(aTHX_ orig_o);
2009 prev_rpeepp = PL_rpeepp;
2010 PL_rpeepp = my_rpeep;
2012 =head2 Pluggable runops
2014 The compile tree is executed in a runops function. There are two runops
2015 functions, in F<run.c> and in F<dump.c>. C<Perl_runops_debug> is used
2016 with DEBUGGING and C<Perl_runops_standard> is used otherwise. For fine
2017 control over the execution of the compile tree it is possible to provide
2018 your own runops function.
2020 It's probably best to copy one of the existing runops functions and
2021 change it to suit your needs. Then, in the BOOT section of your XS
2024 PL_runops = my_runops;
2026 This function should be as efficient as possible to keep your programs
2027 running as fast as possible.
2029 =head2 Compile-time scope hooks
2031 As of perl 5.14 it is possible to hook into the compile-time lexical
2032 scope mechanism using C<Perl_blockhook_register>. This is used like
2035 STATIC void my_start_hook(pTHX_ int full);
2036 STATIC BHK my_hooks;
2039 BhkENTRY_set(&my_hooks, bhk_start, my_start_hook);
2040 Perl_blockhook_register(aTHX_ &my_hooks);
2042 This will arrange to have C<my_start_hook> called at the start of
2043 compiling every lexical scope. The available hooks are:
2047 =item C<void bhk_start(pTHX_ int full)>
2049 This is called just after starting a new lexical scope. Note that Perl
2054 creates two scopes: the first starts at the C<(> and has C<full == 1>,
2055 the second starts at the C<{> and has C<full == 0>. Both end at the
2056 C<}>, so calls to C<start> and C<pre/post_end> will match. Anything
2057 pushed onto the save stack by this hook will be popped just before the
2058 scope ends (between the C<pre_> and C<post_end> hooks, in fact).
2060 =item C<void bhk_pre_end(pTHX_ OP **o)>
2062 This is called at the end of a lexical scope, just before unwinding the
2063 stack. I<o> is the root of the optree representing the scope; it is a
2064 double pointer so you can replace the OP if you need to.
2066 =item C<void bhk_post_end(pTHX_ OP **o)>
2068 This is called at the end of a lexical scope, just after unwinding the
2069 stack. I<o> is as above. Note that it is possible for calls to C<pre_>
2070 and C<post_end> to nest, if there is something on the save stack that
2073 =item C<void bhk_eval(pTHX_ OP *const o)>
2075 This is called just before starting to compile an C<eval STRING>, C<do
2076 FILE>, C<require> or C<use>, after the eval has been set up. I<o> is the
2077 OP that requested the eval, and will normally be an C<OP_ENTEREVAL>,
2078 C<OP_DOFILE> or C<OP_REQUIRE>.
2082 Once you have your hook functions, you need a C<BHK> structure to put
2083 them in. It's best to allocate it statically, since there is no way to
2084 free it once it's registered. The function pointers should be inserted
2085 into this structure using the C<BhkENTRY_set> macro, which will also set
2086 flags indicating which entries are valid. If you do need to allocate
2087 your C<BHK> dynamically for some reason, be sure to zero it before you
2090 Once registered, there is no mechanism to switch these hooks off, so if
2091 that is necessary you will need to do this yourself. An entry in C<%^H>
2092 is probably the best way, so the effect is lexically scoped; however it
2093 is also possible to use the C<BhkDISABLE> and C<BhkENABLE> macros to
2094 temporarily switch entries on and off. You should also be aware that
2095 generally speaking at least one scope will have opened before your
2096 extension is loaded, so you will see some C<pre/post_end> pairs that
2097 didn't have a matching C<start>.
2099 =head1 Examining internal data structures with the C<dump> functions
2101 To aid debugging, the source file F<dump.c> contains a number of
2102 functions which produce formatted output of internal data structures.
2104 The most commonly used of these functions is C<Perl_sv_dump>; it's used
2105 for dumping SVs, AVs, HVs, and CVs. The C<Devel::Peek> module calls
2106 C<sv_dump> to produce debugging output from Perl-space, so users of that
2107 module should already be familiar with its format.
2109 C<Perl_op_dump> can be used to dump an C<OP> structure or any of its
2110 derivatives, and produces output similar to C<perl -Dx>; in fact,
2111 C<Perl_dump_eval> will dump the main root of the code being evaluated,
2112 exactly like C<-Dx>.
