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, inversion
43 lists, used in regular expression data structures, are scalars, each
44 consisting of an array of UVs which are accessed through PVs. But,
45 using it for non-strings requires care, as the underlying assumption of
46 much of the internals is that PVs are just for strings. Often, for
47 example, a trailing NUL is tacked on automatically. The non-string use
48 is documented only in this paragraph.)
50 The seven routines are:
55 SV* newSVpv(const char*, STRLEN);
56 SV* newSVpvn(const char*, STRLEN);
57 SV* newSVpvf(const char*, ...);
60 C<STRLEN> is an integer type (Size_t, usually defined as size_t in
61 F<config.h>) guaranteed to be large enough to represent the size of
62 any string that perl can handle.
64 In the unlikely case of a SV requiring more complex initialisation, you
65 can create an empty SV with newSV(len). If C<len> is 0 an empty SV of
66 type NULL is returned, else an SV of type PV is returned with len + 1 (for
67 the NUL) bytes of storage allocated, accessible via SvPVX. In both cases
68 the SV has the undef value.
70 SV *sv = newSV(0); /* no storage allocated */
71 SV *sv = newSV(10); /* 10 (+1) bytes of uninitialised storage
74 To change the value of an I<already-existing> SV, there are eight routines:
76 void sv_setiv(SV*, IV);
77 void sv_setuv(SV*, UV);
78 void sv_setnv(SV*, double);
79 void sv_setpv(SV*, const char*);
80 void sv_setpvn(SV*, const char*, STRLEN)
81 void sv_setpvf(SV*, const char*, ...);
82 void sv_vsetpvfn(SV*, const char*, STRLEN, va_list *,
84 void sv_setsv(SV*, SV*);
86 Notice that you can choose to specify the length of the string to be
87 assigned by using C<sv_setpvn>, C<newSVpvn>, or C<newSVpv>, or you may
88 allow Perl to calculate the length by using C<sv_setpv> or by specifying
89 0 as the second argument to C<newSVpv>. Be warned, though, that Perl will
90 determine the string's length by using C<strlen>, which depends on the
91 string terminating with a NUL character, and not otherwise containing
94 The arguments of C<sv_setpvf> are processed like C<sprintf>, and the
95 formatted output becomes the value.
97 C<sv_vsetpvfn> is an analogue of C<vsprintf>, but it allows you to specify
98 either a pointer to a variable argument list or the address and length of
99 an array of SVs. The last argument points to a boolean; on return, if that
100 boolean is true, then locale-specific information has been used to format
101 the string, and the string's contents are therefore untrustworthy (see
102 L<perlsec>). This pointer may be NULL if that information is not
103 important. Note that this function requires you to specify the length of
106 The C<sv_set*()> functions are not generic enough to operate on values
107 that have "magic". See L<Magic Virtual Tables> later in this document.
109 All SVs that contain strings should be terminated with a NUL character.
110 If it is not NUL-terminated there is a risk of
111 core dumps and corruptions from code which passes the string to C
112 functions or system calls which expect a NUL-terminated string.
113 Perl's own functions typically add a trailing NUL for this reason.
114 Nevertheless, you should be very careful when you pass a string stored
115 in an SV to a C function or system call.
117 To access the actual value that an SV points to, you can use the macros:
122 SvPV(SV*, STRLEN len)
125 which will automatically coerce the actual scalar type into an IV, UV, double,
128 In the C<SvPV> macro, the length of the string returned is placed into the
129 variable C<len> (this is a macro, so you do I<not> use C<&len>). If you do
130 not care what the length of the data is, use the C<SvPV_nolen> macro.
131 Historically the C<SvPV> macro with the global variable C<PL_na> has been
132 used in this case. But that can be quite inefficient because C<PL_na> must
133 be accessed in thread-local storage in threaded Perl. In any case, remember
134 that Perl allows arbitrary strings of data that may both contain NULs and
135 might not be terminated by a NUL.
137 Also remember that C doesn't allow you to safely say C<foo(SvPV(s, len),
138 len);>. It might work with your 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 extends the string stored in the first
198 SV with the string stored in the second SV. It also forces the second SV
199 to be interpreted as a string.
201 The C<sv_cat*()> functions are not generic enough to operate on values that
202 have "magic". See L<Magic Virtual Tables> later in this document.
204 If you know the name of a scalar variable, you can get a pointer to its SV
205 by using the following:
207 SV* get_sv("package::varname", 0);
209 This returns NULL if the variable does not exist.
211 If you want to know if this variable (or any other SV) is actually C<defined>,
216 The scalar C<undef> value is stored in an SV instance called C<PL_sv_undef>.
218 Its address can be used whenever an C<SV*> is needed. Make sure that
219 you don't try to compare a random sv with C<&PL_sv_undef>. For example
220 when interfacing Perl code, it'll work correctly for:
224 But won't work when called as:
229 So to repeat always use SvOK() to check whether an sv is defined.
231 Also you have to be careful when using C<&PL_sv_undef> as a value in
232 AVs or HVs (see L<AVs, HVs and undefined values>).
234 There are also the two values C<PL_sv_yes> and C<PL_sv_no>, which contain
235 boolean TRUE and FALSE values, respectively. Like C<PL_sv_undef>, their
236 addresses can be used whenever an C<SV*> is needed.
238 Do not be fooled into thinking that C<(SV *) 0> is the same as C<&PL_sv_undef>.
242 if (I-am-to-return-a-real-value) {
243 sv = sv_2mortal(newSViv(42));
247 This code tries to return a new SV (which contains the value 42) if it should
248 return a real value, or undef otherwise. Instead it has returned a NULL
249 pointer which, somewhere down the line, will cause a segmentation violation,
250 bus error, or just weird results. Change the zero to C<&PL_sv_undef> in the
251 first line and all will be well.
253 To free an SV that you've created, call C<SvREFCNT_dec(SV*)>. Normally this
254 call is not necessary (see L<Reference Counts and Mortality>).
258 Perl provides the function C<sv_chop> to efficiently remove characters
259 from the beginning of a string; you give it an SV and a pointer to
260 somewhere inside the PV, and it discards everything before the
261 pointer. The efficiency comes by means of a little hack: instead of
262 actually removing the characters, C<sv_chop> sets the flag C<OOK>
263 (offset OK) to signal to other functions that the offset hack is in
264 effect, and it puts the number of bytes chopped off into the IV field
265 of the SV. It then moves the PV pointer (called C<SvPVX>) forward that
266 many bytes, and adjusts C<SvCUR> and C<SvLEN>.
268 Hence, at this point, the start of the buffer that we allocated lives
269 at C<SvPVX(sv) - SvIV(sv)> in memory and the PV pointer is pointing
270 into the middle of this allocated storage.
272 This is best demonstrated by example:
274 % ./perl -Ilib -MDevel::Peek -le '$a="12345"; $a=~s/.//; Dump($a)'
275 SV = PVIV(0x8128450) at 0x81340f0
277 FLAGS = (POK,OOK,pPOK)
279 PV = 0x8135781 ( "1" . ) "2345"\0
283 Here the number of bytes chopped off (1) is put into IV, and
284 C<Devel::Peek::Dump> helpfully reminds us that this is an offset. The
285 portion of the string between the "real" and the "fake" beginnings is
286 shown in parentheses, and the values of C<SvCUR> and C<SvLEN> reflect
287 the fake beginning, not the real one.
289 Something similar to the offset hack is performed on AVs to enable
290 efficient shifting and splicing off the beginning of the array; while
291 C<AvARRAY> points to the first element in the array that is visible from
292 Perl, C<AvALLOC> points to the real start of the C array. These are
293 usually the same, but a C<shift> operation can be carried out by
294 increasing C<AvARRAY> by one and decreasing C<AvFILL> and C<AvMAX>.
295 Again, the location of the real start of the C array only comes into
296 play when freeing the array. See C<av_shift> in F<av.c>.
298 =head2 What's Really Stored in an SV?
300 Recall that the usual method of determining the type of scalar you have is
301 to use C<Sv*OK> macros. Because a scalar can be both a number and a string,
302 usually these macros will always return TRUE and calling the C<Sv*V>
303 macros will do the appropriate conversion of string to integer/double or
304 integer/double to string.
306 If you I<really> need to know if you have an integer, double, or string
307 pointer in an SV, you can use the following three macros instead:
313 These will tell you if you truly have an integer, double, or string pointer
314 stored in your SV. The "p" stands for private.
316 There are various ways in which the private and public flags may differ.
317 For example, a tied SV may have a valid underlying value in the IV slot
318 (so SvIOKp is true), but the data should be accessed via the FETCH
319 routine rather than directly, so SvIOK is false. Another is when
320 numeric conversion has occurred and precision has been lost: only the
321 private flag is set on 'lossy' values. So when an NV is converted to an
322 IV with loss, SvIOKp, SvNOKp and SvNOK will be set, while SvIOK wont be.
324 In general, though, it's best to use the C<Sv*V> macros.
326 =head2 Working with AVs
328 There are two ways to create and load an AV. The first method creates an
333 The second method both creates the AV and initially populates it with SVs:
335 AV* av_make(I32 num, SV **ptr);
337 The second argument points to an array containing C<num> C<SV*>'s. Once the
338 AV has been created, the SVs can be destroyed, if so desired.
340 Once the AV has been created, the following operations are possible on it:
342 void av_push(AV*, SV*);
345 void av_unshift(AV*, I32 num);
347 These should be familiar operations, with the exception of C<av_unshift>.
348 This routine adds C<num> elements at the front of the array with the C<undef>
349 value. You must then use C<av_store> (described below) to assign values
350 to these new elements.
352 Here are some other functions:
355 SV** av_fetch(AV*, I32 key, I32 lval);
356 SV** av_store(AV*, I32 key, SV* val);
358 The C<av_len> function returns the highest index value in an array (just
359 like $#array in Perl). If the array is empty, -1 is returned. The
360 C<av_fetch> function returns the value at index C<key>, but if C<lval>
361 is non-zero, then C<av_fetch> will store an undef value at that index.
362 The C<av_store> function stores the value C<val> at index C<key>, and does
363 not increment the reference count of C<val>. Thus the caller is responsible
364 for taking care of that, and if C<av_store> returns NULL, the caller will
365 have to decrement the reference count to avoid a memory leak. Note that
366 C<av_fetch> and C<av_store> both return C<SV**>'s, not C<SV*>'s as their
373 void av_extend(AV*, I32 key);
375 The C<av_clear> function deletes all the elements in the AV* array, but
376 does not actually delete the array itself. The C<av_undef> function will
377 delete all the elements in the array plus the array itself. The
378 C<av_extend> function extends the array so that it contains at least C<key+1>
379 elements. If C<key+1> is less than the currently allocated length of the array,
380 then nothing is done.
382 If you know the name of an array variable, you can get a pointer to its AV
383 by using the following:
385 AV* get_av("package::varname", 0);
387 This returns NULL if the variable does not exist.
389 See L<Understanding the Magic of Tied Hashes and Arrays> for more
390 information on how to use the array access functions on tied arrays.
392 =head2 Working with HVs
394 To create an HV, you use the following routine:
398 Once the HV has been created, the following operations are possible on it:
400 SV** hv_store(HV*, const char* key, U32 klen, SV* val, U32 hash);
401 SV** hv_fetch(HV*, const char* key, U32 klen, I32 lval);
403 The C<klen> parameter is the length of the key being passed in (Note that
404 you cannot pass 0 in as a value of C<klen> to tell Perl to measure the
405 length of the key). The C<val> argument contains the SV pointer to the
406 scalar being stored, and C<hash> is the precomputed hash value (zero if
407 you want C<hv_store> to calculate it for you). The C<lval> parameter
408 indicates whether this fetch is actually a part of a store operation, in
409 which case a new undefined value will be added to the HV with the supplied
410 key and C<hv_fetch> will return as if the value had already existed.
412 Remember that C<hv_store> and C<hv_fetch> return C<SV**>'s and not just
413 C<SV*>. To access the scalar value, you must first dereference the return
414 value. However, you should check to make sure that the return value is
415 not NULL before dereferencing it.
417 The first of these two functions checks if a hash table entry exists, and the
420 bool hv_exists(HV*, const char* key, U32 klen);
421 SV* hv_delete(HV*, const char* key, U32 klen, I32 flags);
423 If C<flags> does not include the C<G_DISCARD> flag then C<hv_delete> will
424 create and return a mortal copy of the deleted value.
