3 perlguts - Introduction to the Perl API
7 This document attempts to describe how to use the Perl API, as well as
8 containing 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,
34 =head2 Working with SVs
36 An SV can be created and loaded with one command. There are four types of
37 values that can be loaded: an integer value (IV), a double (NV),
38 a string (PV), and another scalar (SV).
44 SV* newSVpv(const char*, int);
45 SV* newSVpvn(const char*, int);
46 SV* newSVpvf(const char*, ...);
49 To change the value of an *already-existing* SV, there are seven routines:
51 void sv_setiv(SV*, IV);
52 void sv_setuv(SV*, UV);
53 void sv_setnv(SV*, double);
54 void sv_setpv(SV*, const char*);
55 void sv_setpvn(SV*, const char*, int)
56 void sv_setpvf(SV*, const char*, ...);
57 void sv_setpvfn(SV*, const char*, STRLEN, va_list *, SV **, I32, bool);
58 void sv_setsv(SV*, SV*);
60 Notice that you can choose to specify the length of the string to be
61 assigned by using C<sv_setpvn>, C<newSVpvn>, or C<newSVpv>, or you may
62 allow Perl to calculate the length by using C<sv_setpv> or by specifying
63 0 as the second argument to C<newSVpv>. Be warned, though, that Perl will
64 determine the string's length by using C<strlen>, which depends on the
65 string terminating with a NUL character.
67 The arguments of C<sv_setpvf> are processed like C<sprintf>, and the
68 formatted output becomes the value.
70 C<sv_setpvfn> is an analogue of C<vsprintf>, but it allows you to specify
71 either a pointer to a variable argument list or the address and length of
72 an array of SVs. The last argument points to a boolean; on return, if that
73 boolean is true, then locale-specific information has been used to format
74 the string, and the string's contents are therefore untrustworthy (see
75 L<perlsec>). This pointer may be NULL if that information is not
76 important. Note that this function requires you to specify the length of
79 The C<sv_set*()> functions are not generic enough to operate on values
80 that have "magic". See L<Magic Virtual Tables> later in this document.
82 All SVs that contain strings should be terminated with a NUL character.
83 If it is not NUL-terminated there is a risk of
84 core dumps and corruptions from code which passes the string to C
85 functions or system calls which expect a NUL-terminated string.
86 Perl's own functions typically add a trailing NUL for this reason.
87 Nevertheless, you should be very careful when you pass a string stored
88 in an SV to a C function or system call.
90 To access the actual value that an SV points to, you can use the macros:
98 which will automatically coerce the actual scalar type into an IV, UV, double,
101 In the C<SvPV> macro, the length of the string returned is placed into the
102 variable C<len> (this is a macro, so you do I<not> use C<&len>). If you do
103 not care what the length of the data is, use the C<SvPV_nolen> macro.
104 Historically the C<SvPV> macro with the global variable C<PL_na> has been
105 used in this case. But that can be quite inefficient because C<PL_na> must
106 be accessed in thread-local storage in threaded Perl. In any case, remember
107 that Perl allows arbitrary strings of data that may both contain NULs and
108 might not be terminated by a NUL.
110 Also remember that C doesn't allow you to safely say C<foo(SvPV(s, len),
111 len);>. It might work with your compiler, but it won't work for everyone.
112 Break this sort of statement up into separate assignments:
120 If you want to know if the scalar value is TRUE, you can use:
124 Although Perl will automatically grow strings for you, if you need to force
125 Perl to allocate more memory for your SV, you can use the macro
127 SvGROW(SV*, STRLEN newlen)
129 which will determine if more memory needs to be allocated. If so, it will
130 call the function C<sv_grow>. Note that C<SvGROW> can only increase, not
131 decrease, the allocated memory of an SV and that it does not automatically
132 add a byte for the a trailing NUL (perl's own string functions typically do
133 C<SvGROW(sv, len + 1)>).
135 If you have an SV and want to know what kind of data Perl thinks is stored
136 in it, you can use the following macros to check the type of SV you have.
142 You can get and set the current length of the string stored in an SV with
143 the following macros:
146 SvCUR_set(SV*, I32 val)
148 You can also get a pointer to the end of the string stored in the SV
153 But note that these last three macros are valid only if C<SvPOK()> is true.
155 If you want to append something to the end of string stored in an C<SV*>,
156 you can use the following functions:
158 void sv_catpv(SV*, const char*);
159 void sv_catpvn(SV*, const char*, STRLEN);
160 void sv_catpvf(SV*, const char*, ...);
161 void sv_catpvfn(SV*, const char*, STRLEN, va_list *, SV **, I32, bool);
162 void sv_catsv(SV*, SV*);
164 The first function calculates the length of the string to be appended by
165 using C<strlen>. In the second, you specify the length of the string
166 yourself. The third function processes its arguments like C<sprintf> and
167 appends the formatted output. The fourth function works like C<vsprintf>.
168 You can specify the address and length of an array of SVs instead of the
169 va_list argument. The fifth function extends the string stored in the first
170 SV with the string stored in the second SV. It also forces the second SV
171 to be interpreted as a string.
173 The C<sv_cat*()> functions are not generic enough to operate on values that
174 have "magic". See L<Magic Virtual Tables> later in this document.
176 If you know the name of a scalar variable, you can get a pointer to its SV
177 by using the following:
179 SV* get_sv("package::varname", FALSE);
181 This returns NULL if the variable does not exist.
183 If you want to know if this variable (or any other SV) is actually C<defined>,
188 The scalar C<undef> value is stored in an SV instance called C<PL_sv_undef>. Its
189 address can be used whenever an C<SV*> is needed.
191 There are also the two values C<PL_sv_yes> and C<PL_sv_no>, which contain Boolean
192 TRUE and FALSE values, respectively. Like C<PL_sv_undef>, their addresses can
193 be used whenever an C<SV*> is needed.
195 Do not be fooled into thinking that C<(SV *) 0> is the same as C<&PL_sv_undef>.
199 if (I-am-to-return-a-real-value) {
200 sv = sv_2mortal(newSViv(42));
204 This code tries to return a new SV (which contains the value 42) if it should
205 return a real value, or undef otherwise. Instead it has returned a NULL
206 pointer which, somewhere down the line, will cause a segmentation violation,
207 bus error, or just weird results. Change the zero to C<&PL_sv_undef> in the first
208 line and all will be well.
210 To free an SV that you've created, call C<SvREFCNT_dec(SV*)>. Normally this
211 call is not necessary (see L<Reference Counts and Mortality>).
213 =head2 What's Really Stored in an SV?
215 Recall that the usual method of determining the type of scalar you have is
216 to use C<Sv*OK> macros. Because a scalar can be both a number and a string,
217 usually these macros will always return TRUE and calling the C<Sv*V>
218 macros will do the appropriate conversion of string to integer/double or
219 integer/double to string.
221 If you I<really> need to know if you have an integer, double, or string
222 pointer in an SV, you can use the following three macros instead:
228 These will tell you if you truly have an integer, double, or string pointer
229 stored in your SV. The "p" stands for private.
231 In general, though, it's best to use the C<Sv*V> macros.
233 =head2 Working with AVs
235 There are two ways to create and load an AV. The first method creates an
240 The second method both creates the AV and initially populates it with SVs:
242 AV* av_make(I32 num, SV **ptr);
244 The second argument points to an array containing C<num> C<SV*>'s. Once the
245 AV has been created, the SVs can be destroyed, if so desired.
247 Once the AV has been created, the following operations are possible on AVs:
249 void av_push(AV*, SV*);
252 void av_unshift(AV*, I32 num);
254 These should be familiar operations, with the exception of C<av_unshift>.
255 This routine adds C<num> elements at the front of the array with the C<undef>
256 value. You must then use C<av_store> (described below) to assign values
257 to these new elements.
259 Here are some other functions:
262 SV** av_fetch(AV*, I32 key, I32 lval);
263 SV** av_store(AV*, I32 key, SV* val);
265 The C<av_len> function returns the highest index value in array (just
266 like $#array in Perl). If the array is empty, -1 is returned. The
267 C<av_fetch> function returns the value at index C<key>, but if C<lval>
268 is non-zero, then C<av_fetch> will store an undef value at that index.
269 The C<av_store> function stores the value C<val> at index C<key>, and does
270 not increment the reference count of C<val>. Thus the caller is responsible
271 for taking care of that, and if C<av_store> returns NULL, the caller will
272 have to decrement the reference count to avoid a memory leak. Note that
273 C<av_fetch> and C<av_store> both return C<SV**>'s, not C<SV*>'s as their
278 void av_extend(AV*, I32 key);
280 The C<av_clear> function deletes all the elements in the AV* array, but
281 does not actually delete the array itself. The C<av_undef> function will
282 delete all the elements in the array plus the array itself. The
283 C<av_extend> function extends the array so that it contains at least C<key+1>
284 elements. If C<key+1> is less than the currently allocated length of the array,
285 then nothing is done.
287 If you know the name of an array variable, you can get a pointer to its AV
288 by using the following:
290 AV* get_av("package::varname", FALSE);
292 This returns NULL if the variable does not exist.
294 See L<Understanding the Magic of Tied Hashes and Arrays> for more
295 information on how to use the array access functions on tied arrays.