2114 Other useful functions are C<Perl_dump_sub>, which turns a C<GV> into an
2115 op tree, C<Perl_dump_packsubs> which calls C<Perl_dump_sub> on all the
2116 subroutines in a package like so: (Thankfully, these are all xsubs, so
2117 there is no op tree)
2119 (gdb) print Perl_dump_packsubs(PL_defstash)
2121 SUB attributes::bootstrap = (xsub 0x811fedc 0)
2123 SUB UNIVERSAL::can = (xsub 0x811f50c 0)
2125 SUB UNIVERSAL::isa = (xsub 0x811f304 0)
2127 SUB UNIVERSAL::VERSION = (xsub 0x811f7ac 0)
2129 SUB DynaLoader::boot_DynaLoader = (xsub 0x805b188 0)
2131 and C<Perl_dump_all>, which dumps all the subroutines in the stash and
2132 the op tree of the main root.
2134 =head1 How multiple interpreters and concurrency are supported
2136 =head2 Background and PERL_IMPLICIT_CONTEXT
2138 The Perl interpreter can be regarded as a closed box: it has an API
2139 for feeding it code or otherwise making it do things, but it also has
2140 functions for its own use. This smells a lot like an object, and
2141 there are ways for you to build Perl so that you can have multiple
2142 interpreters, with one interpreter represented either as a C structure,
2143 or inside a thread-specific structure. These structures contain all
2144 the context, the state of that interpreter.
2146 One macro controls the major Perl build flavor: MULTIPLICITY. The
2147 MULTIPLICITY build has a C structure that packages all the interpreter
2148 state. With multiplicity-enabled perls, PERL_IMPLICIT_CONTEXT is also
2149 normally defined, and enables the support for passing in a "hidden" first
2150 argument that represents all three data structures. MULTIPLICITY makes
2151 multi-threaded perls possible (with the ithreads threading model, related
2152 to the macro USE_ITHREADS.)
2154 Two other "encapsulation" macros are the PERL_GLOBAL_STRUCT and
2155 PERL_GLOBAL_STRUCT_PRIVATE (the latter turns on the former, and the
2156 former turns on MULTIPLICITY.) The PERL_GLOBAL_STRUCT causes all the
2157 internal variables of Perl to be wrapped inside a single global struct,
2158 struct perl_vars, accessible as (globals) &PL_Vars or PL_VarsPtr or
2159 the function Perl_GetVars(). The PERL_GLOBAL_STRUCT_PRIVATE goes
2160 one step further, there is still a single struct (allocated in main()
2161 either from heap or from stack) but there are no global data symbols
2162 pointing to it. In either case the global struct should be initialized
2163 as the very first thing in main() using Perl_init_global_struct() and
2164 correspondingly tear it down after perl_free() using Perl_free_global_struct(),
2165 please see F<miniperlmain.c> for usage details. You may also need
2166 to use C<dVAR> in your coding to "declare the global variables"
2167 when you are using them. dTHX does this for you automatically.
2169 To see whether you have non-const data you can use a BSD-compatible C<nm>:
2171 nm libperl.a | grep -v ' [TURtr] '
2173 If this displays any C<D> or C<d> symbols, you have non-const data.
2175 For backward compatibility reasons defining just PERL_GLOBAL_STRUCT
2176 doesn't actually hide all symbols inside a big global struct: some
2177 PerlIO_xxx vtables are left visible. The PERL_GLOBAL_STRUCT_PRIVATE
2178 then hides everything (see how the PERLIO_FUNCS_DECL is used).
2180 All this obviously requires a way for the Perl internal functions to be
2181 either subroutines taking some kind of structure as the first
2182 argument, or subroutines taking nothing as the first argument. To
2183 enable these two very different ways of building the interpreter,
2184 the Perl source (as it does in so many other situations) makes heavy
2185 use of macros and subroutine naming conventions.
2187 First problem: deciding which functions will be public API functions and
2188 which will be private. All functions whose names begin C<S_> are private
2189 (think "S" for "secret" or "static"). All other functions begin with
2190 "Perl_", but just because a function begins with "Perl_" does not mean it is
2191 part of the API. (See L</Internal
2192 Functions>.) The easiest way to be B<sure> a
2193 function is part of the API is to find its entry in L<perlapi>.
2194 If it exists in L<perlapi>, it's part of the API. If it doesn't, and you
2195 think it should be (i.e., you need it for your extension), send mail via
2196 L<perlbug> explaining why you think it should be.
2198 Second problem: there must be a syntax so that the same subroutine
2199 declarations and calls can pass a structure as their first argument,
2200 or pass nothing. To solve this, the subroutines are named and
2201 declared in a particular way. Here's a typical start of a static
2202 function used within the Perl guts:
2205 S_incline(pTHX_ char *s)
2207 STATIC becomes "static" in C, and may be #define'd to nothing in some
2208 configurations in the future.
