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_vsetpvfn(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_vsetpvfn> 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 STRLEN is an integer type (Size_t, usually defined as size_t in
80 config.h) guaranteed to be large enough to represent the size of
81 any string that perl can handle.
83 The C<sv_set*()> functions are not generic enough to operate on values
84 that have "magic". See L<Magic Virtual Tables> later in this document.
86 All SVs that contain strings should be terminated with a NUL character.
87 If it is not NUL-terminated there is a risk of
88 core dumps and corruptions from code which passes the string to C
89 functions or system calls which expect a NUL-terminated string.
90 Perl's own functions typically add a trailing NUL for this reason.
91 Nevertheless, you should be very careful when you pass a string stored
92 in an SV to a C function or system call.
94 To access the actual value that an SV points to, you can use the macros:
102 which will automatically coerce the actual scalar type into an IV, UV, double,
105 In the C<SvPV> macro, the length of the string returned is placed into the
106 variable C<len> (this is a macro, so you do I<not> use C<&len>). If you do
107 not care what the length of the data is, use the C<SvPV_nolen> macro.
108 Historically the C<SvPV> macro with the global variable C<PL_na> has been
109 used in this case. But that can be quite inefficient because C<PL_na> must
110 be accessed in thread-local storage in threaded Perl. In any case, remember
111 that Perl allows arbitrary strings of data that may both contain NULs and
112 might not be terminated by a NUL.
114 Also remember that C doesn't allow you to safely say C<foo(SvPV(s, len),
115 len);>. It might work with your compiler, but it won't work for everyone.
116 Break this sort of statement up into separate assignments:
124 If you want to know if the scalar value is TRUE, you can use:
128 Although Perl will automatically grow strings for you, if you need to force
129 Perl to allocate more memory for your SV, you can use the macro
131 SvGROW(SV*, STRLEN newlen)
133 which will determine if more memory needs to be allocated. If so, it will
134 call the function C<sv_grow>. Note that C<SvGROW> can only increase, not
135 decrease, the allocated memory of an SV and that it does not automatically
136 add a byte for the a trailing NUL (perl's own string functions typically do
137 C<SvGROW(sv, len + 1)>).
139 If you have an SV and want to know what kind of data Perl thinks is stored
140 in it, you can use the following macros to check the type of SV you have.
146 You can get and set the current length of the string stored in an SV with
147 the following macros:
150 SvCUR_set(SV*, I32 val)
152 You can also get a pointer to the end of the string stored in the SV
157 But note that these last three macros are valid only if C<SvPOK()> is true.
159 If you want to append something to the end of string stored in an C<SV*>,
160 you can use the following functions:
162 void sv_catpv(SV*, const char*);
163 void sv_catpvn(SV*, const char*, STRLEN);
164 void sv_catpvf(SV*, const char*, ...);
165 void sv_vcatpvfn(SV*, const char*, STRLEN, va_list *, SV **, I32, bool);
166 void sv_catsv(SV*, SV*);
168 The first function calculates the length of the string to be appended by
169 using C<strlen>. In the second, you specify the length of the string
170 yourself. The third function processes its arguments like C<sprintf> and
171 appends the formatted output. The fourth function works like C<vsprintf>.
172 You can specify the address and length of an array of SVs instead of the
173 va_list argument. The fifth function extends the string stored in the first
174 SV with the string stored in the second SV. It also forces the second SV
175 to be interpreted as a string.
177 The C<sv_cat*()> functions are not generic enough to operate on values that
178 have "magic". See L<Magic Virtual Tables> later in this document.
180 If you know the name of a scalar variable, you can get a pointer to its SV
181 by using the following:
183 SV* get_sv("package::varname", FALSE);
185 This returns NULL if the variable does not exist.
187 If you want to know if this variable (or any other SV) is actually C<defined>,
192 The scalar C<undef> value is stored in an SV instance called C<PL_sv_undef>. Its
193 address can be used whenever an C<SV*> is needed.
195 There are also the two values C<PL_sv_yes> and C<PL_sv_no>, which contain Boolean
196 TRUE and FALSE values, respectively. Like C<PL_sv_undef>, their addresses can
197 be used whenever an C<SV*> is needed.
199 Do not be fooled into thinking that C<(SV *) 0> is the same as C<&PL_sv_undef>.
203 if (I-am-to-return-a-real-value) {
204 sv = sv_2mortal(newSViv(42));
208 This code tries to return a new SV (which contains the value 42) if it should
209 return a real value, or undef otherwise. Instead it has returned a NULL
210 pointer which, somewhere down the line, will cause a segmentation violation,
211 bus error, or just weird results. Change the zero to C<&PL_sv_undef> in the first
212 line and all will be well.
214 To free an SV that you've created, call C<SvREFCNT_dec(SV*)>. Normally this
215 call is not necessary (see L<Reference Counts and Mortality>).
219 Perl provides the function C<sv_chop> to efficiently remove characters
220 from the beginning of a string; you give it an SV and a pointer to
221 somewhere inside the PV, and it discards everything before the
222 pointer. The efficiency comes by means of a little hack: instead of
223 actually removing the characters, C<sv_chop> sets the flag C<OOK>
224 (offset OK) to signal to other functions that the offset hack is in
225 effect, and it puts the number of bytes chopped off into the IV field
226 of the SV. It then moves the PV pointer (called C<SvPVX>) forward that
227 many bytes, and adjusts C<SvCUR> and C<SvLEN>.
229 Hence, at this point, the start of the buffer that we allocated lives
230 at C<SvPVX(sv) - SvIV(sv)> in memory and the PV pointer is pointing
231 into the middle of this allocated storage.