297 =head2 Working with HVs
299 To create an HV, you use the following routine:
303 Once the HV has been created, the following operations are possible on HVs:
305 SV** hv_store(HV*, const char* key, U32 klen, SV* val, U32 hash);
306 SV** hv_fetch(HV*, const char* key, U32 klen, I32 lval);
308 The C<klen> parameter is the length of the key being passed in (Note that
309 you cannot pass 0 in as a value of C<klen> to tell Perl to measure the
310 length of the key). The C<val> argument contains the SV pointer to the
311 scalar being stored, and C<hash> is the precomputed hash value (zero if
312 you want C<hv_store> to calculate it for you). The C<lval> parameter
313 indicates whether this fetch is actually a part of a store operation, in
314 which case a new undefined value will be added to the HV with the supplied
315 key and C<hv_fetch> will return as if the value had already existed.
317 Remember that C<hv_store> and C<hv_fetch> return C<SV**>'s and not just
318 C<SV*>. To access the scalar value, you must first dereference the return
319 value. However, you should check to make sure that the return value is
320 not NULL before dereferencing it.
322 These two functions check if a hash table entry exists, and deletes it.
324 bool hv_exists(HV*, const char* key, U32 klen);
325 SV* hv_delete(HV*, const char* key, U32 klen, I32 flags);
327 If C<flags> does not include the C<G_DISCARD> flag then C<hv_delete> will
328 create and return a mortal copy of the deleted value.
330 And more miscellaneous functions:
335 Like their AV counterparts, C<hv_clear> deletes all the entries in the hash
336 table but does not actually delete the hash table. The C<hv_undef> deletes
337 both the entries and the hash table itself.
339 Perl keeps the actual data in linked list of structures with a typedef of HE.
340 These contain the actual key and value pointers (plus extra administrative
341 overhead). The key is a string pointer; the value is an C<SV*>. However,
342 once you have an C<HE*>, to get the actual key and value, use the routines
345 I32 hv_iterinit(HV*);
346 /* Prepares starting point to traverse hash table */
347 HE* hv_iternext(HV*);
348 /* Get the next entry, and return a pointer to a
349 structure that has both the key and value */
350 char* hv_iterkey(HE* entry, I32* retlen);
351 /* Get the key from an HE structure and also return
352 the length of the key string */
353 SV* hv_iterval(HV*, HE* entry);
354 /* Return a SV pointer to the value of the HE
356 SV* hv_iternextsv(HV*, char** key, I32* retlen);
357 /* This convenience routine combines hv_iternext,
358 hv_iterkey, and hv_iterval. The key and retlen
359 arguments are return values for the key and its
360 length. The value is returned in the SV* argument */
362 If you know the name of a hash variable, you can get a pointer to its HV
363 by using the following:
365 HV* get_hv("package::varname", FALSE);
367 This returns NULL if the variable does not exist.
369 The hash algorithm is defined in the C<PERL_HASH(hash, key, klen)> macro:
373 hash = (hash * 33) + *key++;
374 hash = hash + (hash >> 5); /* after 5.6 */
376 The last step was added in version 5.6 to improve distribution of
377 lower bits in the resulting hash value.
379 See L<Understanding the Magic of Tied Hashes and Arrays> for more
380 information on how to use the hash access functions on tied hashes.
382 =head2 Hash API Extensions
384 Beginning with version 5.004, the following functions are also supported:
386 HE* hv_fetch_ent (HV* tb, SV* key, I32 lval, U32 hash);
387 HE* hv_store_ent (HV* tb, SV* key, SV* val, U32 hash);
389 bool hv_exists_ent (HV* tb, SV* key, U32 hash);
390 SV* hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash);
392 SV* hv_iterkeysv (HE* entry);
394 Note that these functions take C<SV*> keys, which simplifies writing
395 of extension code that deals with hash structures. These functions
396 also allow passing of C<SV*> keys to C<tie> functions without forcing
397 you to stringify the keys (unlike the previous set of functions).
399 They also return and accept whole hash entries (C<HE*>), making their
400 use more efficient (since the hash number for a particular string
401 doesn't have to be recomputed every time). See L<perlapi> for detailed
404 The following macros must always be used to access the contents of hash
405 entries. Note that the arguments to these macros must be simple
406 variables, since they may get evaluated more than once. See
407 L<perlapi> for detailed descriptions of these macros.
409 HePV(HE* he, STRLEN len)
413 HeSVKEY_force(HE* he)
414 HeSVKEY_set(HE* he, SV* sv)
416 These two lower level macros are defined, but must only be used when
417 dealing with keys that are not C<SV*>s:
422 Note that both C<hv_store> and C<hv_store_ent> do not increment the
423 reference count of the stored C<val>, which is the caller's responsibility.
424 If these functions return a NULL value, the caller will usually have to
425 decrement the reference count of C<val> to avoid a memory leak.
429 References are a special type of scalar that point to other data types
430 (including references).
432 To create a reference, use either of the following functions:
434 SV* newRV_inc((SV*) thing);
435 SV* newRV_noinc((SV*) thing);
437 The C<thing> argument can be any of an C<SV*>, C<AV*>, or C<HV*>. The
438 functions are identical except that C<newRV_inc> increments the reference
439 count of the C<thing>, while C<newRV_noinc> does not. For historical
440 reasons, C<newRV> is a synonym for C<newRV_inc>.
442 Once you have a reference, you can use the following macro to dereference
447 then call the appropriate routines, casting the returned C<SV*> to either an
448 C<AV*> or C<HV*>, if required.
450 To determine if an SV is a reference, you can use the following macro:
454 To discover what type of value the reference refers to, use the following
455 macro and then check the return value.
459 The most useful types that will be returned are:
468 SVt_PVGV Glob (possible a file handle)
469 SVt_PVMG Blessed or Magical Scalar
471 See the sv.h header file for more details.
473 =head2 Blessed References and Class Objects
475 References are also used to support object-oriented programming. In the
476 OO lexicon, an object is simply a reference that has been blessed into a
477 package (or class). Once blessed, the programmer may now use the reference
478 to access the various methods in the class.
480 A reference can be blessed into a package with the following function:
482 SV* sv_bless(SV* sv, HV* stash);
484 The C<sv> argument must be a reference. The C<stash> argument specifies
485 which class the reference will belong to. See
486 L<Stashes and Globs> for information on converting class names into stashes.
488 /* Still under construction */
490 Upgrades rv to reference if not already one. Creates new SV for rv to
491 point to. If C<classname> is non-null, the SV is blessed into the specified
492 class. SV is returned.
494 SV* newSVrv(SV* rv, const char* classname);
496 Copies integer or double into an SV whose reference is C<rv>. SV is blessed
497 if C<classname> is non-null.
499 SV* sv_setref_iv(SV* rv, const char* classname, IV iv);
500 SV* sv_setref_nv(SV* rv, const char* classname, NV iv);
502 Copies the pointer value (I<the address, not the string!>) into an SV whose
503 reference is rv. SV is blessed if C<classname> is non-null.
505 SV* sv_setref_pv(SV* rv, const char* classname, PV iv);
507 Copies string into an SV whose reference is C<rv>. Set length to 0 to let
508 Perl calculate the string length. SV is blessed if C<classname> is non-null.
510 SV* sv_setref_pvn(SV* rv, const char* classname, PV iv, STRLEN length);
512 Tests whether the SV is blessed into the specified class. It does not
513 check inheritance relationships.
515 int sv_isa(SV* sv, const char* name);
517 Tests whether the SV is a reference to a blessed object.
519 int sv_isobject(SV* sv);
521 Tests whether the SV is derived from the specified class. SV can be either
522 a reference to a blessed object or a string containing a class name. This
523 is the function implementing the C<UNIVERSAL::isa> functionality.
525 bool sv_derived_from(SV* sv, const char* name);
527 To check if you've got an object derived from a specific class you have
530 if (sv_isobject(sv) && sv_derived_from(sv, class)) { ... }
532 =head2 Creating New Variables
534 To create a new Perl variable with an undef value which can be accessed from
535 your Perl script, use the following routines, depending on the variable type.
537 SV* get_sv("package::varname", TRUE);
538 AV* get_av("package::varname", TRUE);
539 HV* get_hv("package::varname", TRUE);
541 Notice the use of TRUE as the second parameter. The new variable can now
542 be set, using the routines appropriate to the data type.
544 There are additional macros whose values may be bitwise OR'ed with the
545 C<TRUE> argument to enable certain extra features. Those bits are:
547 GV_ADDMULTI Marks the variable as multiply defined, thus preventing the
548 "Name <varname> used only once: possible typo" warning.
549 GV_ADDWARN Issues the warning "Had to create <varname> unexpectedly" if
550 the variable did not exist before the function was called.
552 If you do not specify a package name, the variable is created in the current
555 =head2 Reference Counts and Mortality
557 Perl uses an reference count-driven garbage collection mechanism. SVs,
558 AVs, or HVs (xV for short in the following) start their life with a
559 reference count of 1. If the reference count of an xV ever drops to 0,
560 then it will be destroyed and its memory made available for reuse.
562 This normally doesn't happen at the Perl level unless a variable is
563 undef'ed or the last variable holding a reference to it is changed or
564 overwritten. At the internal level, however, reference counts can be
565 manipulated with the following macros:
567 int SvREFCNT(SV* sv);
568 SV* SvREFCNT_inc(SV* sv);
569 void SvREFCNT_dec(SV* sv);
571 However, there is one other function which manipulates the reference
572 count of its argument. The C<newRV_inc> function, you will recall,
573 creates a reference to the specified argument. As a side effect,
574 it increments the argument's reference count. If this is not what
575 you want, use C<newRV_noinc> instead.