426 And more miscellaneous functions:
431 Like their AV counterparts, C<hv_clear> deletes all the entries in the hash
432 table but does not actually delete the hash table. The C<hv_undef> deletes
433 both the entries and the hash table itself.
435 Perl keeps the actual data in a linked list of structures with a typedef of HE.
436 These contain the actual key and value pointers (plus extra administrative
437 overhead). The key is a string pointer; the value is an C<SV*>. However,
438 once you have an C<HE*>, to get the actual key and value, use the routines
441 I32 hv_iterinit(HV*);
442 /* Prepares starting point to traverse hash table */
443 HE* hv_iternext(HV*);
444 /* Get the next entry, and return a pointer to a
445 structure that has both the key and value */
446 char* hv_iterkey(HE* entry, I32* retlen);
447 /* Get the key from an HE structure and also return
448 the length of the key string */
449 SV* hv_iterval(HV*, HE* entry);
450 /* Return an SV pointer to the value of the HE
452 SV* hv_iternextsv(HV*, char** key, I32* retlen);
453 /* This convenience routine combines hv_iternext,
454 hv_iterkey, and hv_iterval. The key and retlen
455 arguments are return values for the key and its
456 length. The value is returned in the SV* argument */
458 If you know the name of a hash variable, you can get a pointer to its HV
459 by using the following:
461 HV* get_hv("package::varname", 0);
463 This returns NULL if the variable does not exist.
465 The hash algorithm is defined in the C<PERL_HASH(hash, key, klen)> macro:
469 hash = (hash * 33) + *key++;
470 hash = hash + (hash >> 5); /* after 5.6 */
472 The last step was added in version 5.6 to improve distribution of
473 lower bits in the resulting hash value.
475 See L<Understanding the Magic of Tied Hashes and Arrays> for more
476 information on how to use the hash access functions on tied hashes.
478 =head2 Hash API Extensions
480 Beginning with version 5.004, the following functions are also supported:
482 HE* hv_fetch_ent (HV* tb, SV* key, I32 lval, U32 hash);
483 HE* hv_store_ent (HV* tb, SV* key, SV* val, U32 hash);
485 bool hv_exists_ent (HV* tb, SV* key, U32 hash);
486 SV* hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash);
488 SV* hv_iterkeysv (HE* entry);
490 Note that these functions take C<SV*> keys, which simplifies writing
491 of extension code that deals with hash structures. These functions
492 also allow passing of C<SV*> keys to C<tie> functions without forcing
493 you to stringify the keys (unlike the previous set of functions).
495 They also return and accept whole hash entries (C<HE*>), making their
496 use more efficient (since the hash number for a particular string
497 doesn't have to be recomputed every time). See L<perlapi> for detailed
500 The following macros must always be used to access the contents of hash
501 entries. Note that the arguments to these macros must be simple
502 variables, since they may get evaluated more than once. See
503 L<perlapi> for detailed descriptions of these macros.
505 HePV(HE* he, STRLEN len)
509 HeSVKEY_force(HE* he)
510 HeSVKEY_set(HE* he, SV* sv)
512 These two lower level macros are defined, but must only be used when
513 dealing with keys that are not C<SV*>s:
518 Note that both C<hv_store> and C<hv_store_ent> do not increment the
519 reference count of the stored C<val>, which is the caller's responsibility.
520 If these functions return a NULL value, the caller will usually have to
521 decrement the reference count of C<val> to avoid a memory leak.
523 =head2 AVs, HVs and undefined values
525 Sometimes you have to store undefined values in AVs or HVs. Although
526 this may be a rare case, it can be tricky. That's because you're
527 used to using C<&PL_sv_undef> if you need an undefined SV.
529 For example, intuition tells you that this XS code:
532 av_store( av, 0, &PL_sv_undef );
534 is equivalent to this Perl code:
539 Unfortunately, this isn't true. AVs use C<&PL_sv_undef> as a marker
540 for indicating that an array element has not yet been initialized.
541 Thus, C<exists $av[0]> would be true for the above Perl code, but
542 false for the array generated by the XS code.
544 Other problems can occur when storing C<&PL_sv_undef> in HVs:
546 hv_store( hv, "key", 3, &PL_sv_undef, 0 );
548 This will indeed make the value C<undef>, but if you try to modify
549 the value of C<key>, you'll get the following error:
551 Modification of non-creatable hash value attempted
553 In perl 5.8.0, C<&PL_sv_undef> was also used to mark placeholders
554 in restricted hashes. This caused such hash entries not to appear
555 when iterating over the hash or when checking for the keys
556 with the C<hv_exists> function.
558 You can run into similar problems when you store C<&PL_sv_yes> or
559 C<&PL_sv_no> into AVs or HVs. Trying to modify such elements
560 will give you the following error:
562 Modification of a read-only value attempted
564 To make a long story short, you can use the special variables
565 C<&PL_sv_undef>, C<&PL_sv_yes> and C<&PL_sv_no> with AVs and
566 HVs, but you have to make sure you know what you're doing.
568 Generally, if you want to store an undefined value in an AV
569 or HV, you should not use C<&PL_sv_undef>, but rather create a
570 new undefined value using the C<newSV> function, for example:
572 av_store( av, 42, newSV(0) );
573 hv_store( hv, "foo", 3, newSV(0), 0 );
577 References are a special type of scalar that point to other data types
578 (including other references).
580 To create a reference, use either of the following functions:
582 SV* newRV_inc((SV*) thing);
583 SV* newRV_noinc((SV*) thing);
585 The C<thing> argument can be any of an C<SV*>, C<AV*>, or C<HV*>. The
586 functions are identical except that C<newRV_inc> increments the reference
587 count of the C<thing>, while C<newRV_noinc> does not. For historical
588 reasons, C<newRV> is a synonym for C<newRV_inc>.
590 Once you have a reference, you can use the following macro to dereference
595 then call the appropriate routines, casting the returned C<SV*> to either an
596 C<AV*> or C<HV*>, if required.
598 To determine if an SV is a reference, you can use the following macro:
602 To discover what type of value the reference refers to, use the following
603 macro and then check the return value.
607 The most useful types that will be returned are:
616 SVt_PVGV Glob (possibly a file handle)
617 SVt_PVMG Blessed or Magical Scalar
619 See the F<sv.h> header file 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 times. Thinking of "Mortalization"
804 as deferred C<SvREFCNT_dec> should help to minimize such problems.
805 For example if you are passing an SV which you I<know> has a high enough REFCNT
806 to survive its use on the stack you need not do any mortalization.
807 If you are not sure then doing an C<SvREFCNT_inc> and C<sv_2mortal>, or
808 making a C<sv_mortalcopy> is safer.
810 The mortal routines are not just for SVs; AVs and HVs can be
811 made mortal by passing their address (type-casted to C<SV*>) to the
812 C<sv_2mortal> or C<sv_mortalcopy> routines.
814 =head2 Stashes and Globs
816 A B<stash> is a hash that contains all variables that are defined
817 within a package. Each key of the stash is a symbol
818 name (shared by all the different types of objects that have the same
819 name), and each value in the hash table is a GV (Glob Value). This GV
820 in turn contains references to the various objects of that name,
821 including (but not limited to) the following:
830 There is a single stash called C<PL_defstash> that holds the items that exist
831 in the C<main> package. To get at the items in other packages, append the
832 string "::" to the package name. The items in the C<Foo> package are in
833 the stash C<Foo::> in PL_defstash. The items in the C<Bar::Baz> package are
834 in the stash C<Baz::> in C<Bar::>'s stash.
836 To get the stash pointer for a particular package, use the function:
838 HV* gv_stashpv(const char* name, I32 flags)
839 HV* gv_stashsv(SV*, I32 flags)
841 The first function takes a literal string, the second uses the string stored
842 in the SV. Remember that a stash is just a hash table, so you get back an
843 C<HV*>. The C<flags> flag will create a new package if it is set to GV_ADD.
845 The name that C<gv_stash*v> wants is the name of the package whose symbol table
846 you want. The default package is called C<main>. If you have multiply nested
847 packages, pass their names to C<gv_stash*v>, separated by C<::> as in the Perl
850 Alternately, if you have an SV that is a blessed reference, you can find
851 out the stash pointer by using:
853 HV* SvSTASH(SvRV(SV*));
855 then use the following to get the package name itself:
857 char* HvNAME(HV* stash);
859 If you need to bless or re-bless an object you can use the following
862 SV* sv_bless(SV*, HV* stash)
864 where the first argument, an C<SV*>, must be a reference, and the second
865 argument is a stash. The returned C<SV*> can now be used in the same way
868 For more information on references and blessings, consult L<perlref>.
870 =head2 Double-Typed SVs
872 Scalar variables normally contain only one type of value, an integer,
873 double, pointer, or reference. Perl will automatically convert the
874 actual scalar data from the stored type into the requested type.
876 Some scalar variables contain more than one type of scalar data. For
877 example, the variable C<$!> contains either the numeric value of C<errno>
878 or its string equivalent from either C<strerror> or C<sys_errlist[]>.
880 To force multiple data values into an SV, you must do two things: use the
881 C<sv_set*v> routines to add the additional scalar type, then set a flag
882 so that Perl will believe it contains more than one type of data. The
883 four macros to set the flags are:
890 The particular macro you must use depends on which C<sv_set*v> routine
891 you called first. This is because every C<sv_set*v> routine turns on
892 only the bit for the particular type of data being set, and turns off
895 For example, to create a new Perl variable called "dberror" that contains
896 both the numeric and descriptive string error values, you could use the
900 extern char *dberror_list;
902 SV* sv = get_sv("dberror", GV_ADD);
903 sv_setiv(sv, (IV) dberror);
904 sv_setpv(sv, dberror_list[dberror]);
907 If the order of C<sv_setiv> and C<sv_setpv> had been reversed, then the
908 macro C<SvPOK_on> would need to be called instead of C<SvIOK_on>.
910 =head2 Magic Variables
912 [This section still under construction. Ignore everything here. Post no
913 bills. Everything not permitted is forbidden.]
915 Any SV may be magical, that is, it has special features that a normal
916 SV does not have. These features are stored in the SV structure in a
917 linked list of C<struct magic>'s, typedef'ed to C<MAGIC>.
930 Note this is current as of patchlevel 0, and could change at any time.
932 =head2 Assigning Magic
934 Perl adds magic to an SV using the sv_magic function:
936 void sv_magic(SV* sv, SV* obj, int how, const char* name, I32 namlen);
938 The C<sv> argument is a pointer to the SV that is to acquire a new magical
941 If C<sv> is not already magical, Perl uses the C<SvUPGRADE> macro to
942 convert C<sv> to type C<SVt_PVMG>. Perl then continues by adding new magic
943 to the beginning of the linked list of magical features. Any prior entry
944 of the same type of magic is deleted. Note that this can be overridden,
945 and multiple instances of the same type of magic can be associated with an
948 The C<name> and C<namlen> arguments are used to associate a string with
949 the magic, typically the name of a variable. C<namlen> is stored in the
950 C<mg_len> field and if C<name> is non-null then either a C<savepvn> copy of
951 C<name> or C<name> itself is stored in the C<mg_ptr> field, depending on
952 whether C<namlen> is greater than zero or equal to zero respectively. As a
953 special case, if C<(name && namlen == HEf_SVKEY)> then C<name> is assumed
954 to contain an C<SV*> and is stored as-is with its REFCNT incremented.
956 The sv_magic function uses C<how> to determine which, if any, predefined
957 "Magic Virtual Table" should be assigned to the C<mg_virtual> field.
958 See the L<Magic Virtual Tables> section below. The C<how> argument is also
959 stored in the C<mg_type> field. The value of C<how> should be chosen
960 from the set of macros C<PERL_MAGIC_foo> found in F<perl.h>. Note that before
961 these macros were added, Perl internals used to directly use character
962 literals, so you may occasionally come across old code or documentation
963 referring to 'U' magic rather than C<PERL_MAGIC_uvar> for example.
965 The C<obj> argument is stored in the C<mg_obj> field of the C<MAGIC>
966 structure. If it is not the same as the C<sv> argument, the reference
967 count of the C<obj> object is incremented. If it is the same, or if
968 the C<how> argument is C<PERL_MAGIC_arylen>, or if it is a NULL pointer,
969 then C<obj> is merely stored, without the reference count being incremented.