233 This is best demonstrated by example:
235 % ./perl -Ilib -MDevel::Peek -le '$a="12345"; $a=~s/.//; Dump($a)'
236 SV = PVIV(0x8128450) at 0x81340f0
238 FLAGS = (POK,OOK,pPOK)
240 PV = 0x8135781 ( "1" . ) "2345"\0
244 Here the number of bytes chopped off (1) is put into IV, and
245 C<Devel::Peek::Dump> helpfully reminds us that this is an offset. The
246 portion of the string between the "real" and the "fake" beginnings is
247 shown in parentheses, and the values of C<SvCUR> and C<SvLEN> reflect
248 the fake beginning, not the real one.
250 Something similar to the offset hack is performed on AVs to enable
251 efficient shifting and splicing off the beginning of the array; while
252 C<AvARRAY> points to the first element in the array that is visible from
253 Perl, C<AvALLOC> points to the real start of the C array. These are
254 usually the same, but a C<shift> operation can be carried out by
255 increasing C<AvARRAY> by one and decreasing C<AvFILL> and C<AvLEN>.
256 Again, the location of the real start of the C array only comes into
257 play when freeing the array. See C<av_shift> in F<av.c>.
259 =head2 What's Really Stored in an SV?
261 Recall that the usual method of determining the type of scalar you have is
262 to use C<Sv*OK> macros. Because a scalar can be both a number and a string,
263 usually these macros will always return TRUE and calling the C<Sv*V>
264 macros will do the appropriate conversion of string to integer/double or
265 integer/double to string.
267 If you I<really> need to know if you have an integer, double, or string
268 pointer in an SV, you can use the following three macros instead:
274 These will tell you if you truly have an integer, double, or string pointer
275 stored in your SV. The "p" stands for private.
277 The are various ways in which the private and public flags may differ.
278 For example, a tied SV may have a valid underlying value in the IV slot
279 (so SvIOKp is true), but the data should be accessed via the FETCH
280 routine rather than directly, so SvIOK is false. Another is when
281 numeric conversion has occured and precision has been lost: only the
282 private flag is set on 'lossy' values. So when an NV is converted to an
283 IV with loss, SvIOKp, SvNOKp and SvNOK will be set, while SvIOK wont be.
285 In general, though, it's best to use the C<Sv*V> macros.
287 =head2 Working with AVs
289 There are two ways to create and load an AV. The first method creates an
294 The second method both creates the AV and initially populates it with SVs:
296 AV* av_make(I32 num, SV **ptr);
298 The second argument points to an array containing C<num> C<SV*>'s. Once the
299 AV has been created, the SVs can be destroyed, if so desired.
301 Once the AV has been created, the following operations are possible on AVs:
303 void av_push(AV*, SV*);
306 void av_unshift(AV*, I32 num);
308 These should be familiar operations, with the exception of C<av_unshift>.
309 This routine adds C<num> elements at the front of the array with the C<undef>
310 value. You must then use C<av_store> (described below) to assign values
311 to these new elements.
313 Here are some other functions:
316 SV** av_fetch(AV*, I32 key, I32 lval);
317 SV** av_store(AV*, I32 key, SV* val);
319 The C<av_len> function returns the highest index value in array (just
320 like $#array in Perl). If the array is empty, -1 is returned. The
321 C<av_fetch> function returns the value at index C<key>, but if C<lval>
322 is non-zero, then C<av_fetch> will store an undef value at that index.
323 The C<av_store> function stores the value C<val> at index C<key>, and does
324 not increment the reference count of C<val>. Thus the caller is responsible
325 for taking care of that, and if C<av_store> returns NULL, the caller will
326 have to decrement the reference count to avoid a memory leak. Note that
327 C<av_fetch> and C<av_store> both return C<SV**>'s, not C<SV*>'s as their
332 void av_extend(AV*, I32 key);
334 The C<av_clear> function deletes all the elements in the AV* array, but
335 does not actually delete the array itself. The C<av_undef> function will
336 delete all the elements in the array plus the array itself. The
337 C<av_extend> function extends the array so that it contains at least C<key+1>
338 elements. If C<key+1> is less than the currently allocated length of the array,
339 then nothing is done.
341 If you know the name of an array variable, you can get a pointer to its AV
342 by using the following:
344 AV* get_av("package::varname", FALSE);
346 This returns NULL if the variable does not exist.
348 See L<Understanding the Magic of Tied Hashes and Arrays> for more
349 information on how to use the array access functions on tied arrays.
351 =head2 Working with HVs
353 To create an HV, you use the following routine:
357 Once the HV has been created, the following operations are possible on HVs:
359 SV** hv_store(HV*, const char* key, U32 klen, SV* val, U32 hash);
360 SV** hv_fetch(HV*, const char* key, U32 klen, I32 lval);
362 The C<klen> parameter is the length of the key being passed in (Note that
363 you cannot pass 0 in as a value of C<klen> to tell Perl to measure the
364 length of the key). The C<val> argument contains the SV pointer to the
365 scalar being stored, and C<hash> is the precomputed hash value (zero if
366 you want C<hv_store> to calculate it for you). The C<lval> parameter
367 indicates whether this fetch is actually a part of a store operation, in
368 which case a new undefined value will be added to the HV with the supplied
369 key and C<hv_fetch> will return as if the value had already existed.
371 Remember that C<hv_store> and C<hv_fetch> return C<SV**>'s and not just
372 C<SV*>. To access the scalar value, you must first dereference the return
373 value. However, you should check to make sure that the return value is
374 not NULL before dereferencing it.