577 For example, imagine you want to return a reference from an XSUB function.
578 Inside the XSUB routine, you create an SV which initially has a reference
579 count of one. Then you call C<newRV_inc>, passing it the just-created SV.
580 This returns the reference as a new SV, but the reference count of the
581 SV you passed to C<newRV_inc> has been incremented to two. Now you
582 return the reference from the XSUB routine and forget about the SV.
583 But Perl hasn't! Whenever the returned reference is destroyed, the
584 reference count of the original SV is decreased to one and nothing happens.
585 The SV will hang around without any way to access it until Perl itself
586 terminates. This is a memory leak.
588 The correct procedure, then, is to use C<newRV_noinc> instead of
589 C<newRV_inc>. Then, if and when the last reference is destroyed,
590 the reference count of the SV will go to zero and it will be destroyed,
591 stopping any memory leak.
593 There are some convenience functions available that can help with the
594 destruction of xVs. These functions introduce the concept of "mortality".
595 An xV that is mortal has had its reference count marked to be decremented,
596 but not actually decremented, until "a short time later". Generally the
597 term "short time later" means a single Perl statement, such as a call to
598 an XSUB function. The actual determinant for when mortal xVs have their
599 reference count decremented depends on two macros, SAVETMPS and FREETMPS.
600 See L<perlcall> and L<perlxs> for more details on these macros.
602 "Mortalization" then is at its simplest a deferred C<SvREFCNT_dec>.
603 However, if you mortalize a variable twice, the reference count will
604 later be decremented twice.
606 You should be careful about creating mortal variables. Strange things
607 can happen if you make the same value mortal within multiple contexts,
608 or if you make a variable mortal multiple times.
610 To create a mortal variable, use the functions:
614 SV* sv_mortalcopy(SV*)
616 The first call creates a mortal SV, the second converts an existing
617 SV to a mortal SV (and thus defers a call to C<SvREFCNT_dec>), and the
618 third creates a mortal copy of an existing SV.
620 The mortal routines are not just for SVs -- AVs and HVs can be
621 made mortal by passing their address (type-casted to C<SV*>) to the
622 C<sv_2mortal> or C<sv_mortalcopy> routines.
624 =head2 Stashes and Globs
626 A "stash" is a hash that contains all of the different objects that
627 are contained within a package. Each key of the stash is a symbol
628 name (shared by all the different types of objects that have the same
629 name), and each value in the hash table is a GV (Glob Value). This GV
630 in turn contains references to the various objects of that name,
631 including (but not limited to) the following:
640 There is a single stash called "PL_defstash" that holds the items that exist
641 in the "main" package. To get at the items in other packages, append the
642 string "::" to the package name. The items in the "Foo" package are in
643 the stash "Foo::" in PL_defstash. The items in the "Bar::Baz" package are
644 in the stash "Baz::" in "Bar::"'s stash.
646 To get the stash pointer for a particular package, use the function:
648 HV* gv_stashpv(const char* name, I32 create)
649 HV* gv_stashsv(SV*, I32 create)
651 The first function takes a literal string, the second uses the string stored
652 in the SV. Remember that a stash is just a hash table, so you get back an
653 C<HV*>. The C<create> flag will create a new package if it is set.
655 The name that C<gv_stash*v> wants is the name of the package whose symbol table
656 you want. The default package is called C<main>. If you have multiply nested
657 packages, pass their names to C<gv_stash*v>, separated by C<::> as in the Perl
660 Alternately, if you have an SV that is a blessed reference, you can find
661 out the stash pointer by using:
663 HV* SvSTASH(SvRV(SV*));
665 then use the following to get the package name itself:
667 char* HvNAME(HV* stash);
669 If you need to bless or re-bless an object you can use the following
672 SV* sv_bless(SV*, HV* stash)
674 where the first argument, an C<SV*>, must be a reference, and the second
675 argument is a stash. The returned C<SV*> can now be used in the same way
678 For more information on references and blessings, consult L<perlref>.
680 =head2 Double-Typed SVs
682 Scalar variables normally contain only one type of value, an integer,
683 double, pointer, or reference. Perl will automatically convert the
684 actual scalar data from the stored type into the requested type.
686 Some scalar variables contain more than one type of scalar data. For
687 example, the variable C<$!> contains either the numeric value of C<errno>
688 or its string equivalent from either C<strerror> or C<sys_errlist[]>.
690 To force multiple data values into an SV, you must do two things: use the
691 C<sv_set*v> routines to add the additional scalar type, then set a flag
692 so that Perl will believe it contains more than one type of data. The
693 four macros to set the flags are:
700 The particular macro you must use depends on which C<sv_set*v> routine
701 you called first. This is because every C<sv_set*v> routine turns on
702 only the bit for the particular type of data being set, and turns off
705 For example, to create a new Perl variable called "dberror" that contains
706 both the numeric and descriptive string error values, you could use the
710 extern char *dberror_list;
712 SV* sv = get_sv("dberror", TRUE);
713 sv_setiv(sv, (IV) dberror);
714 sv_setpv(sv, dberror_list[dberror]);
717 If the order of C<sv_setiv> and C<sv_setpv> had been reversed, then the
718 macro C<SvPOK_on> would need to be called instead of C<SvIOK_on>.
720 =head2 Magic Variables
722 [This section still under construction. Ignore everything here. Post no
723 bills. Everything not permitted is forbidden.]
725 Any SV may be magical, that is, it has special features that a normal
726 SV does not have. These features are stored in the SV structure in a
727 linked list of C<struct magic>'s, typedef'ed to C<MAGIC>.
740 Note this is current as of patchlevel 0, and could change at any time.
742 =head2 Assigning Magic
744 Perl adds magic to an SV using the sv_magic function:
746 void sv_magic(SV* sv, SV* obj, int how, const char* name, I32 namlen);
748 The C<sv> argument is a pointer to the SV that is to acquire a new magical
751 If C<sv> is not already magical, Perl uses the C<SvUPGRADE> macro to
752 set the C<SVt_PVMG> flag for the C<sv>. Perl then continues by adding
753 it to the beginning of the linked list of magical features. Any prior
754 entry of the same type of magic is deleted. Note that this can be
755 overridden, and multiple instances of the same type of magic can be
756 associated with an SV.
758 The C<name> and C<namlen> arguments are used to associate a string with
759 the magic, typically the name of a variable. C<namlen> is stored in the
760 C<mg_len> field and if C<name> is non-null and C<namlen> >= 0 a malloc'd
761 copy of the name is stored in C<mg_ptr> field.
763 The sv_magic function uses C<how> to determine which, if any, predefined
764 "Magic Virtual Table" should be assigned to the C<mg_virtual> field.
765 See the "Magic Virtual Table" section below. The C<how> argument is also
766 stored in the C<mg_type> field.
768 The C<obj> argument is stored in the C<mg_obj> field of the C<MAGIC>
769 structure. If it is not the same as the C<sv> argument, the reference
770 count of the C<obj> object is incremented. If it is the same, or if
771 the C<how> argument is "#", or if it is a NULL pointer, then C<obj> is
772 merely stored, without the reference count being incremented.
774 There is also a function to add magic to an C<HV>:
776 void hv_magic(HV *hv, GV *gv, int how);
778 This simply calls C<sv_magic> and coerces the C<gv> argument into an C<SV>.
780 To remove the magic from an SV, call the function sv_unmagic:
782 void sv_unmagic(SV *sv, int type);
784 The C<type> argument should be equal to the C<how> value when the C<SV>
785 was initially made magical.
787 =head2 Magic Virtual Tables
789 The C<mg_virtual> field in the C<MAGIC> structure is a pointer to a
790 C<MGVTBL>, which is a structure of function pointers and stands for
791 "Magic Virtual Table" to handle the various operations that might be
792 applied to that variable.
794 The C<MGVTBL> has five pointers to the following routine types:
796 int (*svt_get)(SV* sv, MAGIC* mg);
797 int (*svt_set)(SV* sv, MAGIC* mg);
798 U32 (*svt_len)(SV* sv, MAGIC* mg);
799 int (*svt_clear)(SV* sv, MAGIC* mg);
800 int (*svt_free)(SV* sv, MAGIC* mg);
802 This MGVTBL structure is set at compile-time in C<perl.h> and there are
803 currently 19 types (or 21 with overloading turned on). These different
804 structures contain pointers to various routines that perform additional
805 actions depending on which function is being called.
807 Function pointer Action taken
808 ---------------- ------------
809 svt_get Do something after the value of the SV is retrieved.
810 svt_set Do something after the SV is assigned a value.
811 svt_len Report on the SV's length.
812 svt_clear Clear something the SV represents.
813 svt_free Free any extra storage associated with the SV.
815 For instance, the MGVTBL structure called C<vtbl_sv> (which corresponds
816 to an C<mg_type> of '\0') contains:
818 { magic_get, magic_set, magic_len, 0, 0 }
820 Thus, when an SV is determined to be magical and of type '\0', if a get
821 operation is being performed, the routine C<magic_get> is called. All
822 the various routines for the various magical types begin with C<magic_>.