971 See also C<sv_magicext> in L<perlapi> for a more flexible way to add magic
974 There is also a function to add magic to an C<HV>:
976 void hv_magic(HV *hv, GV *gv, int how);
978 This simply calls C<sv_magic> and coerces the C<gv> argument into an C<SV>.
980 To remove the magic from an SV, call the function sv_unmagic:
982 int sv_unmagic(SV *sv, int type);
984 The C<type> argument should be equal to the C<how> value when the C<SV>
985 was initially made magical.
987 However, note that C<sv_unmagic> removes all magic of a certain C<type> from the
988 C<SV>. If you want to remove only certain magic of a C<type> based on the magic
989 virtual table, use C<sv_unmagicext> instead:
991 int sv_unmagicext(SV *sv, int type, MGVTBL *vtbl);
993 =head2 Magic Virtual Tables
995 The C<mg_virtual> field in the C<MAGIC> structure is a pointer to an
996 C<MGVTBL>, which is a structure of function pointers and stands for
997 "Magic Virtual Table" to handle the various operations that might be
998 applied to that variable.
1000 The C<MGVTBL> has five (or sometimes eight) pointers to the following
1003 int (*svt_get)(SV* sv, MAGIC* mg);
1004 int (*svt_set)(SV* sv, MAGIC* mg);
1005 U32 (*svt_len)(SV* sv, MAGIC* mg);
1006 int (*svt_clear)(SV* sv, MAGIC* mg);
1007 int (*svt_free)(SV* sv, MAGIC* mg);
1009 int (*svt_copy)(SV *sv, MAGIC* mg, SV *nsv,
1010 const char *name, I32 namlen);
1011 int (*svt_dup)(MAGIC *mg, CLONE_PARAMS *param);
1012 int (*svt_local)(SV *nsv, MAGIC *mg);
1015 This MGVTBL structure is set at compile-time in F<perl.h> and there are
1016 currently 32 types. These different structures contain pointers to various
1017 routines that perform additional actions depending on which function is
1020 Function pointer Action taken
1021 ---------------- ------------
1022 svt_get Do something before the value of the SV is
1024 svt_set Do something after the SV is assigned a value.
1025 svt_len Report on the SV's length.
1026 svt_clear Clear something the SV represents.
1027 svt_free Free any extra storage associated with the SV.
1029 svt_copy copy tied variable magic to a tied element
1030 svt_dup duplicate a magic structure during thread cloning
1031 svt_local copy magic to local value during 'local'
1033 For instance, the MGVTBL structure called C<vtbl_sv> (which corresponds
1034 to an C<mg_type> of C<PERL_MAGIC_sv>) contains:
1036 { magic_get, magic_set, magic_len, 0, 0 }
1038 Thus, when an SV is determined to be magical and of type C<PERL_MAGIC_sv>,
1039 if a get operation is being performed, the routine C<magic_get> is
1040 called. All the various routines for the various magical types begin
1041 with C<magic_>. NOTE: the magic routines are not considered part of
1042 the Perl API, and may not be exported by the Perl library.
1044 The last three slots are a recent addition, and for source code
1045 compatibility they are only checked for if one of the three flags
1046 MGf_COPY, MGf_DUP or MGf_LOCAL is set in mg_flags. This means that most
1047 code can continue declaring a vtable as a 5-element value. These three are
1048 currently used exclusively by the threading code, and are highly subject
1051 The current kinds of Magic Virtual Tables are:
1054 This table is generated by regen/mg_vtable.pl. Any changes made here
1057 =for mg_vtable.pl begin
1060 (old-style char and macro) MGVTBL Type of magic
1061 -------------------------- ------ -------------
1062 \0 PERL_MAGIC_sv vtbl_sv Special scalar variable
1063 # PERL_MAGIC_arylen vtbl_arylen Array length ($#ary)
1064 % PERL_MAGIC_rhash (none) extra data for restricted
1066 . PERL_MAGIC_pos vtbl_pos pos() lvalue
1067 : PERL_MAGIC_symtab (none) extra data for symbol
1069 < PERL_MAGIC_backref vtbl_backref for weak ref data
1070 @ PERL_MAGIC_arylen_p (none) to move arylen out of XPVAV
1071 B PERL_MAGIC_bm vtbl_regexp Boyer-Moore
1072 (fast string search)
1073 c PERL_MAGIC_overload_table vtbl_ovrld Holds overload table
1075 D PERL_MAGIC_regdata vtbl_regdata Regex match position data
1077 d PERL_MAGIC_regdatum vtbl_regdatum Regex match position data
1079 E PERL_MAGIC_env vtbl_env %ENV hash
1080 e PERL_MAGIC_envelem vtbl_envelem %ENV hash element
1081 f PERL_MAGIC_fm vtbl_regdata Formline
1083 g PERL_MAGIC_regex_global vtbl_mglob m//g target
1084 H PERL_MAGIC_hints vtbl_hints %^H hash
1085 h PERL_MAGIC_hintselem vtbl_hintselem %^H hash element
1086 I PERL_MAGIC_isa vtbl_isa @ISA array
1087 i PERL_MAGIC_isaelem vtbl_isaelem @ISA array element
1088 k PERL_MAGIC_nkeys vtbl_nkeys scalar(keys()) lvalue
1089 L PERL_MAGIC_dbfile (none) Debugger %_<filename
1090 l PERL_MAGIC_dbline vtbl_dbline Debugger %_<filename
1092 N PERL_MAGIC_shared (none) Shared between threads
1093 n PERL_MAGIC_shared_scalar (none) Shared between threads
1094 o PERL_MAGIC_collxfrm vtbl_collxfrm Locale transformation
1095 P PERL_MAGIC_tied vtbl_pack Tied array or hash
1096 p PERL_MAGIC_tiedelem vtbl_packelem Tied array or hash element
1097 q PERL_MAGIC_tiedscalar vtbl_packelem Tied scalar or handle
1098 r PERL_MAGIC_qr vtbl_regexp precompiled qr// regex
1099 S PERL_MAGIC_sig (none) %SIG hash
1100 s PERL_MAGIC_sigelem vtbl_sigelem %SIG hash element
1101 t PERL_MAGIC_taint vtbl_taint Taintedness
1102 U PERL_MAGIC_uvar vtbl_uvar Available for use by
1104 u PERL_MAGIC_uvar_elem (none) Reserved for use by
1106 V PERL_MAGIC_vstring (none) SV was vstring literal
1107 v PERL_MAGIC_vec vtbl_vec vec() lvalue
1108 w PERL_MAGIC_utf8 vtbl_utf8 Cached UTF-8 information
1109 x PERL_MAGIC_substr vtbl_substr substr() lvalue
1110 y PERL_MAGIC_defelem vtbl_defelem Shadow "foreach" iterator
1111 variable / smart parameter
1113 ] PERL_MAGIC_checkcall vtbl_checkcall inlining/mutation of call
1115 ~ PERL_MAGIC_ext (none) Available for use by
1118 =for mg_vtable.pl end
1120 When an uppercase and lowercase letter both exist in the table, then the
1121 uppercase letter is typically used to represent some kind of composite type
1122 (a list or a hash), and the lowercase letter is used to represent an element
1123 of that composite type. Some internals code makes use of this case
1124 relationship. However, 'v' and 'V' (vec and v-string) are in no way related.
1126 The C<PERL_MAGIC_ext> and C<PERL_MAGIC_uvar> magic types are defined
1127 specifically for use by extensions and will not be used by perl itself.
1128 Extensions can use C<PERL_MAGIC_ext> magic to 'attach' private information
1129 to variables (typically objects). This is especially useful because
1130 there is no way for normal perl code to corrupt this private information
1131 (unlike using extra elements of a hash object).
1133 Similarly, C<PERL_MAGIC_uvar> magic can be used much like tie() to call a
1134 C function any time a scalar's value is used or changed. The C<MAGIC>'s
1135 C<mg_ptr> field points to a C<ufuncs> structure:
1138 I32 (*uf_val)(pTHX_ IV, SV*);
1139 I32 (*uf_set)(pTHX_ IV, SV*);
1143 When the SV is read from or written to, the C<uf_val> or C<uf_set>
1144 function will be called with C<uf_index> as the first arg and a pointer to
1145 the SV as the second. A simple example of how to add C<PERL_MAGIC_uvar>
1146 magic is shown below. Note that the ufuncs structure is copied by
1147 sv_magic, so you can safely allocate it on the stack.
1155 uf.uf_val = &my_get_fn;
1156 uf.uf_set = &my_set_fn;
1158 sv_magic(sv, 0, PERL_MAGIC_uvar, (char*)&uf, sizeof(uf));
1160 Attaching C<PERL_MAGIC_uvar> to arrays is permissible but has no effect.
1162 For hashes there is a specialized hook that gives control over hash
1163 keys (but not values). This hook calls C<PERL_MAGIC_uvar> 'get' magic
1164 if the "set" function in the C<ufuncs> structure is NULL. The hook
1165 is activated whenever the hash is accessed with a key specified as
1166 an C<SV> through the functions C<hv_store_ent>, C<hv_fetch_ent>,
1167 C<hv_delete_ent>, and C<hv_exists_ent>. Accessing the key as a string
1168 through the functions without the C<..._ent> suffix circumvents the
1169 hook. See L<Hash::Util::FieldHash/GUTS> for a detailed description.
1171 Note that because multiple extensions may be using C<PERL_MAGIC_ext>
1172 or C<PERL_MAGIC_uvar> magic, it is important for extensions to take
1173 extra care to avoid conflict. Typically only using the magic on
1174 objects blessed into the same class as the extension is sufficient.
1175 For C<PERL_MAGIC_ext> magic, it is usually a good idea to define an
1176 C<MGVTBL>, even if all its fields will be C<0>, so that individual
1177 C<MAGIC> pointers can be identified as a particular kind of magic
1178 using their magic virtual table. C<mg_findext> provides an easy way
1181 STATIC MGVTBL my_vtbl = { 0, 0, 0, 0, 0, 0, 0, 0 };
1184 if ((mg = mg_findext(sv, PERL_MAGIC_ext, &my_vtbl))) {
1185 /* this is really ours, not another module's PERL_MAGIC_ext */
1186 my_priv_data_t *priv = (my_priv_data_t *)mg->mg_ptr;
1190 Also note that the C<sv_set*()> and C<sv_cat*()> functions described
1191 earlier do B<not> invoke 'set' magic on their targets. This must
1192 be done by the user either by calling the C<SvSETMAGIC()> macro after
1193 calling these functions, or by using one of the C<sv_set*_mg()> or
1194 C<sv_cat*_mg()> functions. Similarly, generic C code must call the
1195 C<SvGETMAGIC()> macro to invoke any 'get' magic if they use an SV
1196 obtained from external sources in functions that don't handle magic.
1197 See L<perlapi> for a description of these functions.
1198 For example, calls to the C<sv_cat*()> functions typically need to be
1199 followed by C<SvSETMAGIC()>, but they don't need a prior C<SvGETMAGIC()>
1200 since their implementation handles 'get' magic.
1202 =head2 Finding Magic
1204 MAGIC *mg_find(SV *sv, int type); /* Finds the magic pointer of that
1207 This routine returns a pointer to a C<MAGIC> structure stored in the SV.
1208 If the SV does not have that magical feature, C<NULL> is returned. If the
1209 SV has multiple instances of that magical feature, the first one will be
1210 returned. C<mg_findext> can be used to find a C<MAGIC> structure of an SV
1211 based on both its magic type and its magic virtual table:
1213 MAGIC *mg_findext(SV *sv, int type, MGVTBL *vtbl);
1215 Also, if the SV passed to C<mg_find> or C<mg_findext> is not of type
1216 SVt_PVMG, Perl may core dump.
1218 int mg_copy(SV* sv, SV* nsv, const char* key, STRLEN klen);
1220 This routine checks to see what types of magic C<sv> has. If the mg_type
1221 field is an uppercase letter, then the mg_obj is copied to C<nsv>, but
1222 the mg_type field is changed to be the lowercase letter.
1224 =head2 Understanding the Magic of Tied Hashes and Arrays
1226 Tied hashes and arrays are magical beasts of the C<PERL_MAGIC_tied>
1229 WARNING: As of the 5.004 release, proper usage of the array and hash
1230 access functions requires understanding a few caveats. Some
1231 of these caveats are actually considered bugs in the API, to be fixed
1232 in later releases, and are bracketed with [MAYCHANGE] below. If
1233 you find yourself actually applying such information in this section, be
1234 aware that the behavior may change in the future, umm, without warning.