376 These two functions check if a hash table entry exists, and deletes it.
378 bool hv_exists(HV*, const char* key, U32 klen);
379 SV* hv_delete(HV*, const char* key, U32 klen, I32 flags);
381 If C<flags> does not include the C<G_DISCARD> flag then C<hv_delete> will
382 create and return a mortal copy of the deleted value.
384 And more miscellaneous functions:
389 Like their AV counterparts, C<hv_clear> deletes all the entries in the hash
390 table but does not actually delete the hash table. The C<hv_undef> deletes
391 both the entries and the hash table itself.
393 Perl keeps the actual data in linked list of structures with a typedef of HE.
394 These contain the actual key and value pointers (plus extra administrative
395 overhead). The key is a string pointer; the value is an C<SV*>. However,
396 once you have an C<HE*>, to get the actual key and value, use the routines
399 I32 hv_iterinit(HV*);
400 /* Prepares starting point to traverse hash table */
401 HE* hv_iternext(HV*);
402 /* Get the next entry, and return a pointer to a
403 structure that has both the key and value */
404 char* hv_iterkey(HE* entry, I32* retlen);
405 /* Get the key from an HE structure and also return
406 the length of the key string */
407 SV* hv_iterval(HV*, HE* entry);
408 /* Return an SV pointer to the value of the HE
410 SV* hv_iternextsv(HV*, char** key, I32* retlen);
411 /* This convenience routine combines hv_iternext,
412 hv_iterkey, and hv_iterval. The key and retlen
413 arguments are return values for the key and its
414 length. The value is returned in the SV* argument */
416 If you know the name of a hash variable, you can get a pointer to its HV
417 by using the following:
419 HV* get_hv("package::varname", FALSE);
421 This returns NULL if the variable does not exist.
423 The hash algorithm is defined in the C<PERL_HASH(hash, key, klen)> macro:
427 hash = (hash * 33) + *key++;
428 hash = hash + (hash >> 5); /* after 5.6 */
430 The last step was added in version 5.6 to improve distribution of
431 lower bits in the resulting hash value.
433 See L<Understanding the Magic of Tied Hashes and Arrays> for more
434 information on how to use the hash access functions on tied hashes.
436 =head2 Hash API Extensions
438 Beginning with version 5.004, the following functions are also supported:
440 HE* hv_fetch_ent (HV* tb, SV* key, I32 lval, U32 hash);
441 HE* hv_store_ent (HV* tb, SV* key, SV* val, U32 hash);
443 bool hv_exists_ent (HV* tb, SV* key, U32 hash);
444 SV* hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash);
446 SV* hv_iterkeysv (HE* entry);
448 Note that these functions take C<SV*> keys, which simplifies writing
449 of extension code that deals with hash structures. These functions
450 also allow passing of C<SV*> keys to C<tie> functions without forcing
451 you to stringify the keys (unlike the previous set of functions).
453 They also return and accept whole hash entries (C<HE*>), making their
454 use more efficient (since the hash number for a particular string
455 doesn't have to be recomputed every time). See L<perlapi> for detailed
458 The following macros must always be used to access the contents of hash
459 entries. Note that the arguments to these macros must be simple
460 variables, since they may get evaluated more than once. See
461 L<perlapi> for detailed descriptions of these macros.
463 HePV(HE* he, STRLEN len)
467 HeSVKEY_force(HE* he)
468 HeSVKEY_set(HE* he, SV* sv)
470 These two lower level macros are defined, but must only be used when
471 dealing with keys that are not C<SV*>s:
476 Note that both C<hv_store> and C<hv_store_ent> do not increment the
477 reference count of the stored C<val>, which is the caller's responsibility.
478 If these functions return a NULL value, the caller will usually have to
479 decrement the reference count of C<val> to avoid a memory leak.
483 References are a special type of scalar that point to other data types
484 (including references).
486 To create a reference, use either of the following functions:
488 SV* newRV_inc((SV*) thing);
489 SV* newRV_noinc((SV*) thing);
491 The C<thing> argument can be any of an C<SV*>, C<AV*>, or C<HV*>. The
492 functions are identical except that C<newRV_inc> increments the reference
493 count of the C<thing>, while C<newRV_noinc> does not. For historical
494 reasons, C<newRV> is a synonym for C<newRV_inc>.
496 Once you have a reference, you can use the following macro to dereference
501 then call the appropriate routines, casting the returned C<SV*> to either an
502 C<AV*> or C<HV*>, if required.
504 To determine if an SV is a reference, you can use the following macro:
508 To discover what type of value the reference refers to, use the following
509 macro and then check the return value.
513 The most useful types that will be returned are:
522 SVt_PVGV Glob (possible a file handle)
523 SVt_PVMG Blessed or Magical Scalar
525 See the sv.h header file for more details.
527 =head2 Blessed References and Class Objects
529 References are also used to support object-oriented programming. In the
530 OO lexicon, an object is simply a reference that has been blessed into a
531 package (or class). Once blessed, the programmer may now use the reference
532 to access the various methods in the class.
534 A reference can be blessed into a package with the following function:
536 SV* sv_bless(SV* sv, HV* stash);
538 The C<sv> argument must be a reference. The C<stash> argument specifies
539 which class the reference will belong to. See
540 L<Stashes and Globs> for information on converting class names into stashes.
542 /* Still under construction */
544 Upgrades rv to reference if not already one. Creates new SV for rv to
545 point to. If C<classname> is non-null, the SV is blessed into the specified
546 class. SV is returned.