823 NOTE: the magic routines are not considered part of the Perl API, and may
824 not be exported by the Perl library.
826 The current kinds of Magic Virtual Tables are:
828 mg_type MGVTBL Type of magic
829 ------- ------ ----------------------------
830 \0 vtbl_sv Special scalar variable
831 A vtbl_amagic %OVERLOAD hash
832 a vtbl_amagicelem %OVERLOAD hash element
833 c (none) Holds overload table (AMT) on stash
834 B vtbl_bm Boyer-Moore (fast string search)
835 D vtbl_regdata Regex match position data (@+ and @- vars)
836 d vtbl_regdatum Regex match position data element
838 e vtbl_envelem %ENV hash element
839 f vtbl_fm Formline ('compiled' format)
840 g vtbl_mglob m//g target / study()ed string
841 I vtbl_isa @ISA array
842 i vtbl_isaelem @ISA array element
843 k vtbl_nkeys scalar(keys()) lvalue
844 L (none) Debugger %_<filename
845 l vtbl_dbline Debugger %_<filename element
846 o vtbl_collxfrm Locale transformation
847 P vtbl_pack Tied array or hash
848 p vtbl_packelem Tied array or hash element
849 q vtbl_packelem Tied scalar or handle
851 s vtbl_sigelem %SIG hash element
852 t vtbl_taint Taintedness
853 U vtbl_uvar Available for use by extensions
854 v vtbl_vec vec() lvalue
855 x vtbl_substr substr() lvalue
856 y vtbl_defelem Shadow "foreach" iterator variable /
857 smart parameter vivification
858 * vtbl_glob GV (typeglob)
859 # vtbl_arylen Array length ($#ary)
860 . vtbl_pos pos() lvalue
861 ~ (none) Available for use by extensions
863 When an uppercase and lowercase letter both exist in the table, then the
864 uppercase letter is used to represent some kind of composite type (a list
865 or a hash), and the lowercase letter is used to represent an element of
868 The '~' and 'U' magic types are defined specifically for use by
869 extensions and will not be used by perl itself. Extensions can use
870 '~' magic to 'attach' private information to variables (typically
871 objects). This is especially useful because there is no way for
872 normal perl code to corrupt this private information (unlike using
873 extra elements of a hash object).
875 Similarly, 'U' magic can be used much like tie() to call a C function
876 any time a scalar's value is used or changed. The C<MAGIC>'s
877 C<mg_ptr> field points to a C<ufuncs> structure:
880 I32 (*uf_val)(IV, SV*);
881 I32 (*uf_set)(IV, SV*);
885 When the SV is read from or written to, the C<uf_val> or C<uf_set>
886 function will be called with C<uf_index> as the first arg and a
887 pointer to the SV as the second. A simple example of how to add 'U'
888 magic is shown below. Note that the ufuncs structure is copied by
889 sv_magic, so you can safely allocate it on the stack.
897 uf.uf_val = &my_get_fn;
898 uf.uf_set = &my_set_fn;
900 sv_magic(sv, 0, 'U', (char*)&uf, sizeof(uf));
902 Note that because multiple extensions may be using '~' or 'U' magic,
903 it is important for extensions to take extra care to avoid conflict.
904 Typically only using the magic on objects blessed into the same class
905 as the extension is sufficient. For '~' magic, it may also be
906 appropriate to add an I32 'signature' at the top of the private data
909 Also note that the C<sv_set*()> and C<sv_cat*()> functions described
910 earlier do B<not> invoke 'set' magic on their targets. This must
911 be done by the user either by calling the C<SvSETMAGIC()> macro after
912 calling these functions, or by using one of the C<sv_set*_mg()> or
913 C<sv_cat*_mg()> functions. Similarly, generic C code must call the
914 C<SvGETMAGIC()> macro to invoke any 'get' magic if they use an SV
915 obtained from external sources in functions that don't handle magic.
916 See L<perlapi> for a description of these functions.
917 For example, calls to the C<sv_cat*()> functions typically need to be
918 followed by C<SvSETMAGIC()>, but they don't need a prior C<SvGETMAGIC()>
919 since their implementation handles 'get' magic.
923 MAGIC* mg_find(SV*, int type); /* Finds the magic pointer of that type */
925 This routine returns a pointer to the C<MAGIC> structure stored in the SV.
926 If the SV does not have that magical feature, C<NULL> is returned. Also,
927 if the SV is not of type SVt_PVMG, Perl may core dump.
929 int mg_copy(SV* sv, SV* nsv, const char* key, STRLEN klen);
931 This routine checks to see what types of magic C<sv> has. If the mg_type
932 field is an uppercase letter, then the mg_obj is copied to C<nsv>, but
933 the mg_type field is changed to be the lowercase letter.
935 =head2 Understanding the Magic of Tied Hashes and Arrays
937 Tied hashes and arrays are magical beasts of the 'P' magic type.
939 WARNING: As of the 5.004 release, proper usage of the array and hash
940 access functions requires understanding a few caveats. Some
941 of these caveats are actually considered bugs in the API, to be fixed
942 in later releases, and are bracketed with [MAYCHANGE] below. If
943 you find yourself actually applying such information in this section, be
944 aware that the behavior may change in the future, umm, without warning.
946 The perl tie function associates a variable with an object that implements
947 the various GET, SET etc methods. To perform the equivalent of the perl
948 tie function from an XSUB, you must mimic this behaviour. The code below
949 carries out the necessary steps - firstly it creates a new hash, and then
950 creates a second hash which it blesses into the class which will implement
951 the tie methods. Lastly it ties the two hashes together, and returns a
952 reference to the new tied hash. Note that the code below does NOT call the
953 TIEHASH method in the MyTie class -
954 see L<Calling Perl Routines from within C Programs> for details on how
965 tie = newRV_noinc((SV*)newHV());
966 stash = gv_stashpv("MyTie", TRUE);
967 sv_bless(tie, stash);
968 hv_magic(hash, tie, 'P');
969 RETVAL = newRV_noinc(hash);
973 The C<av_store> function, when given a tied array argument, merely
974 copies the magic of the array onto the value to be "stored", using
975 C<mg_copy>. It may also return NULL, indicating that the value did not
976 actually need to be stored in the array. [MAYCHANGE] After a call to
977 C<av_store> on a tied array, the caller will usually need to call
978 C<mg_set(val)> to actually invoke the perl level "STORE" method on the
979 TIEARRAY object. If C<av_store> did return NULL, a call to
980 C<SvREFCNT_dec(val)> will also be usually necessary to avoid a memory
983 The previous paragraph is applicable verbatim to tied hash access using the
984 C<hv_store> and C<hv_store_ent> functions as well.
986 C<av_fetch> and the corresponding hash functions C<hv_fetch> and
987 C<hv_fetch_ent> actually return an undefined mortal value whose magic
988 has been initialized using C<mg_copy>. Note the value so returned does not
989 need to be deallocated, as it is already mortal. [MAYCHANGE] But you will
990 need to call C<mg_get()> on the returned value in order to actually invoke
991 the perl level "FETCH" method on the underlying TIE object. Similarly,
992 you may also call C<mg_set()> on the return value after possibly assigning
993 a suitable value to it using C<sv_setsv>, which will invoke the "STORE"
994 method on the TIE object. [/MAYCHANGE]
997 In other words, the array or hash fetch/store functions don't really
998 fetch and store actual values in the case of tied arrays and hashes. They
999 merely call C<mg_copy> to attach magic to the values that were meant to be
1000 "stored" or "fetched". Later calls to C<mg_get> and C<mg_set> actually
1001 do the job of invoking the TIE methods on the underlying objects. Thus
1002 the magic mechanism currently implements a kind of lazy access to arrays
1005 Currently (as of perl version 5.004), use of the hash and array access
1006 functions requires the user to be aware of whether they are operating on
1007 "normal" hashes and arrays, or on their tied variants. The API may be
1008 changed to provide more transparent access to both tied and normal data
1009 types in future versions.
1012 You would do well to understand that the TIEARRAY and TIEHASH interfaces
1013 are mere sugar to invoke some perl method calls while using the uniform hash
1014 and array syntax. The use of this sugar imposes some overhead (typically
1015 about two to four extra opcodes per FETCH/STORE operation, in addition to
1016 the creation of all the mortal variables required to invoke the methods).
1017 This overhead will be comparatively small if the TIE methods are themselves
1018 substantial, but if they are only a few statements long, the overhead
1019 will not be insignificant.
1021 =head2 Localizing changes
1023 Perl has a very handy construction
1030 This construction is I<approximately> equivalent to
1039 The biggest difference is that the first construction would
1040 reinstate the initial value of $var, irrespective of how control exits
1041 the block: C<goto>, C<return>, C<die>/C<eval> etc. It is a little bit
1042 more efficient as well.
1044 There is a way to achieve a similar task from C via Perl API: create a
1045 I<pseudo-block>, and arrange for some changes to be automatically
1046 undone at the end of it, either explicit, or via a non-local exit (via
1047 die()). A I<block>-like construct is created by a pair of
1048 C<ENTER>/C<LEAVE> macros (see L<perlcall/"Returning a Scalar">).
1049 Such a construct may be created specially for some important localized
1050 task, or an existing one (like boundaries of enclosing Perl
1051 subroutine/block, or an existing pair for freeing TMPs) may be
1052 used. (In the second case the overhead of additional localization must
1053 be almost negligible.) Note that any XSUB is automatically enclosed in
1054 an C<ENTER>/C<LEAVE> pair.