1236 The perl tie function associates a variable with an object that implements
1237 the various GET, SET, etc methods. To perform the equivalent of the perl
1238 tie function from an XSUB, you must mimic this behaviour. The code below
1239 carries out the necessary steps - firstly it creates a new hash, and then
1240 creates a second hash which it blesses into the class which will implement
1241 the tie methods. Lastly it ties the two hashes together, and returns a
1242 reference to the new tied hash. Note that the code below does NOT call the
1243 TIEHASH method in the MyTie class -
1244 see L<Calling Perl Routines from within C Programs> for details on how
1255 tie = newRV_noinc((SV*)newHV());
1256 stash = gv_stashpv("MyTie", GV_ADD);
1257 sv_bless(tie, stash);
1258 hv_magic(hash, (GV*)tie, PERL_MAGIC_tied);
1259 RETVAL = newRV_noinc(hash);
1263 The C<av_store> function, when given a tied array argument, merely
1264 copies the magic of the array onto the value to be "stored", using
1265 C<mg_copy>. It may also return NULL, indicating that the value did not
1266 actually need to be stored in the array. [MAYCHANGE] After a call to
1267 C<av_store> on a tied array, the caller will usually need to call
1268 C<mg_set(val)> to actually invoke the perl level "STORE" method on the
1269 TIEARRAY object. If C<av_store> did return NULL, a call to
1270 C<SvREFCNT_dec(val)> will also be usually necessary to avoid a memory
1273 The previous paragraph is applicable verbatim to tied hash access using the
1274 C<hv_store> and C<hv_store_ent> functions as well.
1276 C<av_fetch> and the corresponding hash functions C<hv_fetch> and
1277 C<hv_fetch_ent> actually return an undefined mortal value whose magic
1278 has been initialized using C<mg_copy>. Note the value so returned does not
1279 need to be deallocated, as it is already mortal. [MAYCHANGE] But you will
1280 need to call C<mg_get()> on the returned value in order to actually invoke
1281 the perl level "FETCH" method on the underlying TIE object. Similarly,
1282 you may also call C<mg_set()> on the return value after possibly assigning
1283 a suitable value to it using C<sv_setsv>, which will invoke the "STORE"
1284 method on the TIE object. [/MAYCHANGE]
1287 In other words, the array or hash fetch/store functions don't really
1288 fetch and store actual values in the case of tied arrays and hashes. They
1289 merely call C<mg_copy> to attach magic to the values that were meant to be
1290 "stored" or "fetched". Later calls to C<mg_get> and C<mg_set> actually
1291 do the job of invoking the TIE methods on the underlying objects. Thus
1292 the magic mechanism currently implements a kind of lazy access to arrays
1295 Currently (as of perl version 5.004), use of the hash and array access
1296 functions requires the user to be aware of whether they are operating on
1297 "normal" hashes and arrays, or on their tied variants. The API may be
1298 changed to provide more transparent access to both tied and normal data
1299 types in future versions.
1302 You would do well to understand that the TIEARRAY and TIEHASH interfaces
1303 are mere sugar to invoke some perl method calls while using the uniform hash
1304 and array syntax. The use of this sugar imposes some overhead (typically
1305 about two to four extra opcodes per FETCH/STORE operation, in addition to
1306 the creation of all the mortal variables required to invoke the methods).
1307 This overhead will be comparatively small if the TIE methods are themselves
1308 substantial, but if they are only a few statements long, the overhead
1309 will not be insignificant.
1311 =head2 Localizing changes
1313 Perl has a very handy construction
1320 This construction is I<approximately> equivalent to
1329 The biggest difference is that the first construction would
1330 reinstate the initial value of $var, irrespective of how control exits
1331 the block: C<goto>, C<return>, C<die>/C<eval>, etc. It is a little bit
1332 more efficient as well.
1334 There is a way to achieve a similar task from C via Perl API: create a
1335 I<pseudo-block>, and arrange for some changes to be automatically
1336 undone at the end of it, either explicit, or via a non-local exit (via
1337 die()). A I<block>-like construct is created by a pair of
1338 C<ENTER>/C<LEAVE> macros (see L<perlcall/"Returning a Scalar">).
1339 Such a construct may be created specially for some important localized
1340 task, or an existing one (like boundaries of enclosing Perl
1341 subroutine/block, or an existing pair for freeing TMPs) may be
1342 used. (In the second case the overhead of additional localization must
1343 be almost negligible.) Note that any XSUB is automatically enclosed in
1344 an C<ENTER>/C<LEAVE> pair.
1346 Inside such a I<pseudo-block> the following service is available:
1350 =item C<SAVEINT(int i)>
1352 =item C<SAVEIV(IV i)>
1354 =item C<SAVEI32(I32 i)>
1356 =item C<SAVELONG(long i)>
1358 These macros arrange things to restore the value of integer variable
1359 C<i> at the end of enclosing I<pseudo-block>.
1361 =item C<SAVESPTR(s)>
1363 =item C<SAVEPPTR(p)>
1365 These macros arrange things to restore the value of pointers C<s> and
1366 C<p>. C<s> must be a pointer of a type which survives conversion to
1367 C<SV*> and back, C<p> should be able to survive conversion to C<char*>
1370 =item C<SAVEFREESV(SV *sv)>
1372 The refcount of C<sv> would be decremented at the end of
1373 I<pseudo-block>. This is similar to C<sv_2mortal> in that it is also a
1374 mechanism for doing a delayed C<SvREFCNT_dec>. However, while C<sv_2mortal>
1375 extends the lifetime of C<sv> until the beginning of the next statement,
1376 C<SAVEFREESV> extends it until the end of the enclosing scope. These
1377 lifetimes can be wildly different.
1379 Also compare C<SAVEMORTALIZESV>.
1381 =item C<SAVEMORTALIZESV(SV *sv)>
1383 Just like C<SAVEFREESV>, but mortalizes C<sv> at the end of the current
1384 scope instead of decrementing its reference count. This usually has the
1385 effect of keeping C<sv> alive until the statement that called the currently
1386 live scope has finished executing.
1388 =item C<SAVEFREEOP(OP *op)>
1390 The C<OP *> is op_free()ed at the end of I<pseudo-block>.
1392 =item C<SAVEFREEPV(p)>
1394 The chunk of memory which is pointed to by C<p> is Safefree()ed at the
1395 end of I<pseudo-block>.
1397 =item C<SAVECLEARSV(SV *sv)>
1399 Clears a slot in the current scratchpad which corresponds to C<sv> at
1400 the end of I<pseudo-block>.
1402 =item C<SAVEDELETE(HV *hv, char *key, I32 length)>
1404 The key C<key> of C<hv> is deleted at the end of I<pseudo-block>. The
1405 string pointed to by C<key> is Safefree()ed. If one has a I<key> in
1406 short-lived storage, the corresponding string may be reallocated like
1409 SAVEDELETE(PL_defstash, savepv(tmpbuf), strlen(tmpbuf));
1411 =item C<SAVEDESTRUCTOR(DESTRUCTORFUNC_NOCONTEXT_t f, void *p)>
1413 At the end of I<pseudo-block> the function C<f> is called with the
1416 =item C<SAVEDESTRUCTOR_X(DESTRUCTORFUNC_t f, void *p)>
1418 At the end of I<pseudo-block> the function C<f> is called with the
1419 implicit context argument (if any), and C<p>.
1421 =item C<SAVESTACK_POS()>
1423 The current offset on the Perl internal stack (cf. C<SP>) is restored
1424 at the end of I<pseudo-block>.
1428 The following API list contains functions, thus one needs to
1429 provide pointers to the modifiable data explicitly (either C pointers,
1430 or Perlish C<GV *>s). Where the above macros take C<int>, a similar
1431 function takes C<int *>.
1435 =item C<SV* save_scalar(GV *gv)>
1437 Equivalent to Perl code C<local $gv>.
1439 =item C<AV* save_ary(GV *gv)>
1441 =item C<HV* save_hash(GV *gv)>
1443 Similar to C<save_scalar>, but localize C<@gv> and C<%gv>.
1445 =item C<void save_item(SV *item)>
1447 Duplicates the current value of C<SV>, on the exit from the current
1448 C<ENTER>/C<LEAVE> I<pseudo-block> will restore the value of C<SV>
1449 using the stored value. It doesn't handle magic. Use C<save_scalar> if
1452 =item C<void save_list(SV **sarg, I32 maxsarg)>
1454 A variant of C<save_item> which takes multiple arguments via an array
1455 C<sarg> of C<SV*> of length C<maxsarg>.
1457 =item C<SV* save_svref(SV **sptr)>
1459 Similar to C<save_scalar>, but will reinstate an C<SV *>.
1461 =item C<void save_aptr(AV **aptr)>
1463 =item C<void save_hptr(HV **hptr)>
1465 Similar to C<save_svref>, but localize C<AV *> and C<HV *>.
1469 The C<Alias> module implements localization of the basic types within the
1470 I<caller's scope>. People who are interested in how to localize things in
1471 the containing scope should take a look there too.
1475 =head2 XSUBs and the Argument Stack
1477 The XSUB mechanism is a simple way for Perl programs to access C subroutines.
1478 An XSUB routine will have a stack that contains the arguments from the Perl
1479 program, and a way to map from the Perl data structures to a C equivalent.
1481 The stack arguments are accessible through the C<ST(n)> macro, which returns
1482 the C<n>'th stack argument. Argument 0 is the first argument passed in the
1483 Perl subroutine call. These arguments are C<SV*>, and can be used anywhere
1486 Most of the time, output from the C routine can be handled through use of
1487 the RETVAL and OUTPUT directives. However, there are some cases where the
1488 argument stack is not already long enough to handle all the return values.
1489 An example is the POSIX tzname() call, which takes no arguments, but returns
1490 two, the local time zone's standard and summer time abbreviations.
1492 To handle this situation, the PPCODE directive is used and the stack is
1493 extended using the macro:
1497 where C<SP> is the macro that represents the local copy of the stack pointer,
1498 and C<num> is the number of elements the stack should be extended by.
1500 Now that there is room on the stack, values can be pushed on it using C<PUSHs>
1501 macro. The pushed values will often need to be "mortal" (See
1502 L</Reference Counts and Mortality>):
1504 PUSHs(sv_2mortal(newSViv(an_integer)))
1505 PUSHs(sv_2mortal(newSVuv(an_unsigned_integer)))
1506 PUSHs(sv_2mortal(newSVnv(a_double)))
1507 PUSHs(sv_2mortal(newSVpv("Some String",0)))
1508 /* Although the last example is better written as the more
1510 PUSHs(newSVpvs_flags("Some String", SVs_TEMP))
1512 And now the Perl program calling C<tzname>, the two values will be assigned
1515 ($standard_abbrev, $summer_abbrev) = POSIX::tzname;
1517 An alternate (and possibly simpler) method to pushing values on the stack is
1522 This macro automatically adjusts the stack for you, if needed. Thus, you
1523 do not need to call C<EXTEND> to extend the stack.
1525 Despite their suggestions in earlier versions of this document the macros
1526 C<(X)PUSH[iunp]> are I<not> suited to XSUBs which return multiple results.
1527 For that, either stick to the C<(X)PUSHs> macros shown above, or use the new
1528 C<m(X)PUSH[iunp]> macros instead; see L</Putting a C value on Perl stack>.
1530 For more information, consult L<perlxs> and L<perlxstut>.
1532 =head2 Autoloading with XSUBs
1534 If an AUTOLOAD routine is an XSUB, as with Perl subroutines, Perl puts the
1535 fully-qualified name of the autoloaded subroutine in the $AUTOLOAD variable
1536 of the XSUB's package.
1538 But it also puts the same information in certain fields of the XSUB itself:
1540 HV *stash = CvSTASH(cv);
1541 const char *subname = SvPVX(cv);
1542 STRLEN name_length = SvCUR(cv); /* in bytes */
1543 U32 is_utf8 = SvUTF8(cv);
1545 C<SvPVX(cv)> contains just the sub name itself, not including the package.
1546 For an AUTOLOAD routine in UNIVERSAL or one of its superclasses,
1547 C<CvSTASH(cv)> returns NULL during a method call on a nonexistent package.
1549 B<Note>: Setting $AUTOLOAD stopped working in 5.6.1, which did not support
1550 XS AUTOLOAD subs at all. Perl 5.8.0 introduced the use of fields in the
1551 XSUB itself. Perl 5.16.0 restored the setting of $AUTOLOAD. If you need
1552 to support 5.8-5.14, use the XSUB's fields.