548 SV* newSVrv(SV* rv, const char* classname);
550 Copies integer, unsigned integer or double into an SV whose reference is C<rv>. SV is blessed
551 if C<classname> is non-null.
553 SV* sv_setref_iv(SV* rv, const char* classname, IV iv);
554 SV* sv_setref_uv(SV* rv, const char* classname, UV uv);
555 SV* sv_setref_nv(SV* rv, const char* classname, NV iv);
557 Copies the pointer value (I<the address, not the string!>) into an SV whose
558 reference is rv. SV is blessed if C<classname> is non-null.
560 SV* sv_setref_pv(SV* rv, const char* classname, PV iv);
562 Copies string into an SV whose reference is C<rv>. Set length to 0 to let
563 Perl calculate the string length. SV is blessed if C<classname> is non-null.
565 SV* sv_setref_pvn(SV* rv, const char* classname, PV iv, STRLEN length);
567 Tests whether the SV is blessed into the specified class. It does not
568 check inheritance relationships.
570 int sv_isa(SV* sv, const char* name);
572 Tests whether the SV is a reference to a blessed object.
574 int sv_isobject(SV* sv);
576 Tests whether the SV is derived from the specified class. SV can be either
577 a reference to a blessed object or a string containing a class name. This
578 is the function implementing the C<UNIVERSAL::isa> functionality.
580 bool sv_derived_from(SV* sv, const char* name);
582 To check if you've got an object derived from a specific class you have
585 if (sv_isobject(sv) && sv_derived_from(sv, class)) { ... }
587 =head2 Creating New Variables
589 To create a new Perl variable with an undef value which can be accessed from
590 your Perl script, use the following routines, depending on the variable type.
592 SV* get_sv("package::varname", TRUE);
593 AV* get_av("package::varname", TRUE);
594 HV* get_hv("package::varname", TRUE);
596 Notice the use of TRUE as the second parameter. The new variable can now
597 be set, using the routines appropriate to the data type.
599 There are additional macros whose values may be bitwise OR'ed with the
600 C<TRUE> argument to enable certain extra features. Those bits are:
606 Marks the variable as multiply defined, thus preventing the:
608 Name <varname> used only once: possible typo
616 Had to create <varname> unexpectedly
618 if the variable did not exist before the function was called.
622 If you do not specify a package name, the variable is created in the current
625 =head2 Reference Counts and Mortality
627 Perl uses a reference count-driven garbage collection mechanism. SVs,
628 AVs, or HVs (xV for short in the following) start their life with a
629 reference count of 1. If the reference count of an xV ever drops to 0,
630 then it will be destroyed and its memory made available for reuse.
632 This normally doesn't happen at the Perl level unless a variable is
633 undef'ed or the last variable holding a reference to it is changed or
634 overwritten. At the internal level, however, reference counts can be
635 manipulated with the following macros:
637 int SvREFCNT(SV* sv);
638 SV* SvREFCNT_inc(SV* sv);
639 void SvREFCNT_dec(SV* sv);
641 However, there is one other function which manipulates the reference
642 count of its argument. The C<newRV_inc> function, you will recall,
643 creates a reference to the specified argument. As a side effect,
644 it increments the argument's reference count. If this is not what
645 you want, use C<newRV_noinc> instead.
647 For example, imagine you want to return a reference from an XSUB function.
648 Inside the XSUB routine, you create an SV which initially has a reference
649 count of one. Then you call C<newRV_inc>, passing it the just-created SV.
650 This returns the reference as a new SV, but the reference count of the
651 SV you passed to C<newRV_inc> has been incremented to two. Now you
652 return the reference from the XSUB routine and forget about the SV.
653 But Perl hasn't! Whenever the returned reference is destroyed, the
654 reference count of the original SV is decreased to one and nothing happens.
655 The SV will hang around without any way to access it until Perl itself
656 terminates. This is a memory leak.
658 The correct procedure, then, is to use C<newRV_noinc> instead of
659 C<newRV_inc>. Then, if and when the last reference is destroyed,
660 the reference count of the SV will go to zero and it will be destroyed,
661 stopping any memory leak.
663 There are some convenience functions available that can help with the
664 destruction of xVs. These functions introduce the concept of "mortality".
665 An xV that is mortal has had its reference count marked to be decremented,
666 but not actually decremented, until "a short time later". Generally the
667 term "short time later" means a single Perl statement, such as a call to
668 an XSUB function. The actual determinant for when mortal xVs have their
669 reference count decremented depends on two macros, SAVETMPS and FREETMPS.
670 See L<perlcall> and L<perlxs> for more details on these macros.
672 "Mortalization" then is at its simplest a deferred C<SvREFCNT_dec>.
673 However, if you mortalize a variable twice, the reference count will
674 later be decremented twice.
676 "Mortal" SVs are mainly used for SVs that are placed on perl's stack.
677 For example an SV which is created just to pass a number to a called sub
678 is made mortal to have it cleaned up automatically when stack is popped.
679 Similarly results returned by XSUBs (which go in the stack) are often
682 To create a mortal variable, use the functions:
686 SV* sv_mortalcopy(SV*)
688 The first call creates a mortal SV (with no value), the second converts an existing
689 SV to a mortal SV (and thus defers a call to C<SvREFCNT_dec>), and the
690 third creates a mortal copy of an existing SV.