1056 Inside such a I<pseudo-block> the following service is available:
1060 =item C<SAVEINT(int i)>
1062 =item C<SAVEIV(IV i)>
1064 =item C<SAVEI32(I32 i)>
1066 =item C<SAVELONG(long i)>
1068 These macros arrange things to restore the value of integer variable
1069 C<i> at the end of enclosing I<pseudo-block>.
1071 =item C<SAVESPTR(s)>
1073 =item C<SAVEPPTR(p)>
1075 These macros arrange things to restore the value of pointers C<s> and
1076 C<p>. C<s> must be a pointer of a type which survives conversion to
1077 C<SV*> and back, C<p> should be able to survive conversion to C<char*>
1080 =item C<SAVEFREESV(SV *sv)>
1082 The refcount of C<sv> would be decremented at the end of
1083 I<pseudo-block>. This is similar to C<sv_2mortal>, which should (?) be
1086 =item C<SAVEFREEOP(OP *op)>
1088 The C<OP *> is op_free()ed at the end of I<pseudo-block>.
1090 =item C<SAVEFREEPV(p)>
1092 The chunk of memory which is pointed to by C<p> is Safefree()ed at the
1093 end of I<pseudo-block>.
1095 =item C<SAVECLEARSV(SV *sv)>
1097 Clears a slot in the current scratchpad which corresponds to C<sv> at
1098 the end of I<pseudo-block>.
1100 =item C<SAVEDELETE(HV *hv, char *key, I32 length)>
1102 The key C<key> of C<hv> is deleted at the end of I<pseudo-block>. The
1103 string pointed to by C<key> is Safefree()ed. If one has a I<key> in
1104 short-lived storage, the corresponding string may be reallocated like
1107 SAVEDELETE(PL_defstash, savepv(tmpbuf), strlen(tmpbuf));
1109 =item C<SAVEDESTRUCTOR(DESTRUCTORFUNC_NOCONTEXT_t f, void *p)>
1111 At the end of I<pseudo-block> the function C<f> is called with the
1114 =item C<SAVEDESTRUCTOR_X(DESTRUCTORFUNC_t f, void *p)>
1116 At the end of I<pseudo-block> the function C<f> is called with the
1117 implicit context argument (if any), and C<p>.
1119 =item C<SAVESTACK_POS()>
1121 The current offset on the Perl internal stack (cf. C<SP>) is restored
1122 at the end of I<pseudo-block>.
1126 The following API list contains functions, thus one needs to
1127 provide pointers to the modifiable data explicitly (either C pointers,
1128 or Perlish C<GV *>s). Where the above macros take C<int>, a similar
1129 function takes C<int *>.
1133 =item C<SV* save_scalar(GV *gv)>
1135 Equivalent to Perl code C<local $gv>.
1137 =item C<AV* save_ary(GV *gv)>
1139 =item C<HV* save_hash(GV *gv)>
1141 Similar to C<save_scalar>, but localize C<@gv> and C<%gv>.
1143 =item C<void save_item(SV *item)>
1145 Duplicates the current value of C<SV>, on the exit from the current
1146 C<ENTER>/C<LEAVE> I<pseudo-block> will restore the value of C<SV>
1147 using the stored value.
1149 =item C<void save_list(SV **sarg, I32 maxsarg)>
1151 A variant of C<save_item> which takes multiple arguments via an array
1152 C<sarg> of C<SV*> of length C<maxsarg>.
1154 =item C<SV* save_svref(SV **sptr)>
1156 Similar to C<save_scalar>, but will reinstate a C<SV *>.
1158 =item C<void save_aptr(AV **aptr)>
1160 =item C<void save_hptr(HV **hptr)>
1162 Similar to C<save_svref>, but localize C<AV *> and C<HV *>.
1166 The C<Alias> module implements localization of the basic types within the
1167 I<caller's scope>. People who are interested in how to localize things in
1168 the containing scope should take a look there too.
1172 =head2 XSUBs and the Argument Stack
1174 The XSUB mechanism is a simple way for Perl programs to access C subroutines.
1175 An XSUB routine will have a stack that contains the arguments from the Perl
1176 program, and a way to map from the Perl data structures to a C equivalent.
1178 The stack arguments are accessible through the C<ST(n)> macro, which returns
1179 the C<n>'th stack argument. Argument 0 is the first argument passed in the
1180 Perl subroutine call. These arguments are C<SV*>, and can be used anywhere
1183 Most of the time, output from the C routine can be handled through use of
1184 the RETVAL and OUTPUT directives. However, there are some cases where the
1185 argument stack is not already long enough to handle all the return values.
1186 An example is the POSIX tzname() call, which takes no arguments, but returns
1187 two, the local time zone's standard and summer time abbreviations.
1189 To handle this situation, the PPCODE directive is used and the stack is
1190 extended using the macro:
1194 where C<SP> is the macro that represents the local copy of the stack pointer,
1195 and C<num> is the number of elements the stack should be extended by.
1197 Now that there is room on the stack, values can be pushed on it using the
1198 macros to push IVs, doubles, strings, and SV pointers respectively:
1205 And now the Perl program calling C<tzname>, the two values will be assigned
1208 ($standard_abbrev, $summer_abbrev) = POSIX::tzname;
1210 An alternate (and possibly simpler) method to pushing values on the stack is
1218 These macros automatically adjust the stack for you, if needed. Thus, you
1219 do not need to call C<EXTEND> to extend the stack.
1221 For more information, consult L<perlxs> and L<perlxstut>.
1223 =head2 Calling Perl Routines from within C Programs
1225 There are four routines that can be used to call a Perl subroutine from
1226 within a C program. These four are:
1228 I32 call_sv(SV*, I32);
1229 I32 call_pv(const char*, I32);
1230 I32 call_method(const char*, I32);
1231 I32 call_argv(const char*, I32, register char**);
1233 The routine most often used is C<call_sv>. The C<SV*> argument
1234 contains either the name of the Perl subroutine to be called, or a
1235 reference to the subroutine. The second argument consists of flags
1236 that control the context in which the subroutine is called, whether
1237 or not the subroutine is being passed arguments, how errors should be
1238 trapped, and how to treat return values.
1240 All four routines return the number of arguments that the subroutine returned
1243 These routines used to be called C<perl_call_sv> etc., before Perl v5.6.0,
1244 but those names are now deprecated; macros of the same name are provided for
1247 When using any of these routines (except C<call_argv>), the programmer
1248 must manipulate the Perl stack. These include the following macros and
1263 For a detailed description of calling conventions from C to Perl,
1264 consult L<perlcall>.
1266 =head2 Memory Allocation
1268 All memory meant to be used with the Perl API functions should be manipulated
1269 using the macros described in this section. The macros provide the necessary
1270 transparency between differences in the actual malloc implementation that is
1273 It is suggested that you enable the version of malloc that is distributed
1274 with Perl. It keeps pools of various sizes of unallocated memory in
1275 order to satisfy allocation requests more quickly. However, on some
1276 platforms, it may cause spurious malloc or free errors.
1278 New(x, pointer, number, type);
1279 Newc(x, pointer, number, type, cast);
1280 Newz(x, pointer, number, type);
1282 These three macros are used to initially allocate memory.
1284 The first argument C<x> was a "magic cookie" that was used to keep track
1285 of who called the macro, to help when debugging memory problems. However,
1286 the current code makes no use of this feature (most Perl developers now
1287 use run-time memory checkers), so this argument can be any number.
1289 The second argument C<pointer> should be the name of a variable that will
1290 point to the newly allocated memory.
1292 The third and fourth arguments C<number> and C<type> specify how many of
1293 the specified type of data structure should be allocated. The argument
1294 C<type> is passed to C<sizeof>. The final argument to C<Newc>, C<cast>,
1295 should be used if the C<pointer> argument is different from the C<type>
1298 Unlike the C<New> and C<Newc> macros, the C<Newz> macro calls C<memzero>
1299 to zero out all the newly allocated memory.
1301 Renew(pointer, number, type);
1302 Renewc(pointer, number, type, cast);
1305 These three macros are used to change a memory buffer size or to free a
1306 piece of memory no longer needed. The arguments to C<Renew> and C<Renewc>
1307 match those of C<New> and C<Newc> with the exception of not needing the
1308 "magic cookie" argument.
1310 Move(source, dest, number, type);
1311 Copy(source, dest, number, type);
1312 Zero(dest, number, type);
1314 These three macros are used to move, copy, or zero out previously allocated
1315 memory. The C<source> and C<dest> arguments point to the source and
1316 destination starting points. Perl will move, copy, or zero out C<number>
1317 instances of the size of the C<type> data structure (using the C<sizeof>
1320 Here is a handy table of equivalents between ordinary C and Perl's
1321 memory abstraction layer:
1332 strndup savepvn (Hey, strndup doesn't exist!)