1554 =head2 Calling Perl Routines from within C Programs
1556 There are four routines that can be used to call a Perl subroutine from
1557 within a C program. These four are:
1559 I32 call_sv(SV*, I32);
1560 I32 call_pv(const char*, I32);
1561 I32 call_method(const char*, I32);
1562 I32 call_argv(const char*, I32, register char**);
1564 The routine most often used is C<call_sv>. The C<SV*> argument
1565 contains either the name of the Perl subroutine to be called, or a
1566 reference to the subroutine. The second argument consists of flags
1567 that control the context in which the subroutine is called, whether
1568 or not the subroutine is being passed arguments, how errors should be
1569 trapped, and how to treat return values.
1571 All four routines return the number of arguments that the subroutine returned
1574 These routines used to be called C<perl_call_sv>, etc., before Perl v5.6.0,
1575 but those names are now deprecated; macros of the same name are provided for
1578 When using any of these routines (except C<call_argv>), the programmer
1579 must manipulate the Perl stack. These include the following macros and
1594 For a detailed description of calling conventions from C to Perl,
1595 consult L<perlcall>.
1597 =head2 Memory Allocation
1601 All memory meant to be used with the Perl API functions should be manipulated
1602 using the macros described in this section. The macros provide the necessary
1603 transparency between differences in the actual malloc implementation that is
1606 It is suggested that you enable the version of malloc that is distributed
1607 with Perl. It keeps pools of various sizes of unallocated memory in
1608 order to satisfy allocation requests more quickly. However, on some
1609 platforms, it may cause spurious malloc or free errors.
1611 The following three macros are used to initially allocate memory :
1613 Newx(pointer, number, type);
1614 Newxc(pointer, number, type, cast);
1615 Newxz(pointer, number, type);
1617 The first argument C<pointer> should be the name of a variable that will
1618 point to the newly allocated memory.
1620 The second and third arguments C<number> and C<type> specify how many of
1621 the specified type of data structure should be allocated. The argument
1622 C<type> is passed to C<sizeof>. The final argument to C<Newxc>, C<cast>,
1623 should be used if the C<pointer> argument is different from the C<type>
1626 Unlike the C<Newx> and C<Newxc> macros, the C<Newxz> macro calls C<memzero>
1627 to zero out all the newly allocated memory.
1631 Renew(pointer, number, type);
1632 Renewc(pointer, number, type, cast);
1635 These three macros are used to change a memory buffer size or to free a
1636 piece of memory no longer needed. The arguments to C<Renew> and C<Renewc>
1637 match those of C<New> and C<Newc> with the exception of not needing the
1638 "magic cookie" argument.
1642 Move(source, dest, number, type);
1643 Copy(source, dest, number, type);
1644 Zero(dest, number, type);
1646 These three macros are used to move, copy, or zero out previously allocated
1647 memory. The C<source> and C<dest> arguments point to the source and
1648 destination starting points. Perl will move, copy, or zero out C<number>
1649 instances of the size of the C<type> data structure (using the C<sizeof>
1654 The most recent development releases of Perl have been experimenting with
1655 removing Perl's dependency on the "normal" standard I/O suite and allowing
1656 other stdio implementations to be used. This involves creating a new
1657 abstraction layer that then calls whichever implementation of stdio Perl
1658 was compiled with. All XSUBs should now use the functions in the PerlIO
1659 abstraction layer and not make any assumptions about what kind of stdio
1662 For a complete description of the PerlIO abstraction, consult L<perlapio>.
1664 =head2 Putting a C value on Perl stack
1666 A lot of opcodes (this is an elementary operation in the internal perl
1667 stack machine) put an SV* on the stack. However, as an optimization
1668 the corresponding SV is (usually) not recreated each time. The opcodes
1669 reuse specially assigned SVs (I<target>s) which are (as a corollary)
1670 not constantly freed/created.
1672 Each of the targets is created only once (but see
1673 L<Scratchpads and recursion> below), and when an opcode needs to put
1674 an integer, a double, or a string on stack, it just sets the
1675 corresponding parts of its I<target> and puts the I<target> on stack.
1677 The macro to put this target on stack is C<PUSHTARG>, and it is
1678 directly used in some opcodes, as well as indirectly in zillions of
1679 others, which use it via C<(X)PUSH[iunp]>.
1681 Because the target is reused, you must be careful when pushing multiple
1682 values on the stack. The following code will not do what you think:
1687 This translates as "set C<TARG> to 10, push a pointer to C<TARG> onto
1688 the stack; set C<TARG> to 20, push a pointer to C<TARG> onto the stack".
1689 At the end of the operation, the stack does not contain the values 10
1690 and 20, but actually contains two pointers to C<TARG>, which we have set
1693 If you need to push multiple different values then you should either use
1694 the C<(X)PUSHs> macros, or else use the new C<m(X)PUSH[iunp]> macros,
1695 none of which make use of C<TARG>. The C<(X)PUSHs> macros simply push an
1696 SV* on the stack, which, as noted under L</XSUBs and the Argument Stack>,
1697 will often need to be "mortal". The new C<m(X)PUSH[iunp]> macros make
1698 this a little easier to achieve by creating a new mortal for you (via
1699 C<(X)PUSHmortal>), pushing that onto the stack (extending it if necessary
1700 in the case of the C<mXPUSH[iunp]> macros), and then setting its value.
1701 Thus, instead of writing this to "fix" the example above:
1703 XPUSHs(sv_2mortal(newSViv(10)))
1704 XPUSHs(sv_2mortal(newSViv(20)))
1706 you can simply write:
1711 On a related note, if you do use C<(X)PUSH[iunp]>, then you're going to
1712 need a C<dTARG> in your variable declarations so that the C<*PUSH*>
1713 macros can make use of the local variable C<TARG>. See also C<dTARGET>
1718 The question remains on when the SVs which are I<target>s for opcodes
1719 are created. The answer is that they are created when the current
1720 unit--a subroutine or a file (for opcodes for statements outside of
1721 subroutines)--is compiled. During this time a special anonymous Perl
1722 array is created, which is called a scratchpad for the current unit.
1724 A scratchpad keeps SVs which are lexicals for the current unit and are
1725 targets for opcodes. One can deduce that an SV lives on a scratchpad
1726 by looking on its flags: lexicals have C<SVs_PADMY> set, and
1727 I<target>s have C<SVs_PADTMP> set.
1729 The correspondence between OPs and I<target>s is not 1-to-1. Different
1730 OPs in the compile tree of the unit can use the same target, if this
1731 would not conflict with the expected life of the temporary.
1733 =head2 Scratchpads and recursion
1735 In fact it is not 100% true that a compiled unit contains a pointer to
1736 the scratchpad AV. In fact it contains a pointer to an AV of
1737 (initially) one element, and this element is the scratchpad AV. Why do
1738 we need an extra level of indirection?
1740 The answer is B<recursion>, and maybe B<threads>. Both
1741 these can create several execution pointers going into the same
1742 subroutine. For the subroutine-child not write over the temporaries
1743 for the subroutine-parent (lifespan of which covers the call to the
1744 child), the parent and the child should have different
1745 scratchpads. (I<And> the lexicals should be separate anyway!)
1747 So each subroutine is born with an array of scratchpads (of length 1).
1748 On each entry to the subroutine it is checked that the current
1749 depth of the recursion is not more than the length of this array, and
1750 if it is, new scratchpad is created and pushed into the array.
1752 The I<target>s on this scratchpad are C<undef>s, but they are already
1753 marked with correct flags.
1755 =head1 Compiled code
1759 Here we describe the internal form your code is converted to by
1760 Perl. Start with a simple example:
1764 This is converted to a tree similar to this one:
1772 (but slightly more complicated). This tree reflects the way Perl
1773 parsed your code, but has nothing to do with the execution order.
1774 There is an additional "thread" going through the nodes of the tree
1775 which shows the order of execution of the nodes. In our simplified
1776 example above it looks like:
1778 $b ---> $c ---> + ---> $a ---> assign-to
1780 But with the actual compile tree for C<$a = $b + $c> it is different:
1781 some nodes I<optimized away>. As a corollary, though the actual tree
1782 contains more nodes than our simplified example, the execution order
1783 is the same as in our example.
1785 =head2 Examining the tree
1787 If you have your perl compiled for debugging (usually done with
1788 C<-DDEBUGGING> on the C<Configure> command line), you may examine the
1789 compiled tree by specifying C<-Dx> on the Perl command line. The
1790 output takes several lines per node, and for C<$b+$c> it looks like
1795 FLAGS = (SCALAR,KIDS)
1797 TYPE = null ===> (4)
1799 FLAGS = (SCALAR,KIDS)
1801 3 TYPE = gvsv ===> 4
1807 TYPE = null ===> (5)
1809 FLAGS = (SCALAR,KIDS)
1811 4 TYPE = gvsv ===> 5
1817 This tree has 5 nodes (one per C<TYPE> specifier), only 3 of them are
1818 not optimized away (one per number in the left column). The immediate
1819 children of the given node correspond to C<{}> pairs on the same level
1820 of indentation, thus this listing corresponds to the tree:
1828 The execution order is indicated by C<===E<gt>> marks, thus it is C<3
1829 4 5 6> (node C<6> is not included into above listing), i.e.,
1830 C<gvsv gvsv add whatever>.
1832 Each of these nodes represents an op, a fundamental operation inside the
1833 Perl core. The code which implements each operation can be found in the
1834 F<pp*.c> files; the function which implements the op with type C<gvsv>
1835 is C<pp_gvsv>, and so on. As the tree above shows, different ops have
1836 different numbers of children: C<add> is a binary operator, as one would
1837 expect, and so has two children. To accommodate the various different
1838 numbers of children, there are various types of op data structure, and
1839 they link together in different ways.
1841 The simplest type of op structure is C<OP>: this has no children. Unary
1842 operators, C<UNOP>s, have one child, and this is pointed to by the
1843 C<op_first> field. Binary operators (C<BINOP>s) have not only an
1844 C<op_first> field but also an C<op_last> field. The most complex type of
1845 op is a C<LISTOP>, which has any number of children. In this case, the
1846 first child is pointed to by C<op_first> and the last child by
1847 C<op_last>. The children in between can be found by iteratively
1848 following the C<op_sibling> pointer from the first child to the last.
1850 There are also two other op types: a C<PMOP> holds a regular expression,
1851 and has no children, and a C<LOOP> may or may not have children. If the
1852 C<op_children> field is non-zero, it behaves like a C<LISTOP>. To
1853 complicate matters, if a C<UNOP> is actually a C<null> op after
1854 optimization (see L</Compile pass 2: context propagation>) it will still
1855 have children in accordance with its former type.
1857 Another way to examine the tree is to use a compiler back-end module, such
1860 =head2 Compile pass 1: check routines
1862 The tree is created by the compiler while I<yacc> code feeds it
1863 the constructions it recognizes. Since I<yacc> works bottom-up, so does
1864 the first pass of perl compilation.
1866 What makes this pass interesting for perl developers is that some
1867 optimization may be performed on this pass. This is optimization by
1868 so-called "check routines". The correspondence between node names
1869 and corresponding check routines is described in F<opcode.pl> (do not
1870 forget to run C<make regen_headers> if you modify this file).
1872 A check routine is called when the node is fully constructed except
1873 for the execution-order thread. Since at this time there are no
1874 back-links to the currently constructed node, one can do most any
1875 operation to the top-level node, including freeing it and/or creating
1876 new nodes above/below it.
1878 The check routine returns the node which should be inserted into the
1879 tree (if the top-level node was not modified, check routine returns
1882 By convention, check routines have names C<ck_*>. They are usually
1883 called from C<new*OP> subroutines (or C<convert>) (which in turn are
1884 called from F<perly.y>).
1886 =head2 Compile pass 1a: constant folding
1888 Immediately after the check routine is called the returned node is
1889 checked for being compile-time executable. If it is (the value is
1890 judged to be constant) it is immediately executed, and a I<constant>
1891 node with the "return value" of the corresponding subtree is
1892 substituted instead. The subtree is deleted.