2210 A public function (i.e. part of the internal API, but not necessarily
2211 sanctioned for use in extensions) begins like this:
2214 Perl_sv_setiv(pTHX_ SV* dsv, IV num)
2216 C<pTHX_> is one of a number of macros (in F<perl.h>) that hide the
2217 details of the interpreter's context. THX stands for "thread", "this",
2218 or "thingy", as the case may be. (And no, George Lucas is not involved. :-)
2219 The first character could be 'p' for a B<p>rototype, 'a' for B<a>rgument,
2220 or 'd' for B<d>eclaration, so we have C<pTHX>, C<aTHX> and C<dTHX>, and
2223 When Perl is built without options that set PERL_IMPLICIT_CONTEXT, there is no
2224 first argument containing the interpreter's context. The trailing underscore
2225 in the pTHX_ macro indicates that the macro expansion needs a comma
2226 after the context argument because other arguments follow it. If
2227 PERL_IMPLICIT_CONTEXT is not defined, pTHX_ will be ignored, and the
2228 subroutine is not prototyped to take the extra argument. The form of the
2229 macro without the trailing underscore is used when there are no additional
2232 When a core function calls another, it must pass the context. This
2233 is normally hidden via macros. Consider C<sv_setiv>. It expands into
2234 something like this:
2236 #ifdef PERL_IMPLICIT_CONTEXT
2237 #define sv_setiv(a,b) Perl_sv_setiv(aTHX_ a, b)
2238 /* can't do this for vararg functions, see below */
2240 #define sv_setiv Perl_sv_setiv
2243 This works well, and means that XS authors can gleefully write:
2247 and still have it work under all the modes Perl could have been
2250 This doesn't work so cleanly for varargs functions, though, as macros
2251 imply that the number of arguments is known in advance. Instead we
2252 either need to spell them out fully, passing C<aTHX_> as the first
2253 argument (the Perl core tends to do this with functions like
2254 Perl_warner), or use a context-free version.
2256 The context-free version of Perl_warner is called
2257 Perl_warner_nocontext, and does not take the extra argument. Instead
2258 it does dTHX; to get the context from thread-local storage. We
2259 C<#define warner Perl_warner_nocontext> so that extensions get source
2260 compatibility at the expense of performance. (Passing an arg is
2261 cheaper than grabbing it from thread-local storage.)
2263 You can ignore [pad]THXx when browsing the Perl headers/sources.
2264 Those are strictly for use within the core. Extensions and embedders
2265 need only be aware of [pad]THX.
2267 =head2 So what happened to dTHR?
2269 C<dTHR> was introduced in perl 5.005 to support the older thread model.
2270 The older thread model now uses the C<THX> mechanism to pass context
2271 pointers around, so C<dTHR> is not useful any more. Perl 5.6.0 and
2272 later still have it for backward source compatibility, but it is defined
2275 =head2 How do I use all this in extensions?
2277 When Perl is built with PERL_IMPLICIT_CONTEXT, extensions that call
2278 any functions in the Perl API will need to pass the initial context
2279 argument somehow. The kicker is that you will need to write it in
2280 such a way that the extension still compiles when Perl hasn't been
2281 built with PERL_IMPLICIT_CONTEXT enabled.
2283 There are three ways to do this. First, the easy but inefficient way,
2284 which is also the default, in order to maintain source compatibility
2285 with extensions: whenever F<XSUB.h> is #included, it redefines the aTHX
2286 and aTHX_ macros to call a function that will return the context.
2287 Thus, something like:
2291 in your extension will translate to this when PERL_IMPLICIT_CONTEXT is
2294 Perl_sv_setiv(Perl_get_context(), sv, num);
2296 or to this otherwise:
2298 Perl_sv_setiv(sv, num);
2300 You don't have to do anything new in your extension to get this; since
2301 the Perl library provides Perl_get_context(), it will all just
2304 The second, more efficient way is to use the following template for
2307 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2312 STATIC void my_private_function(int arg1, int arg2);
2315 my_private_function(int arg1, int arg2)
2317 dTHX; /* fetch context */
2318 ... call many Perl API functions ...