1333 memcpy/*(struct foo *) StructCopy
1337 The most recent development releases of Perl has been experimenting with
1338 removing Perl's dependency on the "normal" standard I/O suite and allowing
1339 other stdio implementations to be used. This involves creating a new
1340 abstraction layer that then calls whichever implementation of stdio Perl
1341 was compiled with. All XSUBs should now use the functions in the PerlIO
1342 abstraction layer and not make any assumptions about what kind of stdio
1345 For a complete description of the PerlIO abstraction, consult L<perlapio>.
1347 =head2 Putting a C value on Perl stack
1349 A lot of opcodes (this is an elementary operation in the internal perl
1350 stack machine) put an SV* on the stack. However, as an optimization
1351 the corresponding SV is (usually) not recreated each time. The opcodes
1352 reuse specially assigned SVs (I<target>s) which are (as a corollary)
1353 not constantly freed/created.
1355 Each of the targets is created only once (but see
1356 L<Scratchpads and recursion> below), and when an opcode needs to put
1357 an integer, a double, or a string on stack, it just sets the
1358 corresponding parts of its I<target> and puts the I<target> on stack.
1360 The macro to put this target on stack is C<PUSHTARG>, and it is
1361 directly used in some opcodes, as well as indirectly in zillions of
1362 others, which use it via C<(X)PUSH[pni]>.
1366 The question remains on when the SVs which are I<target>s for opcodes
1367 are created. The answer is that they are created when the current unit --
1368 a subroutine or a file (for opcodes for statements outside of
1369 subroutines) -- is compiled. During this time a special anonymous Perl
1370 array is created, which is called a scratchpad for the current
1373 A scratchpad keeps SVs which are lexicals for the current unit and are
1374 targets for opcodes. One can deduce that an SV lives on a scratchpad
1375 by looking on its flags: lexicals have C<SVs_PADMY> set, and
1376 I<target>s have C<SVs_PADTMP> set.
1378 The correspondence between OPs and I<target>s is not 1-to-1. Different
1379 OPs in the compile tree of the unit can use the same target, if this
1380 would not conflict with the expected life of the temporary.
1382 =head2 Scratchpads and recursion
1384 In fact it is not 100% true that a compiled unit contains a pointer to
1385 the scratchpad AV. In fact it contains a pointer to an AV of
1386 (initially) one element, and this element is the scratchpad AV. Why do
1387 we need an extra level of indirection?
1389 The answer is B<recursion>, and maybe (sometime soon) B<threads>. Both
1390 these can create several execution pointers going into the same
1391 subroutine. For the subroutine-child not write over the temporaries
1392 for the subroutine-parent (lifespan of which covers the call to the
1393 child), the parent and the child should have different
1394 scratchpads. (I<And> the lexicals should be separate anyway!)
1396 So each subroutine is born with an array of scratchpads (of length 1).
1397 On each entry to the subroutine it is checked that the current
1398 depth of the recursion is not more than the length of this array, and
1399 if it is, new scratchpad is created and pushed into the array.
1401 The I<target>s on this scratchpad are C<undef>s, but they are already
1402 marked with correct flags.
1404 =head1 Compiled code
1408 Here we describe the internal form your code is converted to by
1409 Perl. Start with a simple example:
1413 This is converted to a tree similar to this one:
1421 (but slightly more complicated). This tree reflects the way Perl
1422 parsed your code, but has nothing to do with the execution order.
1423 There is an additional "thread" going through the nodes of the tree
1424 which shows the order of execution of the nodes. In our simplified
1425 example above it looks like:
1427 $b ---> $c ---> + ---> $a ---> assign-to
1429 But with the actual compile tree for C<$a = $b + $c> it is different:
1430 some nodes I<optimized away>. As a corollary, though the actual tree
1431 contains more nodes than our simplified example, the execution order
1432 is the same as in our example.
1434 =head2 Examining the tree
1436 If you have your perl compiled for debugging (usually done with C<-D
1437 optimize=-g> on C<Configure> command line), you may examine the
1438 compiled tree by specifying C<-Dx> on the Perl command line. The
1439 output takes several lines per node, and for C<$b+$c> it looks like
1444 FLAGS = (SCALAR,KIDS)
1446 TYPE = null ===> (4)
1448 FLAGS = (SCALAR,KIDS)
1450 3 TYPE = gvsv ===> 4
1456 TYPE = null ===> (5)
1458 FLAGS = (SCALAR,KIDS)
1460 4 TYPE = gvsv ===> 5
1466 This tree has 5 nodes (one per C<TYPE> specifier), only 3 of them are
1467 not optimized away (one per number in the left column). The immediate
1468 children of the given node correspond to C<{}> pairs on the same level
1469 of indentation, thus this listing corresponds to the tree:
1477 The execution order is indicated by C<===E<gt>> marks, thus it is C<3
1478 4 5 6> (node C<6> is not included into above listing), i.e.,
1479 C<gvsv gvsv add whatever>.
1481 =head2 Compile pass 1: check routines
1483 The tree is created by the compiler while I<yacc> code feeds it
1484 the constructions it recognizes. Since I<yacc> works bottom-up, so does
1485 the first pass of perl compilation.
1487 What makes this pass interesting for perl developers is that some
1488 optimization may be performed on this pass. This is optimization by
1489 so-called "check routines". The correspondence between node names
1490 and corresponding check routines is described in F<opcode.pl> (do not
1491 forget to run C<make regen_headers> if you modify this file).
1493 A check routine is called when the node is fully constructed except
1494 for the execution-order thread. Since at this time there are no
1495 back-links to the currently constructed node, one can do most any
1496 operation to the top-level node, including freeing it and/or creating
1497 new nodes above/below it.
1499 The check routine returns the node which should be inserted into the
1500 tree (if the top-level node was not modified, check routine returns
1503 By convention, check routines have names C<ck_*>. They are usually
1504 called from C<new*OP> subroutines (or C<convert>) (which in turn are
1505 called from F<perly.y>).
1507 =head2 Compile pass 1a: constant folding
1509 Immediately after the check routine is called the returned node is
1510 checked for being compile-time executable. If it is (the value is
1511 judged to be constant) it is immediately executed, and a I<constant>
1512 node with the "return value" of the corresponding subtree is
1513 substituted instead. The subtree is deleted.
1515 If constant folding was not performed, the execution-order thread is
1518 =head2 Compile pass 2: context propagation
1520 When a context for a part of compile tree is known, it is propagated
1521 down through the tree. At this time the context can have 5 values
1522 (instead of 2 for runtime context): void, boolean, scalar, list, and
1523 lvalue. In contrast with the pass 1 this pass is processed from top
1524 to bottom: a node's context determines the context for its children.
1526 Additional context-dependent optimizations are performed at this time.
1527 Since at this moment the compile tree contains back-references (via
1528 "thread" pointers), nodes cannot be free()d now. To allow
1529 optimized-away nodes at this stage, such nodes are null()ified instead
1530 of free()ing (i.e. their type is changed to OP_NULL).
1532 =head2 Compile pass 3: peephole optimization
1534 After the compile tree for a subroutine (or for an C<eval> or a file)
1535 is created, an additional pass over the code is performed. This pass
1536 is neither top-down or bottom-up, but in the execution order (with
1537 additional complications for conditionals). These optimizations are
1538 done in the subroutine peep(). Optimizations performed at this stage
1539 are subject to the same restrictions as in the pass 2.
1541 =head1 How multiple interpreters and concurrency are supported
1543 =head2 Background and PERL_IMPLICIT_CONTEXT
1545 The Perl interpreter can be regarded as a closed box: it has an API
1546 for feeding it code or otherwise making it do things, but it also has
1547 functions for its own use. This smells a lot like an object, and
1548 there are ways for you to build Perl so that you can have multiple
1549 interpreters, with one interpreter represented either as a C++ object,
1550 a C structure, or inside a thread. The thread, the C structure, or
1551 the C++ object will contain all the context, the state of that
1554 Three macros control the major Perl build flavors: MULTIPLICITY,
1555 USE_THREADS and PERL_OBJECT. The MULTIPLICITY build has a C structure
1556 that packages all the interpreter state, there is a similar thread-specific
1557 data structure under USE_THREADS, and the PERL_OBJECT build has a C++
1558 class to maintain interpreter state. In all three cases,
1559 PERL_IMPLICIT_CONTEXT is also normally defined, and enables the
1560 support for passing in a "hidden" first argument that represents all three
1563 All this obviously requires a way for the Perl internal functions to be
1564 C++ methods, subroutines taking some kind of structure as the first
1565 argument, or subroutines taking nothing as the first argument. To
1566 enable these three very different ways of building the interpreter,
1567 the Perl source (as it does in so many other situations) makes heavy
1568 use of macros and subroutine naming conventions.
1570 First problem: deciding which functions will be public API functions and
1571 which will be private. All functions whose names begin C<S_> are private
1572 (think "S" for "secret" or "static"). All other functions begin with
1573 "Perl_", but just because a function begins with "Perl_" does not mean it is
1574 part of the API. (See L</Internal Functions>.) The easiest way to be B<sure> a
1575 function is part of the API is to find its entry in L<perlapi>.
1576 If it exists in L<perlapi>, it's part of the API. If it doesn't, and you
1577 think it should be (i.e., you need it for your extension), send mail via
1578 L<perlbug> explaining why you think it should be.
1580 Second problem: there must be a syntax so that the same subroutine
1581 declarations and calls can pass a structure as their first argument,
1582 or pass nothing. To solve this, the subroutines are named and
1583 declared in a particular way. Here's a typical start of a static
1584 function used within the Perl guts:
1587 S_incline(pTHX_ char *s)
1589 STATIC becomes "static" in C, and is #define'd to nothing in C++.
1591 A public function (i.e. part of the internal API, but not necessarily
1592 sanctioned for use in extensions) begins like this:
1595 Perl_sv_setsv(pTHX_ SV* dsv, SV* ssv)
1597 C<pTHX_> is one of a number of macros (in perl.h) that hide the
1598 details of the interpreter's context. THX stands for "thread", "this",
1599 or "thingy", as the case may be. (And no, George Lucas is not involved. :-)
1600 The first character could be 'p' for a B<p>rototype, 'a' for B<a>rgument,
1601 or 'd' for B<d>eclaration.