1894 If constant folding was not performed, the execution-order thread is
1897 =head2 Compile pass 2: context propagation
1899 When a context for a part of compile tree is known, it is propagated
1900 down through the tree. At this time the context can have 5 values
1901 (instead of 2 for runtime context): void, boolean, scalar, list, and
1902 lvalue. In contrast with the pass 1 this pass is processed from top
1903 to bottom: a node's context determines the context for its children.
1905 Additional context-dependent optimizations are performed at this time.
1906 Since at this moment the compile tree contains back-references (via
1907 "thread" pointers), nodes cannot be free()d now. To allow
1908 optimized-away nodes at this stage, such nodes are null()ified instead
1909 of free()ing (i.e. their type is changed to OP_NULL).
1911 =head2 Compile pass 3: peephole optimization
1913 After the compile tree for a subroutine (or for an C<eval> or a file)
1914 is created, an additional pass over the code is performed. This pass
1915 is neither top-down or bottom-up, but in the execution order (with
1916 additional complications for conditionals). Optimizations performed
1917 at this stage are subject to the same restrictions as in the pass 2.
1919 Peephole optimizations are done by calling the function pointed to
1920 by the global variable C<PL_peepp>. By default, C<PL_peepp> just
1921 calls the function pointed to by the global variable C<PL_rpeepp>.
1922 By default, that performs some basic op fixups and optimisations along
1923 the execution-order op chain, and recursively calls C<PL_rpeepp> for
1924 each side chain of ops (resulting from conditionals). Extensions may
1925 provide additional optimisations or fixups, hooking into either the
1926 per-subroutine or recursive stage, like this:
1928 static peep_t prev_peepp;
1929 static void my_peep(pTHX_ OP *o)
1931 /* custom per-subroutine optimisation goes here */
1933 /* custom per-subroutine optimisation may also go here */
1936 prev_peepp = PL_peepp;
1939 static peep_t prev_rpeepp;
1940 static void my_rpeep(pTHX_ OP *o)
1943 for(; o; o = o->op_next) {
1944 /* custom per-op optimisation goes here */
1946 prev_rpeepp(orig_o);
1949 prev_rpeepp = PL_rpeepp;
1950 PL_rpeepp = my_rpeep;
1952 =head2 Pluggable runops
1954 The compile tree is executed in a runops function. There are two runops
1955 functions, in F<run.c> and in F<dump.c>. C<Perl_runops_debug> is used
1956 with DEBUGGING and C<Perl_runops_standard> is used otherwise. For fine
1957 control over the execution of the compile tree it is possible to provide
1958 your own runops function.
1960 It's probably best to copy one of the existing runops functions and
1961 change it to suit your needs. Then, in the BOOT section of your XS
1964 PL_runops = my_runops;
1966 This function should be as efficient as possible to keep your programs
1967 running as fast as possible.
1969 =head2 Compile-time scope hooks
1971 As of perl 5.14 it is possible to hook into the compile-time lexical
1972 scope mechanism using C<Perl_blockhook_register>. This is used like
1975 STATIC void my_start_hook(pTHX_ int full);
1976 STATIC BHK my_hooks;
1979 BhkENTRY_set(&my_hooks, bhk_start, my_start_hook);
1980 Perl_blockhook_register(aTHX_ &my_hooks);
1982 This will arrange to have C<my_start_hook> called at the start of
1983 compiling every lexical scope. The available hooks are:
1987 =item C<void bhk_start(pTHX_ int full)>
1989 This is called just after starting a new lexical scope. Note that Perl
1994 creates two scopes: the first starts at the C<(> and has C<full == 1>,
1995 the second starts at the C<{> and has C<full == 0>. Both end at the
1996 C<}>, so calls to C<start> and C<pre/post_end> will match. Anything
1997 pushed onto the save stack by this hook will be popped just before the
1998 scope ends (between the C<pre_> and C<post_end> hooks, in fact).
2000 =item C<void bhk_pre_end(pTHX_ OP **o)>
2002 This is called at the end of a lexical scope, just before unwinding the
2003 stack. I<o> is the root of the optree representing the scope; it is a
2004 double pointer so you can replace the OP if you need to.
2006 =item C<void bhk_post_end(pTHX_ OP **o)>
2008 This is called at the end of a lexical scope, just after unwinding the
2009 stack. I<o> is as above. Note that it is possible for calls to C<pre_>
2010 and C<post_end> to nest, if there is something on the save stack that
2013 =item C<void bhk_eval(pTHX_ OP *const o)>
2015 This is called just before starting to compile an C<eval STRING>, C<do
2016 FILE>, C<require> or C<use>, after the eval has been set up. I<o> is the
2017 OP that requested the eval, and will normally be an C<OP_ENTEREVAL>,
2018 C<OP_DOFILE> or C<OP_REQUIRE>.
2022 Once you have your hook functions, you need a C<BHK> structure to put
2023 them in. It's best to allocate it statically, since there is no way to
2024 free it once it's registered. The function pointers should be inserted
2025 into this structure using the C<BhkENTRY_set> macro, which will also set
2026 flags indicating which entries are valid. If you do need to allocate
2027 your C<BHK> dynamically for some reason, be sure to zero it before you
2030 Once registered, there is no mechanism to switch these hooks off, so if
2031 that is necessary you will need to do this yourself. An entry in C<%^H>
2032 is probably the best way, so the effect is lexically scoped; however it
2033 is also possible to use the C<BhkDISABLE> and C<BhkENABLE> macros to
2034 temporarily switch entries on and off. You should also be aware that
2035 generally speaking at least one scope will have opened before your
2036 extension is loaded, so you will see some C<pre/post_end> pairs that
2037 didn't have a matching C<start>.
2039 =head1 Examining internal data structures with the C<dump> functions
2041 To aid debugging, the source file F<dump.c> contains a number of
2042 functions which produce formatted output of internal data structures.
2044 The most commonly used of these functions is C<Perl_sv_dump>; it's used
2045 for dumping SVs, AVs, HVs, and CVs. The C<Devel::Peek> module calls
2046 C<sv_dump> to produce debugging output from Perl-space, so users of that
2047 module should already be familiar with its format.
2049 C<Perl_op_dump> can be used to dump an C<OP> structure or any of its
2050 derivatives, and produces output similar to C<perl -Dx>; in fact,
2051 C<Perl_dump_eval> will dump the main root of the code being evaluated,
2052 exactly like C<-Dx>.
2054 Other useful functions are C<Perl_dump_sub>, which turns a C<GV> into an
2055 op tree, C<Perl_dump_packsubs> which calls C<Perl_dump_sub> on all the
2056 subroutines in a package like so: (Thankfully, these are all xsubs, so
2057 there is no op tree)
2059 (gdb) print Perl_dump_packsubs(PL_defstash)
2061 SUB attributes::bootstrap = (xsub 0x811fedc 0)
2063 SUB UNIVERSAL::can = (xsub 0x811f50c 0)
2065 SUB UNIVERSAL::isa = (xsub 0x811f304 0)
2067 SUB UNIVERSAL::VERSION = (xsub 0x811f7ac 0)
2069 SUB DynaLoader::boot_DynaLoader = (xsub 0x805b188 0)
2071 and C<Perl_dump_all>, which dumps all the subroutines in the stash and
2072 the op tree of the main root.
2074 =head1 How multiple interpreters and concurrency are supported
2076 =head2 Background and PERL_IMPLICIT_CONTEXT
2078 The Perl interpreter can be regarded as a closed box: it has an API
2079 for feeding it code or otherwise making it do things, but it also has
2080 functions for its own use. This smells a lot like an object, and
2081 there are ways for you to build Perl so that you can have multiple
2082 interpreters, with one interpreter represented either as a C structure,
2083 or inside a thread-specific structure. These structures contain all
2084 the context, the state of that interpreter.
2086 One macro controls the major Perl build flavor: MULTIPLICITY. The
2087 MULTIPLICITY build has a C structure that packages all the interpreter
2088 state. With multiplicity-enabled perls, PERL_IMPLICIT_CONTEXT is also
2089 normally defined, and enables the support for passing in a "hidden" first
2090 argument that represents all three data structures. MULTIPLICITY makes
2091 multi-threaded perls possible (with the ithreads threading model, related
2092 to the macro USE_ITHREADS.)
2094 Two other "encapsulation" macros are the PERL_GLOBAL_STRUCT and
2095 PERL_GLOBAL_STRUCT_PRIVATE (the latter turns on the former, and the
2096 former turns on MULTIPLICITY.) The PERL_GLOBAL_STRUCT causes all the
2097 internal variables of Perl to be wrapped inside a single global struct,
2098 struct perl_vars, accessible as (globals) &PL_Vars or PL_VarsPtr or
2099 the function Perl_GetVars(). The PERL_GLOBAL_STRUCT_PRIVATE goes
2100 one step further, there is still a single struct (allocated in main()
2101 either from heap or from stack) but there are no global data symbols
2102 pointing to it. In either case the global struct should be initialised
2103 as the very first thing in main() using Perl_init_global_struct() and
2104 correspondingly tear it down after perl_free() using Perl_free_global_struct(),
2105 please see F<miniperlmain.c> for usage details. You may also need
2106 to use C<dVAR> in your coding to "declare the global variables"
2107 when you are using them. dTHX does this for you automatically.
2109 To see whether you have non-const data you can use a BSD-compatible C<nm>:
2111 nm libperl.a | grep -v ' [TURtr] '
2113 If this displays any C<D> or C<d> symbols, you have non-const data.
2115 For backward compatibility reasons defining just PERL_GLOBAL_STRUCT
2116 doesn't actually hide all symbols inside a big global struct: some
2117 PerlIO_xxx vtables are left visible. The PERL_GLOBAL_STRUCT_PRIVATE
2118 then hides everything (see how the PERLIO_FUNCS_DECL is used).
2120 All this obviously requires a way for the Perl internal functions to be
2121 either subroutines taking some kind of structure as the first
2122 argument, or subroutines taking nothing as the first argument. To
2123 enable these two very different ways of building the interpreter,
2124 the Perl source (as it does in so many other situations) makes heavy
2125 use of macros and subroutine naming conventions.
2127 First problem: deciding which functions will be public API functions and
2128 which will be private. All functions whose names begin C<S_> are private
2129 (think "S" for "secret" or "static"). All other functions begin with
2130 "Perl_", but just because a function begins with "Perl_" does not mean it is
2131 part of the API. (See L</Internal Functions>.) The easiest way to be B<sure> a
2132 function is part of the API is to find its entry in L<perlapi>.
2133 If it exists in L<perlapi>, it's part of the API. If it doesn't, and you
2134 think it should be (i.e., you need it for your extension), send mail via
2135 L<perlbug> explaining why you think it should be.
2137 Second problem: there must be a syntax so that the same subroutine
2138 declarations and calls can pass a structure as their first argument,
2139 or pass nothing. To solve this, the subroutines are named and
2140 declared in a particular way. Here's a typical start of a static
2141 function used within the Perl guts:
2144 S_incline(pTHX_ char *s)
2146 STATIC becomes "static" in C, and may be #define'd to nothing in some
2147 configurations in the future.
2149 A public function (i.e. part of the internal API, but not necessarily
2150 sanctioned for use in extensions) begins like this:
2153 Perl_sv_setiv(pTHX_ SV* dsv, IV num)
2155 C<pTHX_> is one of a number of macros (in F<perl.h>) that hide the
2156 details of the interpreter's context. THX stands for "thread", "this",
2157 or "thingy", as the case may be. (And no, George Lucas is not involved. :-)
2158 The first character could be 'p' for a B<p>rototype, 'a' for B<a>rgument,
2159 or 'd' for B<d>eclaration, so we have C<pTHX>, C<aTHX> and C<dTHX>, and
2162 When Perl is built without options that set PERL_IMPLICIT_CONTEXT, there is no
2163 first argument containing the interpreter's context. The trailing underscore
2164 in the pTHX_ macro indicates that the macro expansion needs a comma
2165 after the context argument because other arguments follow it. If
2166 PERL_IMPLICIT_CONTEXT is not defined, pTHX_ will be ignored, and the
2167 subroutine is not prototyped to take the extra argument. The form of the
2168 macro without the trailing underscore is used when there are no additional
2171 When a core function calls another, it must pass the context. This
2172 is normally hidden via macros. Consider C<sv_setiv>. It expands into
2173 something like this:
2175 #ifdef PERL_IMPLICIT_CONTEXT
2176 #define sv_setiv(a,b) Perl_sv_setiv(aTHX_ a, b)
2177 /* can't do this for vararg functions, see below */
2179 #define sv_setiv Perl_sv_setiv
2182 This works well, and means that XS authors can gleefully write:
2186 and still have it work under all the modes Perl could have been
2189 This doesn't work so cleanly for varargs functions, though, as macros
2190 imply that the number of arguments is known in advance. Instead we
2191 either need to spell them out fully, passing C<aTHX_> as the first
2192 argument (the Perl core tends to do this with functions like
2193 Perl_warner), or use a context-free version.
2195 The context-free version of Perl_warner is called
2196 Perl_warner_nocontext, and does not take the extra argument. Instead
2197 it does dTHX; to get the context from thread-local storage. We
2198 C<#define warner Perl_warner_nocontext> so that extensions get source
2199 compatibility at the expense of performance. (Passing an arg is
2200 cheaper than grabbing it from thread-local storage.)