691 Because C<sv_newmortal> gives the new SV no value,it must normally be given one
692 via C<sv_setpv>, C<sv_setiv>, etc. :
694 SV *tmp = sv_newmortal();
695 sv_setiv(tmp, an_integer);
697 As that is multiple C statements it is quite common so see this idiom instead:
699 SV *tmp = sv_2mortal(newSViv(an_integer));
702 You should be careful about creating mortal variables. Strange things
703 can happen if you make the same value mortal within multiple contexts,
704 or if you make a variable mortal multiple times. Thinking of "Mortalization"
705 as deferred C<SvREFCNT_dec> should help to minimize such problems.
706 For example if you are passing an SV which you I<know> has high enough REFCNT
707 to survive its use on the stack you need not do any mortalization.
708 If you are not sure then doing an C<SvREFCNT_inc> and C<sv_2mortal>, or
709 making a C<sv_mortalcopy> is safer.
711 The mortal routines are not just for SVs -- AVs and HVs can be
712 made mortal by passing their address (type-casted to C<SV*>) to the
713 C<sv_2mortal> or C<sv_mortalcopy> routines.
715 =head2 Stashes and Globs
717 A "stash" is a hash that contains all of the different objects that
718 are contained within a package. Each key of the stash is a symbol
719 name (shared by all the different types of objects that have the same
720 name), and each value in the hash table is a GV (Glob Value). This GV
721 in turn contains references to the various objects of that name,
722 including (but not limited to) the following:
731 There is a single stash called "PL_defstash" that holds the items that exist
732 in the "main" package. To get at the items in other packages, append the
733 string "::" to the package name. The items in the "Foo" package are in
734 the stash "Foo::" in PL_defstash. The items in the "Bar::Baz" package are
735 in the stash "Baz::" in "Bar::"'s stash.
737 To get the stash pointer for a particular package, use the function:
739 HV* gv_stashpv(const char* name, I32 create)
740 HV* gv_stashsv(SV*, I32 create)
742 The first function takes a literal string, the second uses the string stored
743 in the SV. Remember that a stash is just a hash table, so you get back an
744 C<HV*>. The C<create> flag will create a new package if it is set.
746 The name that C<gv_stash*v> wants is the name of the package whose symbol table
747 you want. The default package is called C<main>. If you have multiply nested
748 packages, pass their names to C<gv_stash*v>, separated by C<::> as in the Perl
751 Alternately, if you have an SV that is a blessed reference, you can find
752 out the stash pointer by using:
754 HV* SvSTASH(SvRV(SV*));
756 then use the following to get the package name itself:
758 char* HvNAME(HV* stash);
760 If you need to bless or re-bless an object you can use the following
763 SV* sv_bless(SV*, HV* stash)
765 where the first argument, an C<SV*>, must be a reference, and the second
766 argument is a stash. The returned C<SV*> can now be used in the same way
769 For more information on references and blessings, consult L<perlref>.
771 =head2 Double-Typed SVs
773 Scalar variables normally contain only one type of value, an integer,
774 double, pointer, or reference. Perl will automatically convert the
775 actual scalar data from the stored type into the requested type.
777 Some scalar variables contain more than one type of scalar data. For
778 example, the variable C<$!> contains either the numeric value of C<errno>
779 or its string equivalent from either C<strerror> or C<sys_errlist[]>.
781 To force multiple data values into an SV, you must do two things: use the
782 C<sv_set*v> routines to add the additional scalar type, then set a flag
783 so that Perl will believe it contains more than one type of data. The
784 four macros to set the flags are:
791 The particular macro you must use depends on which C<sv_set*v> routine
792 you called first. This is because every C<sv_set*v> routine turns on
793 only the bit for the particular type of data being set, and turns off
796 For example, to create a new Perl variable called "dberror" that contains
797 both the numeric and descriptive string error values, you could use the
801 extern char *dberror_list;
803 SV* sv = get_sv("dberror", TRUE);
804 sv_setiv(sv, (IV) dberror);
805 sv_setpv(sv, dberror_list[dberror]);
808 If the order of C<sv_setiv> and C<sv_setpv> had been reversed, then the
809 macro C<SvPOK_on> would need to be called instead of C<SvIOK_on>.
811 =head2 Magic Variables
813 [This section still under construction. Ignore everything here. Post no
814 bills. Everything not permitted is forbidden.]
816 Any SV may be magical, that is, it has special features that a normal
817 SV does not have. These features are stored in the SV structure in a
818 linked list of C<struct magic>'s, typedef'ed to C<MAGIC>.
831 Note this is current as of patchlevel 0, and could change at any time.
833 =head2 Assigning Magic
835 Perl adds magic to an SV using the sv_magic function:
837 void sv_magic(SV* sv, SV* obj, int how, const char* name, I32 namlen);
839 The C<sv> argument is a pointer to the SV that is to acquire a new magical
842 If C<sv> is not already magical, Perl uses the C<SvUPGRADE> macro to
843 convert C<sv> to type C<SVt_PVMG>. Perl then continues by adding new magic
844 to the beginning of the linked list of magical features. Any prior entry
845 of the same type of magic is deleted. Note that this can be overridden,
846 and multiple instances of the same type of magic can be associated with an
849 The C<name> and C<namlen> arguments are used to associate a string with
850 the magic, typically the name of a variable. C<namlen> is stored in the
851 C<mg_len> field and if C<name> is non-null and C<namlen> E<gt>= 0 a malloc'd
852 copy of the name is stored in C<mg_ptr> field.
854 The sv_magic function uses C<how> to determine which, if any, predefined
855 "Magic Virtual Table" should be assigned to the C<mg_virtual> field.