2323 MODULE = Foo PACKAGE = Foo
2331 my_private_function(arg, 10);
2333 Note that the only two changes from the normal way of writing an
2334 extension is the addition of a C<#define PERL_NO_GET_CONTEXT> before
2335 including the Perl headers, followed by a C<dTHX;> declaration at
2336 the start of every function that will call the Perl API. (You'll
2337 know which functions need this, because the C compiler will complain
2338 that there's an undeclared identifier in those functions.) No changes
2339 are needed for the XSUBs themselves, because the XS() macro is
2340 correctly defined to pass in the implicit context if needed.
2342 The third, even more efficient way is to ape how it is done within
2346 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2351 /* pTHX_ only needed for functions that call Perl API */
2352 STATIC void my_private_function(pTHX_ int arg1, int arg2);
2355 my_private_function(pTHX_ int arg1, int arg2)
2357 /* dTHX; not needed here, because THX is an argument */
2358 ... call Perl API functions ...
2363 MODULE = Foo PACKAGE = Foo
2371 my_private_function(aTHX_ arg, 10);
2373 This implementation never has to fetch the context using a function
2374 call, since it is always passed as an extra argument. Depending on
2375 your needs for simplicity or efficiency, you may mix the previous
2376 two approaches freely.
2378 Never add a comma after C<pTHX> yourself--always use the form of the
2379 macro with the underscore for functions that take explicit arguments,
2380 or the form without the argument for functions with no explicit arguments.
2382 If one is compiling Perl with the C<-DPERL_GLOBAL_STRUCT> the C<dVAR>
2383 definition is needed if the Perl global variables (see F<perlvars.h>
2384 or F<globvar.sym>) are accessed in the function and C<dTHX> is not
2385 used (the C<dTHX> includes the C<dVAR> if necessary). One notices
2386 the need for C<dVAR> only with the said compile-time define, because
2387 otherwise the Perl global variables are visible as-is.
2389 =head2 Should I do anything special if I call perl from multiple threads?
2391 If you create interpreters in one thread and then proceed to call them in
2392 another, you need to make sure perl's own Thread Local Storage (TLS) slot is
2393 initialized correctly in each of those threads.
2395 The C<perl_alloc> and C<perl_clone> API functions will automatically set
2396 the TLS slot to the interpreter they created, so that there is no need to do
2397 anything special if the interpreter is always accessed in the same thread that
2398 created it, and that thread did not create or call any other interpreters
2399 afterwards. If that is not the case, you have to set the TLS slot of the
2400 thread before calling any functions in the Perl API on that particular
2401 interpreter. This is done by calling the C<PERL_SET_CONTEXT> macro in that
2402 thread as the first thing you do:
2404 /* do this before doing anything else with some_perl */
2405 PERL_SET_CONTEXT(some_perl);
2407 ... other Perl API calls on some_perl go here ...
2409 =head2 Future Plans and PERL_IMPLICIT_SYS
2411 Just as PERL_IMPLICIT_CONTEXT provides a way to bundle up everything
2412 that the interpreter knows about itself and pass it around, so too are
2413 there plans to allow the interpreter to bundle up everything it knows
2414 about the environment it's running on. This is enabled with the
2415 PERL_IMPLICIT_SYS macro. Currently it only works with USE_ITHREADS on
2418 This allows the ability to provide an extra pointer (called the "host"
2419 environment) for all the system calls. This makes it possible for
2420 all the system stuff to maintain their own state, broken down into
2421 seven C structures. These are thin wrappers around the usual system
2422 calls (see F<win32/perllib.c>) for the default perl executable, but for a
2423 more ambitious host (like the one that would do fork() emulation) all
2424 the extra work needed to pretend that different interpreters are
2425 actually different "processes", would be done here.
2427 The Perl engine/interpreter and the host are orthogonal entities.
2428 There could be one or more interpreters in a process, and one or
2429 more "hosts", with free association between them.
2431 =head1 Internal Functions
2433 All of Perl's internal functions which will be exposed to the outside
2434 world are prefixed by C<Perl_> so that they will not conflict with XS
2435 functions or functions used in a program in which Perl is embedded.
2436 Similarly, all global variables begin with C<PL_>. (By convention,
2437 static functions start with C<S_>.)
2439 Inside the Perl core (C<PERL_CORE> defined), you can get at the functions
2440 either with or without the C<Perl_> prefix, thanks to a bunch of defines
2441 that live in F<embed.h>. Note that extension code should I<not> set
2442 C<PERL_CORE>; this exposes the full perl internals, and is likely to cause
2443 breakage of the XS in each new perl release.