2202 You can ignore [pad]THXx when browsing the Perl headers/sources.
2203 Those are strictly for use within the core. Extensions and embedders
2204 need only be aware of [pad]THX.
2206 =head2 So what happened to dTHR?
2208 C<dTHR> was introduced in perl 5.005 to support the older thread model.
2209 The older thread model now uses the C<THX> mechanism to pass context
2210 pointers around, so C<dTHR> is not useful any more. Perl 5.6.0 and
2211 later still have it for backward source compatibility, but it is defined
2214 =head2 How do I use all this in extensions?
2216 When Perl is built with PERL_IMPLICIT_CONTEXT, extensions that call
2217 any functions in the Perl API will need to pass the initial context
2218 argument somehow. The kicker is that you will need to write it in
2219 such a way that the extension still compiles when Perl hasn't been
2220 built with PERL_IMPLICIT_CONTEXT enabled.
2222 There are three ways to do this. First, the easy but inefficient way,
2223 which is also the default, in order to maintain source compatibility
2224 with extensions: whenever F<XSUB.h> is #included, it redefines the aTHX
2225 and aTHX_ macros to call a function that will return the context.
2226 Thus, something like:
2230 in your extension will translate to this when PERL_IMPLICIT_CONTEXT is
2233 Perl_sv_setiv(Perl_get_context(), sv, num);
2235 or to this otherwise:
2237 Perl_sv_setiv(sv, num);
2239 You don't have to do anything new in your extension to get this; since
2240 the Perl library provides Perl_get_context(), it will all just
2243 The second, more efficient way is to use the following template for
2246 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2251 STATIC void my_private_function(int arg1, int arg2);
2254 my_private_function(int arg1, int arg2)
2256 dTHX; /* fetch context */
2257 ... call many Perl API functions ...
2262 MODULE = Foo PACKAGE = Foo
2270 my_private_function(arg, 10);
2272 Note that the only two changes from the normal way of writing an
2273 extension is the addition of a C<#define PERL_NO_GET_CONTEXT> before
2274 including the Perl headers, followed by a C<dTHX;> declaration at
2275 the start of every function that will call the Perl API. (You'll
2276 know which functions need this, because the C compiler will complain
2277 that there's an undeclared identifier in those functions.) No changes
2278 are needed for the XSUBs themselves, because the XS() macro is
2279 correctly defined to pass in the implicit context if needed.
2281 The third, even more efficient way is to ape how it is done within
2285 #define PERL_NO_GET_CONTEXT /* we want efficiency */
2290 /* pTHX_ only needed for functions that call Perl API */
2291 STATIC void my_private_function(pTHX_ int arg1, int arg2);
2294 my_private_function(pTHX_ int arg1, int arg2)
2296 /* dTHX; not needed here, because THX is an argument */
2297 ... call Perl API functions ...
2302 MODULE = Foo PACKAGE = Foo
2310 my_private_function(aTHX_ arg, 10);
2312 This implementation never has to fetch the context using a function
2313 call, since it is always passed as an extra argument. Depending on
2314 your needs for simplicity or efficiency, you may mix the previous
2315 two approaches freely.
2317 Never add a comma after C<pTHX> yourself--always use the form of the
2318 macro with the underscore for functions that take explicit arguments,
2319 or the form without the argument for functions with no explicit arguments.
2321 If one is compiling Perl with the C<-DPERL_GLOBAL_STRUCT> the C<dVAR>
2322 definition is needed if the Perl global variables (see F<perlvars.h>
2323 or F<globvar.sym>) are accessed in the function and C<dTHX> is not
2324 used (the C<dTHX> includes the C<dVAR> if necessary). One notices
2325 the need for C<dVAR> only with the said compile-time define, because
2326 otherwise the Perl global variables are visible as-is.
2328 =head2 Should I do anything special if I call perl from multiple threads?
2330 If you create interpreters in one thread and then proceed to call them in
2331 another, you need to make sure perl's own Thread Local Storage (TLS) slot is
2332 initialized correctly in each of those threads.
2334 The C<perl_alloc> and C<perl_clone> API functions will automatically set
2335 the TLS slot to the interpreter they created, so that there is no need to do
2336 anything special if the interpreter is always accessed in the same thread that
2337 created it, and that thread did not create or call any other interpreters
2338 afterwards. If that is not the case, you have to set the TLS slot of the
2339 thread before calling any functions in the Perl API on that particular
2340 interpreter. This is done by calling the C<PERL_SET_CONTEXT> macro in that
2341 thread as the first thing you do:
2343 /* do this before doing anything else with some_perl */
2344 PERL_SET_CONTEXT(some_perl);
2346 ... other Perl API calls on some_perl go here ...
2348 =head2 Future Plans and PERL_IMPLICIT_SYS
2350 Just as PERL_IMPLICIT_CONTEXT provides a way to bundle up everything
2351 that the interpreter knows about itself and pass it around, so too are
2352 there plans to allow the interpreter to bundle up everything it knows
2353 about the environment it's running on. This is enabled with the
2354 PERL_IMPLICIT_SYS macro. Currently it only works with USE_ITHREADS on
2357 This allows the ability to provide an extra pointer (called the "host"
2358 environment) for all the system calls. This makes it possible for
2359 all the system stuff to maintain their own state, broken down into
2360 seven C structures. These are thin wrappers around the usual system
2361 calls (see F<win32/perllib.c>) for the default perl executable, but for a
2362 more ambitious host (like the one that would do fork() emulation) all
2363 the extra work needed to pretend that different interpreters are
2364 actually different "processes", would be done here.
2366 The Perl engine/interpreter and the host are orthogonal entities.
2367 There could be one or more interpreters in a process, and one or
2368 more "hosts", with free association between them.
2370 =head1 Internal Functions
2372 All of Perl's internal functions which will be exposed to the outside
2373 world are prefixed by C<Perl_> so that they will not conflict with XS
2374 functions or functions used in a program in which Perl is embedded.
2375 Similarly, all global variables begin with C<PL_>. (By convention,
2376 static functions start with C<S_>.)
2378 Inside the Perl core (C<PERL_CORE> defined), you can get at the functions
2379 either with or without the C<Perl_> prefix, thanks to a bunch of defines
2380 that live in F<embed.h>. Note that extension code should I<not> set
2381 C<PERL_CORE>; this exposes the full perl internals, and is likely to cause
2382 breakage of the XS in each new perl release.
2384 The file F<embed.h> is generated automatically from
2385 F<embed.pl> and F<embed.fnc>. F<embed.pl> also creates the prototyping
2386 header files for the internal functions, generates the documentation
2387 and a lot of other bits and pieces. It's important that when you add
2388 a new function to the core or change an existing one, you change the
2389 data in the table in F<embed.fnc> as well. Here's a sample entry from
2392 Apd |SV** |av_fetch |AV* ar|I32 key|I32 lval
2394 The second column is the return type, the third column the name. Columns
2395 after that are the arguments. The first column is a set of flags:
2401 This function is a part of the public API. All such functions should also
2402 have 'd', very few do not.
2406 This function has a C<Perl_> prefix; i.e. it is defined as
2411 This function has documentation using the C<apidoc> feature which we'll
2412 look at in a second. Some functions have 'd' but not 'A'; docs are good.
2416 Other available flags are:
2422 This is a static function and is defined as C<STATIC S_whatever>, and
2423 usually called within the sources as C<whatever(...)>.
2427 This does not need an interpreter context, so the definition has no
2428 C<pTHX>, and it follows that callers don't use C<aTHX>. (See
2429 L</Background and PERL_IMPLICIT_CONTEXT>.)
2433 This function never returns; C<croak>, C<exit> and friends.
2437 This function takes a variable number of arguments, C<printf> style.
2438 The argument list should end with C<...>, like this:
2440 Afprd |void |croak |const char* pat|...
2444 This function is part of the experimental development API, and may change
2445 or disappear without notice.
2449 This function should not have a compatibility macro to define, say,
2450 C<Perl_parse> to C<parse>. It must be called as C<Perl_parse>.
2454 This function isn't exported out of the Perl core.
2458 This is implemented as a macro.
2462 This function is explicitly exported.
2466 This function is visible to extensions included in the Perl core.
2470 Binary backward compatibility; this function is a macro but also has
2471 a C<Perl_> implementation (which is exported).
2475 See the comments at the top of C<embed.fnc> for others.
2479 If you edit F<embed.pl> or F<embed.fnc>, you will need to run
2480 C<make regen_headers> to force a rebuild of F<embed.h> and other
2481 auto-generated files.
2483 =head2 Formatted Printing of IVs, UVs, and NVs
2485 If you are printing IVs, UVs, or NVS instead of the stdio(3) style
2486 formatting codes like C<%d>, C<%ld>, C<%f>, you should use the
2487 following macros for portability
2492 UVxf UV in hexadecimal
2497 These will take care of 64-bit integers and long doubles.
2500 printf("IV is %"IVdf"\n", iv);
2502 The IVdf will expand to whatever is the correct format for the IVs.
2504 If you are printing addresses of pointers, use UVxf combined
2505 with PTR2UV(), do not use %lx or %p.
2507 =head2 Pointer-To-Integer and Integer-To-Pointer
2509 Because pointer size does not necessarily equal integer size,
2510 use the follow macros to do it right.
2515 INT2PTR(pointertotype, integer)
2520 SV *sv = INT2PTR(SV*, iv);
2527 =head2 Exception Handling
2529 There are a couple of macros to do very basic exception handling in XS
2530 modules. You have to define C<NO_XSLOCKS> before including F<XSUB.h> to
2531 be able to use these macros:
2536 You can use these macros if you call code that may croak, but you need
2537 to do some cleanup before giving control back to Perl. For example:
2539 dXCPT; /* set up necessary variables */
2542 code_that_may_croak();
2547 /* do cleanup here */
2551 Note that you always have to rethrow an exception that has been
2552 caught. Using these macros, it is not possible to just catch the
2553 exception and ignore it. If you have to ignore the exception, you
2554 have to use the C<call_*> function.
2556 The advantage of using the above macros is that you don't have
2557 to setup an extra function for C<call_*>, and that using these
2558 macros is faster than using C<call_*>.
2560 =head2 Source Documentation
2562 There's an effort going on to document the internal functions and
2563 automatically produce reference manuals from them - L<perlapi> is one
2564 such manual which details all the functions which are available to XS
2565 writers. L<perlintern> is the autogenerated manual for the functions
2566 which are not part of the API and are supposedly for internal use only.
2568 Source documentation is created by putting POD comments into the C
2572 =for apidoc sv_setiv
2574 Copies an integer into the given SV. Does not handle 'set' magic. See
2580 Please try and supply some documentation if you add functions to the
2583 =head2 Backwards compatibility
2585 The Perl API changes over time. New functions are added or the interfaces
2586 of existing functions are changed. The C<Devel::PPPort> module tries to
2587 provide compatibility code for some of these changes, so XS writers don't
2588 have to code it themselves when supporting multiple versions of Perl.
2590 C<Devel::PPPort> generates a C header file F<ppport.h> that can also
2591 be run as a Perl script. To generate F<ppport.h>, run:
2593 perl -MDevel::PPPort -eDevel::PPPort::WriteFile
2595 Besides checking existing XS code, the script can also be used to retrieve
2596 compatibility information for various API calls using the C<--api-info>
2597 command line switch. For example:
2599 % perl ppport.h --api-info=sv_magicext
2601 For details, see C<perldoc ppport.h>.
2603 =head1 Unicode Support
2605 Perl 5.6.0 introduced Unicode support. It's important for porters and XS
2606 writers to understand this support and make sure that the code they
2607 write does not corrupt Unicode data.