856 See the "Magic Virtual Table" section below. The C<how> argument is also
857 stored in the C<mg_type> field. The value of C<how> should be chosen
858 from the set of macros C<PERL_MAGIC_foo> found perl.h. Note that before
859 these macros were added, Perl internals used to directly use character
860 literals, so you may occasionally come across old code or documentation
861 referring to 'U' magic rather than C<PERL_MAGIC_uvar> for example.
863 The C<obj> argument is stored in the C<mg_obj> field of the C<MAGIC>
864 structure. If it is not the same as the C<sv> argument, the reference
865 count of the C<obj> object is incremented. If it is the same, or if
866 the C<how> argument is C<PERL_MAGIC_arylen>, or if it is a NULL pointer,
867 then C<obj> is merely stored, without the reference count being incremented.
869 There is also a function to add magic to an C<HV>:
871 void hv_magic(HV *hv, GV *gv, int how);
873 This simply calls C<sv_magic> and coerces the C<gv> argument into an C<SV>.
875 To remove the magic from an SV, call the function sv_unmagic:
877 void sv_unmagic(SV *sv, int type);
879 The C<type> argument should be equal to the C<how> value when the C<SV>
880 was initially made magical.
882 =head2 Magic Virtual Tables
884 The C<mg_virtual> field in the C<MAGIC> structure is a pointer to an
885 C<MGVTBL>, which is a structure of function pointers and stands for
886 "Magic Virtual Table" to handle the various operations that might be
887 applied to that variable.
889 The C<MGVTBL> has five pointers to the following routine types:
891 int (*svt_get)(SV* sv, MAGIC* mg);
892 int (*svt_set)(SV* sv, MAGIC* mg);
893 U32 (*svt_len)(SV* sv, MAGIC* mg);
894 int (*svt_clear)(SV* sv, MAGIC* mg);
895 int (*svt_free)(SV* sv, MAGIC* mg);
897 This MGVTBL structure is set at compile-time in C<perl.h> and there are
898 currently 19 types (or 21 with overloading turned on). These different
899 structures contain pointers to various routines that perform additional
900 actions depending on which function is being called.
902 Function pointer Action taken
903 ---------------- ------------
904 svt_get Do something before the value of the SV is retrieved.
905 svt_set Do something after the SV is assigned a value.
906 svt_len Report on the SV's length.
907 svt_clear Clear something the SV represents.
908 svt_free Free any extra storage associated with the SV.
910 For instance, the MGVTBL structure called C<vtbl_sv> (which corresponds
911 to an C<mg_type> of C<PERL_MAGIC_sv>) contains:
913 { magic_get, magic_set, magic_len, 0, 0 }
915 Thus, when an SV is determined to be magical and of type C<PERL_MAGIC_sv>,
916 if a get operation is being performed, the routine C<magic_get> is
917 called. All the various routines for the various magical types begin
918 with C<magic_>. NOTE: the magic routines are not considered part of
919 the Perl API, and may not be exported by the Perl library.
921 The current kinds of Magic Virtual Tables are:
924 (old-style char and macro) MGVTBL Type of magic
925 -------------------------- ------ ----------------------------
926 \0 PERL_MAGIC_sv vtbl_sv Special scalar variable
927 A PERL_MAGIC_overload vtbl_amagic %OVERLOAD hash
928 a PERL_MAGIC_overload_elem vtbl_amagicelem %OVERLOAD hash element
929 c PERL_MAGIC_overload_table (none) Holds overload table (AMT)
931 B PERL_MAGIC_bm vtbl_bm Boyer-Moore (fast string search)
932 D PERL_MAGIC_regdata vtbl_regdata Regex match position data
934 d PERL_MAGIC_regdatum vtbl_regdatum Regex match position data
936 E PERL_MAGIC_env vtbl_env %ENV hash
937 e PERL_MAGIC_envelem vtbl_envelem %ENV hash element
938 f PERL_MAGIC_fm vtbl_fm Formline ('compiled' format)
939 g PERL_MAGIC_regex_global vtbl_mglob m//g target / study()ed string
940 I PERL_MAGIC_isa vtbl_isa @ISA array
941 i PERL_MAGIC_isaelem vtbl_isaelem @ISA array element
942 k PERL_MAGIC_nkeys vtbl_nkeys scalar(keys()) lvalue
943 L PERL_MAGIC_dbfile (none) Debugger %_<filename
944 l PERL_MAGIC_dbline vtbl_dbline Debugger %_<filename element
945 m PERL_MAGIC_mutex vtbl_mutex ???
946 o PERL_MAGIC_collxfrm vtbl_collxfrm Locale collate transformation
947 P PERL_MAGIC_tied vtbl_pack Tied array or hash
948 p PERL_MAGIC_tiedelem vtbl_packelem Tied array or hash element
949 q PERL_MAGIC_tiedscalar vtbl_packelem Tied scalar or handle
950 r PERL_MAGIC_qr vtbl_qr precompiled qr// regex
951 S PERL_MAGIC_sig vtbl_sig %SIG hash
952 s PERL_MAGIC_sigelem vtbl_sigelem %SIG hash element
953 t PERL_MAGIC_taint vtbl_taint Taintedness
954 U PERL_MAGIC_uvar vtbl_uvar Available for use by extensions
955 v PERL_MAGIC_vec vtbl_vec vec() lvalue
956 x PERL_MAGIC_substr vtbl_substr substr() lvalue
957 y PERL_MAGIC_defelem vtbl_defelem Shadow "foreach" iterator
958 variable / smart parameter
960 * PERL_MAGIC_glob vtbl_glob GV (typeglob)
961 # PERL_MAGIC_arylen vtbl_arylen Array length ($#ary)
962 . PERL_MAGIC_pos vtbl_pos pos() lvalue
963 < PERL_MAGIC_backref vtbl_backref ???