2445 The file F<embed.h> is generated automatically from
2446 F<embed.pl> and F<embed.fnc>. F<embed.pl> also creates the prototyping
2447 header files for the internal functions, generates the documentation
2448 and a lot of other bits and pieces. It's important that when you add
2449 a new function to the core or change an existing one, you change the
2450 data in the table in F<embed.fnc> as well. Here's a sample entry from
2453 Apd |SV** |av_fetch |AV* ar|I32 key|I32 lval
2455 The second column is the return type, the third column the name. Columns
2456 after that are the arguments. The first column is a set of flags:
2462 This function is a part of the public
2463 API. All such functions should also
2464 have 'd', very few do not.
2468 This function has a C<Perl_> prefix; i.e. it is defined as
2473 This function has documentation using the C<apidoc> feature which we'll
2474 look at in a second. Some functions have 'd' but not 'A'; docs are good.
2478 Other available flags are:
2484 This is a static function and is defined as C<STATIC S_whatever>, and
2485 usually called within the sources as C<whatever(...)>.
2489 This does not need an interpreter context, so the definition has no
2490 C<pTHX>, and it follows that callers don't use C<aTHX>. (See
2491 L</Background and PERL_IMPLICIT_CONTEXT>.)
2495 This function never returns; C<croak>, C<exit> and friends.
2499 This function takes a variable number of arguments, C<printf> style.
2500 The argument list should end with C<...>, like this:
2502 Afprd |void |croak |const char* pat|...
2506 This function is part of the experimental development API, and may change
2507 or disappear without notice.
2511 This function should not have a compatibility macro to define, say,
2512 C<Perl_parse> to C<parse>. It must be called as C<Perl_parse>.
2516 This function isn't exported out of the Perl core.
2520 This is implemented as a macro.
2524 This function is explicitly exported.
2528 This function is visible to extensions included in the Perl core.
2532 Binary backward compatibility; this function is a macro but also has
2533 a C<Perl_> implementation (which is exported).
2537 See the comments at the top of C<embed.fnc> for others.
2541 If you edit F<embed.pl> or F<embed.fnc>, you will need to run
2542 C<make regen_headers> to force a rebuild of F<embed.h> and other
2543 auto-generated files.
2545 =head2 Formatted Printing of IVs, UVs, and NVs
2547 If you are printing IVs, UVs, or NVS instead of the stdio(3) style
2548 formatting codes like C<%d>, C<%ld>, C<%f>, you should use the
2549 following macros for portability
2554 UVxf UV in hexadecimal
2559 These will take care of 64-bit integers and long doubles.
2562 printf("IV is %"IVdf"\n", iv);
2564 The IVdf will expand to whatever is the correct format for the IVs.
2566 If you are printing addresses of pointers, use UVxf combined
2567 with PTR2UV(), do not use %lx or %p.
2569 =head2 Pointer-To-Integer and Integer-To-Pointer
2571 Because pointer size does not necessarily equal integer size,
2572 use the follow macros to do it right.
2577 INT2PTR(pointertotype, integer)
2582 SV *sv = INT2PTR(SV*, iv);
2589 =head2 Exception Handling
2591 There are a couple of macros to do very basic exception handling in XS
2592 modules. You have to define C<NO_XSLOCKS> before including F<XSUB.h> to
2593 be able to use these macros:
2598 You can use these macros if you call code that may croak, but you need
2599 to do some cleanup before giving control back to Perl. For example:
2601 dXCPT; /* set up necessary variables */
2604 code_that_may_croak();
2609 /* do cleanup here */
2613 Note that you always have to rethrow an exception that has been
2614 caught. Using these macros, it is not possible to just catch the
2615 exception and ignore it. If you have to ignore the exception, you
2616 have to use the C<call_*> function.
2618 The advantage of using the above macros is that you don't have
2619 to setup an extra function for C<call_*>, and that using these
2620 macros is faster than using C<call_*>.
2622 =head2 Source Documentation
2624 There's an effort going on to document the internal functions and
2625 automatically produce reference manuals from them - L<perlapi> is one
2626 such manual which details all the functions which are available to XS
2627 writers. L<perlintern> is the autogenerated manual for the functions
2628 which are not part of the API and are supposedly for internal use only.
2630 Source documentation is created by putting POD comments into the C
2634 =for apidoc sv_setiv
2636 Copies an integer into the given SV. Does not handle 'set' magic. See
2642 Please try and supply some documentation if you add functions to the
2645 =head2 Backwards compatibility
2647 The Perl API changes over time. New functions are
2648 added or the interfaces of existing functions are
2649 changed. The C<Devel::PPPort> module tries to
2650 provide compatibility code for some of these changes, so XS writers don't
2651 have to code it themselves when supporting multiple versions of Perl.