2609 =head2 What B<is> Unicode, anyway?
2611 In the olden, less enlightened times, we all used to use ASCII. Most of
2612 us did, anyway. The big problem with ASCII is that it's American. Well,
2613 no, that's not actually the problem; the problem is that it's not
2614 particularly useful for people who don't use the Roman alphabet. What
2615 used to happen was that particular languages would stick their own
2616 alphabet in the upper range of the sequence, between 128 and 255. Of
2617 course, we then ended up with plenty of variants that weren't quite
2618 ASCII, and the whole point of it being a standard was lost.
2620 Worse still, if you've got a language like Chinese or
2621 Japanese that has hundreds or thousands of characters, then you really
2622 can't fit them into a mere 256, so they had to forget about ASCII
2623 altogether, and build their own systems using pairs of numbers to refer
2626 To fix this, some people formed Unicode, Inc. and
2627 produced a new character set containing all the characters you can
2628 possibly think of and more. There are several ways of representing these
2629 characters, and the one Perl uses is called UTF-8. UTF-8 uses
2630 a variable number of bytes to represent a character. You can learn more
2631 about Unicode and Perl's Unicode model in L<perlunicode>.
2633 =head2 How can I recognise a UTF-8 string?
2635 You can't. This is because UTF-8 data is stored in bytes just like
2636 non-UTF-8 data. The Unicode character 200, (C<0xC8> for you hex types)
2637 capital E with a grave accent, is represented by the two bytes
2638 C<v196.172>. Unfortunately, the non-Unicode string C<chr(196).chr(172)>
2639 has that byte sequence as well. So you can't tell just by looking - this
2640 is what makes Unicode input an interesting problem.
2642 In general, you either have to know what you're dealing with, or you
2643 have to guess. The API function C<is_utf8_string> can help; it'll tell
2644 you if a string contains only valid UTF-8 characters. However, it can't
2645 do the work for you. On a character-by-character basis,
2647 will tell you whether the current character in a string is valid UTF-8.
2649 =head2 How does UTF-8 represent Unicode characters?
2651 As mentioned above, UTF-8 uses a variable number of bytes to store a
2652 character. Characters with values 0...127 are stored in one byte, just
2653 like good ol' ASCII. Character 128 is stored as C<v194.128>; this
2654 continues up to character 191, which is C<v194.191>. Now we've run out of
2655 bits (191 is binary C<10111111>) so we move on; 192 is C<v195.128>. And
2656 so it goes on, moving to three bytes at character 2048.
2658 Assuming you know you're dealing with a UTF-8 string, you can find out
2659 how long the first character in it is with the C<UTF8SKIP> macro:
2661 char *utf = "\305\233\340\240\201";
2664 len = UTF8SKIP(utf); /* len is 2 here */
2666 len = UTF8SKIP(utf); /* len is 3 here */
2668 Another way to skip over characters in a UTF-8 string is to use
2669 C<utf8_hop>, which takes a string and a number of characters to skip
2670 over. You're on your own about bounds checking, though, so don't use it
2673 All bytes in a multi-byte UTF-8 character will have the high bit set,
2674 so you can test if you need to do something special with this
2675 character like this (the UTF8_IS_INVARIANT() is a macro that tests
2676 whether the byte can be encoded as a single byte even in UTF-8):
2679 U8 *utf_end; /* 1 beyond buffer pointed to by utf */
2680 UV uv; /* Note: a UV, not a U8, not a char */
2681 STRLEN len; /* length of character in bytes */
2683 if (!UTF8_IS_INVARIANT(*utf))
2684 /* Must treat this as UTF-8 */
2685 uv = utf8_to_uvchr_buf(utf, utf_end, &len);
2687 /* OK to treat this character as a byte */
2690 You can also see in that example that we use C<utf8_to_uvchr_buf> to get the
2691 value of the character; the inverse function C<uvchr_to_utf8> is available
2692 for putting a UV into UTF-8:
2694 if (!UTF8_IS_INVARIANT(uv))
2695 /* Must treat this as UTF8 */
2696 utf8 = uvchr_to_utf8(utf8, uv);
2698 /* OK to treat this character as a byte */
2701 You B<must> convert characters to UVs using the above functions if
2702 you're ever in a situation where you have to match UTF-8 and non-UTF-8
2703 characters. You may not skip over UTF-8 characters in this case. If you
2704 do this, you'll lose the ability to match hi-bit non-UTF-8 characters;
2705 for instance, if your UTF-8 string contains C<v196.172>, and you skip
2706 that character, you can never match a C<chr(200)> in a non-UTF-8 string.
2709 =head2 How does Perl store UTF-8 strings?
2711 Currently, Perl deals with Unicode strings and non-Unicode strings
2712 slightly differently. A flag in the SV, C<SVf_UTF8>, indicates that the
2713 string is internally encoded as UTF-8. Without it, the byte value is the
2714 codepoint number and vice versa (in other words, the string is encoded
2715 as iso-8859-1, but C<use feature 'unicode_strings'> is needed to get iso-8859-1
2716 semantics). You can check and manipulate this flag with the
2723 This flag has an important effect on Perl's treatment of the string: if
2724 Unicode data is not properly distinguished, regular expressions,
2725 C<length>, C<substr> and other string handling operations will have
2726 undesirable results.
2728 The problem comes when you have, for instance, a string that isn't
2729 flagged as UTF-8, and contains a byte sequence that could be UTF-8 -
2730 especially when combining non-UTF-8 and UTF-8 strings.
2732 Never forget that the C<SVf_UTF8> flag is separate to the PV value; you
2733 need be sure you don't accidentally knock it off while you're
2734 manipulating SVs. More specifically, you cannot expect to do this:
2743 nsv = newSVpvn(p, len);
2745 The C<char*> string does not tell you the whole story, and you can't
2746 copy or reconstruct an SV just by copying the string value. Check if the
2747 old SV has the UTF8 flag set, and act accordingly:
2751 nsv = newSVpvn(p, len);
2755 In fact, your C<frobnicate> function should be made aware of whether or
2756 not it's dealing with UTF-8 data, so that it can handle the string
2759 Since just passing an SV to an XS function and copying the data of
2760 the SV is not enough to copy the UTF8 flags, even less right is just
2761 passing a C<char *> to an XS function.
2763 =head2 How do I convert a string to UTF-8?
2765 If you're mixing UTF-8 and non-UTF-8 strings, it is necessary to upgrade
2766 one of the strings to UTF-8. If you've got an SV, the easiest way to do
2769 sv_utf8_upgrade(sv);
2771 However, you must not do this, for example:
2774 sv_utf8_upgrade(left);
2776 If you do this in a binary operator, you will actually change one of the
2777 strings that came into the operator, and, while it shouldn't be noticeable
2778 by the end user, it can cause problems in deficient code.
2780 Instead, C<bytes_to_utf8> will give you a UTF-8-encoded B<copy> of its
2781 string argument. This is useful for having the data available for
2782 comparisons and so on, without harming the original SV. There's also
2783 C<utf8_to_bytes> to go the other way, but naturally, this will fail if
2784 the string contains any characters above 255 that can't be represented
2787 =head2 Is there anything else I need to know?
2789 Not really. Just remember these things:
2795 There's no way to tell if a string is UTF-8 or not. You can tell if an SV
2796 is UTF-8 by looking at its C<SvUTF8> flag. Don't forget to set the flag if
2797 something should be UTF-8. Treat the flag as part of the PV, even though
2798 it's not - if you pass on the PV to somewhere, pass on the flag too.
2802 If a string is UTF-8, B<always> use C<utf8_to_uvchr_buf> to get at the value,
2803 unless C<UTF8_IS_INVARIANT(*s)> in which case you can use C<*s>.
2807 When writing a character C<uv> to a UTF-8 string, B<always> use
2808 C<uvchr_to_utf8>, unless C<UTF8_IS_INVARIANT(uv))> in which case
2809 you can use C<*s = uv>.
2813 Mixing UTF-8 and non-UTF-8 strings is tricky. Use C<bytes_to_utf8> to get
2814 a new string which is UTF-8 encoded, and then combine them.
2818 =head1 Custom Operators
2820 Custom operator support is a new experimental feature that allows you to
2821 define your own ops. This is primarily to allow the building of
2822 interpreters for other languages in the Perl core, but it also allows
2823 optimizations through the creation of "macro-ops" (ops which perform the
2824 functions of multiple ops which are usually executed together, such as
2825 C<gvsv, gvsv, add>.)
2827 This feature is implemented as a new op type, C<OP_CUSTOM>. The Perl
2828 core does not "know" anything special about this op type, and so it will
2829 not be involved in any optimizations. This also means that you can
2830 define your custom ops to be any op structure - unary, binary, list and
2833 It's important to know what custom operators won't do for you. They
2834 won't let you add new syntax to Perl, directly. They won't even let you
2835 add new keywords, directly. In fact, they won't change the way Perl
2836 compiles a program at all. You have to do those changes yourself, after
2837 Perl has compiled the program. You do this either by manipulating the op
2838 tree using a C<CHECK> block and the C<B::Generate> module, or by adding
2839 a custom peephole optimizer with the C<optimize> module.
2841 When you do this, you replace ordinary Perl ops with custom ops by
2842 creating ops with the type C<OP_CUSTOM> and the C<pp_addr> of your own
2843 PP function. This should be defined in XS code, and should look like
2844 the PP ops in C<pp_*.c>. You are responsible for ensuring that your op
2845 takes the appropriate number of values from the stack, and you are
2846 responsible for adding stack marks if necessary.
2848 You should also "register" your op with the Perl interpreter so that it
2849 can produce sensible error and warning messages. Since it is possible to
2850 have multiple custom ops within the one "logical" op type C<OP_CUSTOM>,
2851 Perl uses the value of C<< o->op_ppaddr >> to determine which custom op
2852 it is dealing with. You should create an C<XOP> structure for each
2853 ppaddr you use, set the properties of the custom op with
2854 C<XopENTRY_set>, and register the structure against the ppaddr using
2855 C<Perl_custom_op_register>. A trivial example might look like:
2858 static OP *my_pp(pTHX);
2861 XopENTRY_set(&my_xop, xop_name, "myxop");
2862 XopENTRY_set(&my_xop, xop_desc, "Useless custom op");
2863 Perl_custom_op_register(aTHX_ my_pp, &my_xop);
2865 The available fields in the structure are:
2871 A short name for your op. This will be included in some error messages,
2872 and will also be returned as C<< $op->name >> by the L<B|B> module, so
2873 it will appear in the output of module like L<B::Concise|B::Concise>.
2877 A short description of the function of the op.
2881 Which of the various C<*OP> structures this op uses. This should be one of
2882 the C<OA_*> constants from F<op.h>, namely
2902 =item OA_PVOP_OR_SVOP
2904 This should be interpreted as 'C<PVOP>' only. The C<_OR_SVOP> is because
2905 the only core C<PVOP>, C<OP_TRANS>, can sometimes be a C<SVOP> instead.
2913 The other C<OA_*> constants should not be used.
2917 This member is of type C<Perl_cpeep_t>, which expands to C<void
2918 (*Perl_cpeep_t)(aTHX_ OP *o, OP *oldop)>. If it is set, this function
2919 will be called from C<Perl_rpeep> when ops of this type are encountered
2920 by the peephole optimizer. I<o> is the OP that needs optimizing;
2921 I<oldop> is the previous OP optimized, whose C<op_next> points to I<o>.
2925 C<B::Generate> directly supports the creation of custom ops by name.
2929 Until May 1997, this document was maintained by Jeff Okamoto
2930 E<lt>okamoto@corp.hp.comE<gt>. It is now maintained as part of Perl
2931 itself by the Perl 5 Porters E<lt>perl5-porters@perl.orgE<gt>.
2933 With lots of help and suggestions from Dean Roehrich, Malcolm Beattie,
2934 Andreas Koenig, Paul Hudson, Ilya Zakharevich, Paul Marquess, Neil
2935 Bowers, Matthew Green, Tim Bunce, Spider Boardman, Ulrich Pfeifer,
2936 Stephen McCamant, and Gurusamy Sarathy.
2940 L<perlapi>, L<perlintern>, L<perlxs>, L<perlembed>