964 ~ PERL_MAGIC_ext (none) Available for use by extensions
966 When an uppercase and lowercase letter both exist in the table, then the
967 uppercase letter is used to represent some kind of composite type (a list
968 or a hash), and the lowercase letter is used to represent an element of
969 that composite type. Some internals code makes use of this case
972 The C<PERL_MAGIC_ext> and C<PERL_MAGIC_uvar> magic types are defined
973 specifically for use by extensions and will not be used by perl itself.
974 Extensions can use C<PERL_MAGIC_ext> magic to 'attach' private information
975 to variables (typically objects). This is especially useful because
976 there is no way for normal perl code to corrupt this private information
977 (unlike using extra elements of a hash object).
979 Similarly, C<PERL_MAGIC_uvar> magic can be used much like tie() to call a
980 C function any time a scalar's value is used or changed. The C<MAGIC>'s
981 C<mg_ptr> field points to a C<ufuncs> structure:
984 I32 (*uf_val)(pTHX_ IV, SV*);
985 I32 (*uf_set)(pTHX_ IV, SV*);
989 When the SV is read from or written to, the C<uf_val> or C<uf_set>
990 function will be called with C<uf_index> as the first arg and a pointer to
991 the SV as the second. A simple example of how to add C<PERL_MAGIC_uvar>
992 magic is shown below. Note that the ufuncs structure is copied by
993 sv_magic, so you can safely allocate it on the stack.
1001 uf.uf_val = &my_get_fn;
1002 uf.uf_set = &my_set_fn;
1004 sv_magic(sv, 0, PERL_MAGIC_uvar, (char*)&uf, sizeof(uf));
1006 Note that because multiple extensions may be using C<PERL_MAGIC_ext>
1007 or C<PERL_MAGIC_uvar> magic, it is important for extensions to take
1008 extra care to avoid conflict. Typically only using the magic on
1009 objects blessed into the same class as the extension is sufficient.
1010 For C<PERL_MAGIC_ext> magic, it may also be appropriate to add an I32
1011 'signature' at the top of the private data area and check that.
1013 Also note that the C<sv_set*()> and C<sv_cat*()> functions described
1014 earlier do B<not> invoke 'set' magic on their targets. This must
1015 be done by the user either by calling the C<SvSETMAGIC()> macro after
1016 calling these functions, or by using one of the C<sv_set*_mg()> or
1017 C<sv_cat*_mg()> functions. Similarly, generic C code must call the
1018 C<SvGETMAGIC()> macro to invoke any 'get' magic if they use an SV
1019 obtained from external sources in functions that don't handle magic.
1020 See L<perlapi> for a description of these functions.
1021 For example, calls to the C<sv_cat*()> functions typically need to be
1022 followed by C<SvSETMAGIC()>, but they don't need a prior C<SvGETMAGIC()>
1023 since their implementation handles 'get' magic.
1025 =head2 Finding Magic
1027 MAGIC* mg_find(SV*, int type); /* Finds the magic pointer of that type */
1029 This routine returns a pointer to the C<MAGIC> structure stored in the SV.
1030 If the SV does not have that magical feature, C<NULL> is returned. Also,
1031 if the SV is not of type SVt_PVMG, Perl may core dump.
1033 int mg_copy(SV* sv, SV* nsv, const char* key, STRLEN klen);
1035 This routine checks to see what types of magic C<sv> has. If the mg_type
1036 field is an uppercase letter, then the mg_obj is copied to C<nsv>, but
1037 the mg_type field is changed to be the lowercase letter.
1039 =head2 Understanding the Magic of Tied Hashes and Arrays
1041 Tied hashes and arrays are magical beasts of the C<PERL_MAGIC_tied>
1044 WARNING: As of the 5.004 release, proper usage of the array and hash
1045 access functions requires understanding a few caveats. Some
1046 of these caveats are actually considered bugs in the API, to be fixed
1047 in later releases, and are bracketed with [MAYCHANGE] below. If
1048 you find yourself actually applying such information in this section, be
1049 aware that the behavior may change in the future, umm, without warning.
1051 The perl tie function associates a variable with an object that implements
1052 the various GET, SET, etc methods. To perform the equivalent of the perl
1053 tie function from an XSUB, you must mimic this behaviour. The code below
1054 carries out the necessary steps - firstly it creates a new hash, and then
1055 creates a second hash which it blesses into the class which will implement
1056 the tie methods. Lastly it ties the two hashes together, and returns a
1057 reference to the new tied hash. Note that the code below does NOT call the
1058 TIEHASH method in the MyTie class -
1059 see L<Calling Perl Routines from within C Programs> for details on how
1070 tie = newRV_noinc((SV*)newHV());
1071 stash = gv_stashpv("MyTie", TRUE);
1072 sv_bless(tie, stash);
1073 hv_magic(hash, (GV*)tie, PERL_MAGIC_tied);