1603 When Perl is built without PERL_IMPLICIT_CONTEXT, there is no first
1604 argument containing the interpreter's context. The trailing underscore
1605 in the pTHX_ macro indicates that the macro expansion needs a comma
1606 after the context argument because other arguments follow it. If
1607 PERL_IMPLICIT_CONTEXT is not defined, pTHX_ will be ignored, and the
1608 subroutine is not prototyped to take the extra argument. The form of the
1609 macro without the trailing underscore is used when there are no additional
1612 When a core function calls another, it must pass the context. This
1613 is normally hidden via macros. Consider C<sv_setsv>. It expands
1614 something like this:
1616 ifdef PERL_IMPLICIT_CONTEXT
1617 define sv_setsv(a,b) Perl_sv_setsv(aTHX_ a, b)
1618 /* can't do this for vararg functions, see below */
1620 define sv_setsv Perl_sv_setsv
1623 This works well, and means that XS authors can gleefully write:
1627 and still have it work under all the modes Perl could have been
1630 Under PERL_OBJECT in the core, that will translate to either:
1632 CPerlObj::Perl_sv_setsv(foo,bar); # in CPerlObj functions,
1633 # C++ takes care of 'this'
1636 pPerl->Perl_sv_setsv(foo,bar); # in truly static functions,
1639 Under PERL_OBJECT in extensions (aka PERL_CAPI), or under
1640 MULTIPLICITY/USE_THREADS w/ PERL_IMPLICIT_CONTEXT in both core
1641 and extensions, it will be:
1643 Perl_sv_setsv(aTHX_ foo, bar); # the canonical Perl "API"
1644 # for all build flavors
1646 This doesn't work so cleanly for varargs functions, though, as macros
1647 imply that the number of arguments is known in advance. Instead we
1648 either need to spell them out fully, passing C<aTHX_> as the first
1649 argument (the Perl core tends to do this with functions like
1650 Perl_warner), or use a context-free version.
1652 The context-free version of Perl_warner is called
1653 Perl_warner_nocontext, and does not take the extra argument. Instead
1654 it does dTHX; to get the context from thread-local storage. We
1655 C<#define warner Perl_warner_nocontext> so that extensions get source
1656 compatibility at the expense of performance. (Passing an arg is
1657 cheaper than grabbing it from thread-local storage.)
1659 You can ignore [pad]THX[xo] when browsing the Perl headers/sources.
1660 Those are strictly for use within the core. Extensions and embedders
1661 need only be aware of [pad]THX.
1663 =head2 How do I use all this in extensions?
1665 When Perl is built with PERL_IMPLICIT_CONTEXT, extensions that call
1666 any functions in the Perl API will need to pass the initial context
1667 argument somehow. The kicker is that you will need to write it in
1668 such a way that the extension still compiles when Perl hasn't been
1669 built with PERL_IMPLICIT_CONTEXT enabled.
1671 There are three ways to do this. First, the easy but inefficient way,
1672 which is also the default, in order to maintain source compatibility
1673 with extensions: whenever XSUB.h is #included, it redefines the aTHX
1674 and aTHX_ macros to call a function that will return the context.
1675 Thus, something like:
1679 in your extension will translate to this when PERL_IMPLICIT_CONTEXT is
1682 Perl_sv_setsv(Perl_get_context(), asv, bsv);
1684 or to this otherwise:
1686 Perl_sv_setsv(asv, bsv);
1688 You have to do nothing new in your extension to get this; since
1689 the Perl library provides Perl_get_context(), it will all just
1692 The second, more efficient way is to use the following template for
1695 #define PERL_NO_GET_CONTEXT /* we want efficiency */
1700 static my_private_function(int arg1, int arg2);
1703 my_private_function(int arg1, int arg2)
1705 dTHX; /* fetch context */
1706 ... call many Perl API functions ...
1711 MODULE = Foo PACKAGE = Foo
1719 my_private_function(arg, 10);
1721 Note that the only two changes from the normal way of writing an
1722 extension is the addition of a C<#define PERL_NO_GET_CONTEXT> before
1723 including the Perl headers, followed by a C<dTHX;> declaration at
1724 the start of every function that will call the Perl API. (You'll
1725 know which functions need this, because the C compiler will complain
1726 that there's an undeclared identifier in those functions.) No changes
1727 are needed for the XSUBs themselves, because the XS() macro is
1728 correctly defined to pass in the implicit context if needed.
1730 The third, even more efficient way is to ape how it is done within
1734 #define PERL_NO_GET_CONTEXT /* we want efficiency */
1739 /* pTHX_ only needed for functions that call Perl API */
1740 static my_private_function(pTHX_ int arg1, int arg2);
1743 my_private_function(pTHX_ int arg1, int arg2)
1745 /* dTHX; not needed here, because THX is an argument */
1746 ... call Perl API functions ...
1751 MODULE = Foo PACKAGE = Foo
1759 my_private_function(aTHX_ arg, 10);
1761 This implementation never has to fetch the context using a function
1762 call, since it is always passed as an extra argument. Depending on
1763 your needs for simplicity or efficiency, you may mix the previous
1764 two approaches freely.
1766 Never add a comma after C<pTHX> yourself--always use the form of the
1767 macro with the underscore for functions that take explicit arguments,
1768 or the form without the argument for functions with no explicit arguments.
1770 =head2 Future Plans and PERL_IMPLICIT_SYS
1772 Just as PERL_IMPLICIT_CONTEXT provides a way to bundle up everything
1773 that the interpreter knows about itself and pass it around, so too are
1774 there plans to allow the interpreter to bundle up everything it knows
1775 about the environment it's running on. This is enabled with the
1776 PERL_IMPLICIT_SYS macro. Currently it only works with PERL_OBJECT,
1777 but is mostly there for MULTIPLICITY and USE_THREADS (see inside
1780 This allows the ability to provide an extra pointer (called the "host"
1781 environment) for all the system calls. This makes it possible for
1782 all the system stuff to maintain their own state, broken down into
1783 seven C structures. These are thin wrappers around the usual system
1784 calls (see win32/perllib.c) for the default perl executable, but for a
1785 more ambitious host (like the one that would do fork() emulation) all
1786 the extra work needed to pretend that different interpreters are
1787 actually different "processes", would be done here.
1789 The Perl engine/interpreter and the host are orthogonal entities.
1790 There could be one or more interpreters in a process, and one or
1791 more "hosts", with free association between them.
1793 =head1 Internal Functions
1795 All of Perl's internal functions which will be exposed to the outside
1796 world are be prefixed by C<Perl_> so that they will not conflict with XS
1797 functions or functions used in a program in which Perl is embedded.
1798 Similarly, all global variables begin with C<PL_>. (By convention,
1799 static functions start with C<S_>)
1801 Inside the Perl core, you can get at the functions either with or
1802 without the C<Perl_> prefix, thanks to a bunch of defines that live in
1803 F<embed.h>. This header file is generated automatically from
1804 F<embed.pl>. F<embed.pl> also creates the prototyping header files for
1805 the internal functions, generates the documentation and a lot of other
1806 bits and pieces. It's important that when you add a new function to the
1807 core or change an existing one, you change the data in the table at the
1808 end of F<embed.pl> as well. Here's a sample entry from that table:
1810 Apd |SV** |av_fetch |AV* ar|I32 key|I32 lval
1812 The second column is the return type, the third column the name. Columns
1813 after that are the arguments. The first column is a set of flags:
1819 This function is a part of the public API.
1823 This function has a C<Perl_> prefix; ie, it is defined as C<Perl_av_fetch>
1827 This function has documentation using the C<apidoc> feature which we'll
1828 look at in a second.
1832 Other available flags are:
1838 This is a static function and is defined as C<S_whatever>.
1842 This does not use C<aTHX_> and C<pTHX> to pass interpreter context. (See
1843 L<perlguts/Background and PERL_IMPLICIT_CONTEXT>.)
1847 This function never returns; C<croak>, C<exit> and friends.
1851 This function takes a variable number of arguments, C<printf> style.
1852 The argument list should end with C<...>, like this:
1854 Afprd |void |croak |const char* pat|...
1858 This function is part of the experimental development API, and may change
1859 or disappear without notice.
1863 This function should not have a compatibility macro to define, say,
1864 C<Perl_parse> to C<parse>. It must be called as C<Perl_parse>.
1868 This function is not a member of C<CPerlObj>. If you don't know
1869 what this means, don't use it.
1873 This function isn't exported out of the Perl core.
1877 If you edit F<embed.pl>, you will need to run C<make regen_headers> to
1878 force a rebuild of F<embed.h> and other auto-generated files.