2653 C<Devel::PPPort> generates a C header file F<ppport.h> that can also
2654 be run as a Perl script. To generate F<ppport.h>, run:
2656 perl -MDevel::PPPort -eDevel::PPPort::WriteFile
2658 Besides checking existing XS code, the script can also be used to retrieve
2659 compatibility information for various API calls using the C<--api-info>
2660 command line switch. For example:
2662 % perl ppport.h --api-info=sv_magicext
2664 For details, see C<perldoc ppport.h>.
2666 =head1 Unicode Support
2668 Perl 5.6.0 introduced Unicode support. It's important for porters and XS
2669 writers to understand this support and make sure that the code they
2670 write does not corrupt Unicode data.
2672 =head2 What B<is> Unicode, anyway?
2674 In the olden, less enlightened times, we all used to use ASCII. Most of
2675 us did, anyway. The big problem with ASCII is that it's American. Well,
2676 no, that's not actually the problem; the problem is that it's not
2677 particularly useful for people who don't use the Roman alphabet. What
2678 used to happen was that particular languages would stick their own
2679 alphabet in the upper range of the sequence, between 128 and 255. Of
2680 course, we then ended up with plenty of variants that weren't quite
2681 ASCII, and the whole point of it being a standard was lost.
2683 Worse still, if you've got a language like Chinese or
2684 Japanese that has hundreds or thousands of characters, then you really
2685 can't fit them into a mere 256, so they had to forget about ASCII
2686 altogether, and build their own systems using pairs of numbers to refer
2689 To fix this, some people formed Unicode, Inc. and
2690 produced a new character set containing all the characters you can
2691 possibly think of and more. There are several ways of representing these
2692 characters, and the one Perl uses is called UTF-8. UTF-8 uses
2693 a variable number of bytes to represent a character. You can learn more
2694 about Unicode and Perl's Unicode model in L<perlunicode>.
2696 =head2 How can I recognise a UTF-8 string?
2698 You can't. This is because UTF-8 data is stored in bytes just like
2699 non-UTF-8 data. The Unicode character 200, (C<0xC8> for you hex types)
2700 capital E with a grave accent, is represented by the two bytes
2701 C<v196.172>. Unfortunately, the non-Unicode string C<chr(196).chr(172)>
2702 has that byte sequence as well. So you can't tell just by looking - this
2703 is what makes Unicode input an interesting problem.
2705 In general, you either have to know what you're dealing with, or you
2706 have to guess. The API function C<is_utf8_string> can help; it'll tell
2707 you if a string contains only valid UTF-8 characters. However, it can't
2708 do the work for you. On a character-by-character basis,
2710 will tell you whether the current character in a string is valid UTF-8.
2712 =head2 How does UTF-8 represent Unicode characters?
2714 As mentioned above, UTF-8 uses a variable number of bytes to store a
2715 character. Characters with values 0...127 are stored in one
2716 byte, just like good ol' ASCII. Character 128 is stored as
2717 C<v194.128>; this continues up to character 191, which is
2718 C<v194.191>. Now we've run out of bits (191 is binary
2719 C<10111111>) so we move on; 192 is C<v195.128>. And
2720 so it goes on, moving to three bytes at character 2048.
2722 Assuming you know you're dealing with a UTF-8 string, you can find out
2723 how long the first character in it is with the C<UTF8SKIP> macro:
2725 char *utf = "\305\233\340\240\201";
2728 len = UTF8SKIP(utf); /* len is 2 here */
2730 len = UTF8SKIP(utf); /* len is 3 here */
2732 Another way to skip over characters in a UTF-8 string is to use
2733 C<utf8_hop>, which takes a string and a number of characters to skip
2734 over. You're on your own about bounds checking, though, so don't use it
2737 All bytes in a multi-byte UTF-8 character will have the high bit set,
2738 so you can test if you need to do something special with this
2739 character like this (the UTF8_IS_INVARIANT() is a macro that tests
2740 whether the byte is encoded as a single byte even in UTF-8):
2743 U8 *utf_end; /* 1 beyond buffer pointed to by utf */
2744 UV uv; /* Note: a UV, not a U8, not a char */
2745 STRLEN len; /* length of character in bytes */
2747 if (!UTF8_IS_INVARIANT(*utf))
2748 /* Must treat this as UTF-8 */
2749 uv = utf8_to_uvchr_buf(utf, utf_end, &len);
2751 /* OK to treat this character as a byte */
2754 You can also see in that example that we use C<utf8_to_uvchr_buf> to get the
2755 value of the character; the inverse function C<uvchr_to_utf8> is available
2756 for putting a UV into UTF-8:
2758 if (!UTF8_IS_INVARIANT(uv))
2759 /* Must treat this as UTF8 */
2760 utf8 = uvchr_to_utf8(utf8, uv);
2762 /* OK to treat this character as a byte */
2765 You B<must> convert characters to UVs using the above functions if
2766 you're ever in a situation where you have to match UTF-8 and non-UTF-8
2767 characters. You may not skip over UTF-8 characters in this case. If you
2768 do this, you'll lose the ability to match hi-bit non-UTF-8 characters;
2769 for instance, if your UTF-8 string contains C<v196.172>, and you skip
2770 that character, you can never match a C<chr(200)> in a non-UTF-8 string.