1880 =head2 Formatted Printing of IVs, UVs, and NVs
1882 If you are printing IVs, UVs, or NVS instead of the stdio(3) style
1883 formatting codes like C<%d>, C<%ld>, C<%f>, you should use the
1884 following macros for portability
1889 UVxf UV in hexadecimal
1894 These will take care of 64-bit integers and long doubles.
1897 printf("IV is %"IVdf"\n", iv);
1899 The IVdf will expand to whatever is the correct format for the IVs.
1901 If you are printing addresses of pointers, use UVxf combined
1902 with PTR2UV(), do not use %lx or %p.
1904 =head2 Pointer-To-Integer and Integer-To-Pointer
1906 Because pointer size does not necessarily equal integer size,
1907 use the follow macros to do it right.
1912 INT2PTR(pointertotype, integer)
1917 SV *sv = INT2PTR(SV*, iv);
1924 =head2 Source Documentation
1926 There's an effort going on to document the internal functions and
1927 automatically produce reference manuals from them - L<perlapi> is one
1928 such manual which details all the functions which are available to XS
1929 writers. L<perlintern> is the autogenerated manual for the functions
1930 which are not part of the API and are supposedly for internal use only.
1932 Source documentation is created by putting POD comments into the C
1936 =for apidoc sv_setiv
1938 Copies an integer into the given SV. Does not handle 'set' magic. See
1944 Please try and supply some documentation if you add functions to the
1947 =head1 Unicode Support
1949 Perl 5.6.0 introduced Unicode support. It's important for porters and XS
1950 writers to understand this support and make sure that the code they
1951 write does not corrupt Unicode data.
1953 =head2 What B<is> Unicode, anyway?
1955 In the olden, less enlightened times, we all used to use ASCII. Most of
1956 us did, anyway. The big problem with ASCII is that it's American. Well,
1957 no, that's not actually the problem; the problem is that it's not
1958 particularly useful for people who don't use the Roman alphabet. What
1959 used to happen was that particular languages would stick their own
1960 alphabet in the upper range of the sequence, between 128 and 255. Of
1961 course, we then ended up with plenty of variants that weren't quite
1962 ASCII, and the whole point of it being a standard was lost.
1964 Worse still, if you've got a language like Chinese or
1965 Japanese that has hundreds or thousands of characters, then you really
1966 can't fit them into a mere 256, so they had to forget about ASCII
1967 altogether, and build their own systems using pairs of numbers to refer
1970 To fix this, some people formed Unicode, Inc. and
1971 produced a new character set containing all the characters you can
1972 possibly think of and more. There are several ways of representing these
1973 characters, and the one Perl uses is called UTF8. UTF8 uses
1974 a variable number of bytes to represent a character, instead of just
1975 one. You can learn more about Unicode at http://www.unicode.org/
1977 =head2 How can I recognise a UTF8 string?
1979 You can't. This is because UTF8 data is stored in bytes just like
1980 non-UTF8 data. The Unicode character 200, (C<0xC8> for you hex types)
1981 capital E with a grave accent, is represented by the two bytes
1982 C<v196.172>. Unfortunately, the non-Unicode string C<chr(196).chr(172)>
1983 has that byte sequence as well. So you can't tell just by looking - this
1984 is what makes Unicode input an interesting problem.
1986 The API function C<is_utf8_string> can help; it'll tell you if a string
1987 contains only valid UTF8 characters. However, it can't do the work for
1988 you. On a character-by-character basis, C<is_utf8_char> will tell you
1989 whether the current character in a string is valid UTF8.
1991 =head2 How does UTF8 represent Unicode characters?
1993 As mentioned above, UTF8 uses a variable number of bytes to store a
1994 character. Characters with values 1...128 are stored in one byte, just
1995 like good ol' ASCII. Character 129 is stored as C<v194.129>; this
1996 continues up to character 191, which is C<v194.191>. Now we've run out of
1997 bits (191 is binary C<10111111>) so we move on; 192 is C<v195.128>. And
1998 so it goes on, moving to three bytes at character 2048.
2000 Assuming you know you're dealing with a UTF8 string, you can find out
2001 how long the first character in it is with the C<UTF8SKIP> macro:
2003 char *utf = "\305\233\340\240\201";
2006 len = UTF8SKIP(utf); /* len is 2 here */
2008 len = UTF8SKIP(utf); /* len is 3 here */
2010 Another way to skip over characters in a UTF8 string is to use
2011 C<utf8_hop>, which takes a string and a number of characters to skip
2012 over. You're on your own about bounds checking, though, so don't use it
2015 All bytes in a multi-byte UTF8 character will have the high bit set, so
2016 you can test if you need to do something special with this character
2022 /* Must treat this as UTF8 */
2023 uv = utf8_to_uv(utf);
2025 /* OK to treat this character as a byte */
2028 You can also see in that example that we use C<utf8_to_uv> to get the
2029 value of the character; the inverse function C<uv_to_utf8> is available
2030 for putting a UV into UTF8:
2033 /* Must treat this as UTF8 */
2034 utf8 = uv_to_utf8(utf8, uv);
2036 /* OK to treat this character as a byte */
2039 You B<must> convert characters to UVs using the above functions if
2040 you're ever in a situation where you have to match UTF8 and non-UTF8
2041 characters. You may not skip over UTF8 characters in this case. If you
2042 do this, you'll lose the ability to match hi-bit non-UTF8 characters;
2043 for instance, if your UTF8 string contains C<v196.172>, and you skip
2044 that character, you can never match a C<chr(200)> in a non-UTF8 string.
2047 =head2 How does Perl store UTF8 strings?
2049 Currently, Perl deals with Unicode strings and non-Unicode strings
2050 slightly differently. If a string has been identified as being UTF-8
2051 encoded, Perl will set a flag in the SV, C<SVf_UTF8>. You can check and
2052 manipulate this flag with the following macros:
2058 This flag has an important effect on Perl's treatment of the string: if
2059 Unicode data is not properly distinguished, regular expressions,
2060 C<length>, C<substr> and other string handling operations will have
2061 undesirable results.
2063 The problem comes when you have, for instance, a string that isn't
2064 flagged is UTF8, and contains a byte sequence that could be UTF8 -
2065 especially when combining non-UTF8 and UTF8 strings.
2067 Never forget that the C<SVf_UTF8> flag is separate to the PV value; you
2068 need be sure you don't accidentally knock it off while you're
2069 manipulating SVs. More specifically, you cannot expect to do this:
2078 nsv = newSVpvn(p, len);
2080 The C<char*> string does not tell you the whole story, and you can't
2081 copy or reconstruct an SV just by copying the string value. Check if the
2082 old SV has the UTF8 flag set, and act accordingly:
2086 nsv = newSVpvn(p, len);
2090 In fact, your C<frobnicate> function should be made aware of whether or
2091 not it's dealing with UTF8 data, so that it can handle the string
2094 =head2 How do I convert a string to UTF8?
2096 If you're mixing UTF8 and non-UTF8 strings, you might find it necessary
2097 to upgrade one of the strings to UTF8. If you've got an SV, the easiest
2100 sv_utf8_upgrade(sv);
2102 However, you must not do this, for example:
2105 sv_utf8_upgrade(left);
2107 If you do this in a binary operator, you will actually change one of the
2108 strings that came into the operator, and, while it shouldn't be noticeable
2109 by the end user, it can cause problems.
2111 Instead, C<bytes_to_utf8> will give you a UTF8-encoded B<copy> of its
2112 string argument. This is useful for having the data available for
2113 comparisons and so on, without harming the original SV. There's also
2114 C<utf8_to_bytes> to go the other way, but naturally, this will fail if
2115 the string contains any characters above 255 that can't be represented
2118 =head2 Is there anything else I need to know?
2120 Not really. Just remember these things:
2126 There's no way to tell if a string is UTF8 or not. You can tell if an SV
2127 is UTF8 by looking at is C<SvUTF8> flag. Don't forget to set the flag if
2128 something should be UTF8. Treat the flag as part of the PV, even though
2129 it's not - if you pass on the PV to somewhere, pass on the flag too.
2133 If a string is UTF8, B<always> use C<utf8_to_uv> to get at the value,
2134 unless C<!(*s & 0x80)> in which case you can use C<*s>.
2138 When writing to a UTF8 string, B<always> use C<uv_to_utf8>, unless
2139 C<uv < 0x80> in which case you can use C<*s = uv>.
2143 Mixing UTF8 and non-UTF8 strings is tricky. Use C<bytes_to_utf8> to get
2144 a new string which is UTF8 encoded. There are tricks you can use to
2145 delay deciding whether you need to use a UTF8 string until you get to a
2146 high character - C<HALF_UPGRADE> is one of those.
2152 Until May 1997, this document was maintained by Jeff Okamoto
2153 <okamoto@corp.hp.com>. It is now maintained as part of Perl itself
2154 by the Perl 5 Porters <perl5-porters@perl.org>.
2156 With lots of help and suggestions from Dean Roehrich, Malcolm Beattie,
2157 Andreas Koenig, Paul Hudson, Ilya Zakharevich, Paul Marquess, Neil
2158 Bowers, Matthew Green, Tim Bunce, Spider Boardman, Ulrich Pfeifer,
2159 Stephen McCamant, and Gurusamy Sarathy.
2161 API Listing originally by Dean Roehrich <roehrich@cray.com>.
2163 Modifications to autogenerate the API listing (L<perlapi>) by Benjamin
2168 perlapi(1), perlintern(1), perlxs(1), perlembed(1)