2773 =head2 How does Perl store UTF-8 strings?
2775 Currently, Perl deals with Unicode strings and non-Unicode strings
2776 slightly differently. A flag in the SV, C<SVf_UTF8>, indicates that the
2777 string is internally encoded as UTF-8. Without it, the byte value is the
2778 codepoint number and vice versa (in other words, the string is encoded
2779 as iso-8859-1, but C<use feature 'unicode_strings'> is needed to get iso-8859-1
2780 semantics). This flag is only meaningful if the SV is C<SvPOK>
2781 or immediately after stringification via C<SvPV> or a similar
2782 macro. You can check and manipulate this flag with the
2789 This flag has an important effect on Perl's treatment of the string: if
2790 Unicode data is not properly distinguished, regular expressions,
2791 C<length>, C<substr> and other string handling operations will have
2792 undesirable results.
2794 The problem comes when you have, for instance, a string that isn't
2795 flagged as UTF-8, and contains a byte sequence that could be UTF-8 -
2796 especially when combining non-UTF-8 and UTF-8 strings.
2798 Never forget that the C<SVf_UTF8> flag is separate to the PV value; you
2799 need be sure you don't accidentally knock it off while you're
2800 manipulating SVs. More specifically, you cannot expect to do this:
2809 nsv = newSVpvn(p, len);
2811 The C<char*> string does not tell you the whole story, and you can't
2812 copy or reconstruct an SV just by copying the string value. Check if the
2813 old SV has the UTF8 flag set (I<after> the C<SvPV> call), and act
2818 nsv = newSVpvn(p, len);
2822 In fact, your C<frobnicate> function should be made aware of whether or
2823 not it's dealing with UTF-8 data, so that it can handle the string
2826 Since just passing an SV to an XS function and copying the data of
2827 the SV is not enough to copy the UTF8 flags, even less right is just
2828 passing a C<char *> to an XS function.
2830 =head2 How do I convert a string to UTF-8?
2832 If you're mixing UTF-8 and non-UTF-8 strings, it is necessary to upgrade
2833 one of the strings to UTF-8. If you've got an SV, the easiest way to do
2836 sv_utf8_upgrade(sv);
2838 However, you must not do this, for example:
2841 sv_utf8_upgrade(left);
2843 If you do this in a binary operator, you will actually change one of the
2844 strings that came into the operator, and, while it shouldn't be noticeable
2845 by the end user, it can cause problems in deficient code.
2847 Instead, C<bytes_to_utf8> will give you a UTF-8-encoded B<copy> of its
2848 string argument. This is useful for having the data available for
2849 comparisons and so on, without harming the original SV. There's also
2850 C<utf8_to_bytes> to go the other way, but naturally, this will fail if
2851 the string contains any characters above 255 that can't be represented
2854 =head2 Is there anything else I need to know?
2856 Not really. Just remember these things:
2862 There's no way to tell if a string is UTF-8 or not. You can tell if an SV
2863 is UTF-8 by looking at its C<SvUTF8> flag after stringifying it
2864 with C<SvPV> or a similar macro. Don't forget to set the flag if
2865 something should be UTF-8. Treat the flag as part of the PV, even though
2866 it's not - if you pass on the PV to somewhere, pass on the flag too.
2870 If a string is UTF-8, B<always> use C<utf8_to_uvchr_buf> to get at the value,
2871 unless C<UTF8_IS_INVARIANT(*s)> in which case you can use C<*s>.
2875 When writing a character C<uv> to a UTF-8 string, B<always> use
2876 C<uvchr_to_utf8>, unless C<UTF8_IS_INVARIANT(uv))> in which case
2877 you can use C<*s = uv>.
2881 Mixing UTF-8 and non-UTF-8 strings is
2882 tricky. Use C<bytes_to_utf8> to get
2883 a new string which is UTF-8 encoded, and then combine them.
2887 =head1 Custom Operators
2889 Custom operator support is an experimental feature that allows you to
2890 define your own ops. This is primarily to allow the building of
2891 interpreters for other languages in the Perl core, but it also allows
2892 optimizations through the creation of "macro-ops" (ops which perform the
2893 functions of multiple ops which are usually executed together, such as
2894 C<gvsv, gvsv, add>.)
2896 This feature is implemented as a new op type, C<OP_CUSTOM>. The Perl
2897 core does not "know" anything special about this op type, and so it will
2898 not be involved in any optimizations. This also means that you can
2899 define your custom ops to be any op structure - unary, binary, list and
2902 It's important to know what custom operators won't do for you. They
2903 won't let you add new syntax to Perl, directly. They won't even let you
2904 add new keywords, directly. In fact, they won't change the way Perl
2905 compiles a program at all. You have to do those changes yourself, after
2906 Perl has compiled the program. You do this either by manipulating the op
2907 tree using a C<CHECK> block and the C<B::Generate> module, or by adding
2908 a custom peephole optimizer with the C<optimize> module.
2910 When you do this, you replace ordinary Perl ops with custom ops by
2911 creating ops with the type C<OP_CUSTOM> and the C<op_ppaddr> of your own
2912 PP function. This should be defined in XS code, and should look like
2913 the PP ops in C<pp_*.c>. You are responsible for ensuring that your op
2914 takes the appropriate number of values from the stack, and you are
2915 responsible for adding stack marks if necessary.
2917 You should also "register" your op with the Perl interpreter so that it
2918 can produce sensible error and warning messages. Since it is possible to
2919 have multiple custom ops within the one "logical" op type C<OP_CUSTOM>,
2920 Perl uses the value of C<< o->op_ppaddr >> to determine which custom op
2921 it is dealing with. You should create an C<XOP> structure for each
2922 ppaddr you use, set the properties of the custom op with
2923 C<XopENTRY_set>, and register the structure against the ppaddr using
2924 C<Perl_custom_op_register>. A trivial example might look like:
2927 static OP *my_pp(pTHX);
2930 XopENTRY_set(&my_xop, xop_name, "myxop");
2931 XopENTRY_set(&my_xop, xop_desc, "Useless custom op");
2932 Perl_custom_op_register(aTHX_ my_pp, &my_xop);
2934 The available fields in the structure are:
2940 A short name for your op. This will be included in some error messages,
2941 and will also be returned as C<< $op->name >> by the L<B|B> module, so
2942 it will appear in the output of module like L<B::Concise|B::Concise>.
2946 A short description of the function of the op.
2950 Which of the various C<*OP> structures this op uses. This should be one of
2951 the C<OA_*> constants from F<op.h>, namely
2971 =item OA_PVOP_OR_SVOP
2973 This should be interpreted as 'C<PVOP>' only. The C<_OR_SVOP> is because
2974 the only core C<PVOP>, C<OP_TRANS>, can sometimes be a C<SVOP> instead.
2982 The other C<OA_*> constants should not be used.
2986 This member is of type C<Perl_cpeep_t>, which expands to C<void
2987 (*Perl_cpeep_t)(aTHX_ OP *o, OP *oldop)>. If it is set, this function
2988 will be called from C<Perl_rpeep> when ops of this type are encountered
2989 by the peephole optimizer. I<o> is the OP that needs optimizing;
2990 I<oldop> is the previous OP optimized, whose C<op_next> points to I<o>.
2994 C<B::Generate> directly supports the creation of custom ops by name.
2998 Until May 1997, this document was maintained by Jeff Okamoto
2999 E<lt>okamoto@corp.hp.comE<gt>. It is now maintained as part of Perl
3000 itself by the Perl 5 Porters E<lt>perl5-porters@perl.orgE<gt>.
3002 With lots of help and suggestions from Dean Roehrich, Malcolm Beattie,
3003 Andreas Koenig, Paul Hudson, Ilya Zakharevich, Paul Marquess, Neil
3004 Bowers, Matthew Green, Tim Bunce, Spider Boardman, Ulrich Pfeifer,
3005 Stephen McCamant, and Gurusamy Sarathy.
3009 L<perlapi>, L<perlintern>, L<perlxs>, L<perlembed>