1074 RETVAL = newRV_noinc(hash);
1078 The C<av_store> function, when given a tied array argument, merely
1079 copies the magic of the array onto the value to be "stored", using
1080 C<mg_copy>. It may also return NULL, indicating that the value did not
1081 actually need to be stored in the array. [MAYCHANGE] After a call to
1082 C<av_store> on a tied array, the caller will usually need to call
1083 C<mg_set(val)> to actually invoke the perl level "STORE" method on the
1084 TIEARRAY object. If C<av_store> did return NULL, a call to
1085 C<SvREFCNT_dec(val)> will also be usually necessary to avoid a memory
1088 The previous paragraph is applicable verbatim to tied hash access using the
1089 C<hv_store> and C<hv_store_ent> functions as well.
1091 C<av_fetch> and the corresponding hash functions C<hv_fetch> and
1092 C<hv_fetch_ent> actually return an undefined mortal value whose magic
1093 has been initialized using C<mg_copy>. Note the value so returned does not
1094 need to be deallocated, as it is already mortal. [MAYCHANGE] But you will
1095 need to call C<mg_get()> on the returned value in order to actually invoke
1096 the perl level "FETCH" method on the underlying TIE object. Similarly,
1097 you may also call C<mg_set()> on the return value after possibly assigning
1098 a suitable value to it using C<sv_setsv>, which will invoke the "STORE"
1099 method on the TIE object. [/MAYCHANGE]
1102 In other words, the array or hash fetch/store functions don't really
1103 fetch and store actual values in the case of tied arrays and hashes. They
1104 merely call C<mg_copy> to attach magic to the values that were meant to be
1105 "stored" or "fetched". Later calls to C<mg_get> and C<mg_set> actually
1106 do the job of invoking the TIE methods on the underlying objects. Thus
1107 the magic mechanism currently implements a kind of lazy access to arrays
1110 Currently (as of perl version 5.004), use of the hash and array access
1111 functions requires the user to be aware of whether they are operating on
1112 "normal" hashes and arrays, or on their tied variants. The API may be
1113 changed to provide more transparent access to both tied and normal data
1114 types in future versions.
1117 You would do well to understand that the TIEARRAY and TIEHASH interfaces
1118 are mere sugar to invoke some perl method calls while using the uniform hash
1119 and array syntax. The use of this sugar imposes some overhead (typically
1120 about two to four extra opcodes per FETCH/STORE operation, in addition to
1121 the creation of all the mortal variables required to invoke the methods).
1122 This overhead will be comparatively small if the TIE methods are themselves
1123 substantial, but if they are only a few statements long, the overhead
1124 will not be insignificant.
1126 =head2 Localizing changes
1128 Perl has a very handy construction
1135 This construction is I<approximately> equivalent to
1144 The biggest difference is that the first construction would
1145 reinstate the initial value of $var, irrespective of how control exits
1146 the block: C<goto>, C<return>, C<die>/C<eval>, etc. It is a little bit
1147 more efficient as well.
1149 There is a way to achieve a similar task from C via Perl API: create a
1150 I<pseudo-block>, and arrange for some changes to be automatically
1151 undone at the end of it, either explicit, or via a non-local exit (via
1152 die()). A I<block>-like construct is created by a pair of
1153 C<ENTER>/C<LEAVE> macros (see L<perlcall/"Returning a Scalar">).
1154 Such a construct may be created specially for some important localized
1155 task, or an existing one (like boundaries of enclosing Perl
1156 subroutine/block, or an existing pair for freeing TMPs) may be
1157 used. (In the second case the overhead of additional localization must
1158 be almost negligible.) Note that any XSUB is automatically enclosed in
1159 an C<ENTER>/C<LEAVE> pair.
1161 Inside such a I<pseudo-block> the following service is available:
1165 =item C<SAVEINT(int i)>
1167 =item C<SAVEIV(IV i)>
1169 =item C<SAVEI32(I32 i)>
1171 =item C<SAVELONG(long i)>
1173 These macros arrange things to restore the value of integer variable
1174 C<i> at the end of enclosing I<pseudo-block>.
1176 =item C<SAVESPTR(s)>
1178 =item C<SAVEPPTR(p)>
1180 These macros arrange things to restore the value of pointers C<s> and
1181 C<p>. C<s> must be a pointer of a type which survives conversion to
1182 C<SV*> and back, C<p> should be able to survive conversion to C<char*>
1185 =item C<SAVEFREESV(SV *sv)>
1187 The refcount of C<sv> would be decremented at the end of
1188 I<pseudo-block>. This is similar to C<sv_2mortal> in that it is also a
1189 mechanism for doing a delayed C<SvREFCNT_dec>. However, while C<sv_2mortal>
1190 extends the lifetime of C<sv> until the beginning of the next statement,
1191 C<SAVEFREESV> extends it until the end of the enclosing scope. These
1192 lifetimes can be wildly different.
1194 Also compare C<SAVEMORTALIZESV>.
1196 =item C<SAVEMORTALIZESV(SV *sv)>
1198 Just like C<SAVEFREESV>, but mortalizes C<sv> at the end of the current
1199 scope instead of decrementing its reference count. This usually has the
1200 effect of keeping C<sv> alive until the statement that called the currently
1201 live scope has finished executing.
1203 =item C<SAVEFREEOP(OP *op)>
1205 The C<OP *> is op_free()ed at the end of I<pseudo-block>.
1207 =item C<SAVEFREEPV(p)>
1209 The chunk of memory which is pointed to by C<p> is Safefree()ed at the
1210 end of I<pseudo-block>.
1212 =item C<SAVECLEARSV(SV *sv)>
1214 Clears a slot in the current scratchpad which corresponds to C<sv> at
1215 the end of I<pseudo-block>.
1217 =item C<SAVEDELETE(HV *hv, char *key, I32 length)>
1219 The key C<key> of C<hv> is deleted at the end of I<pseudo-block>. The
1220 string pointed to by C<key> is Safefree()ed. If one has a I<key> in
1221 short-lived storage, the corresponding string may be reallocated like
1224 SAVEDELETE(PL_defstash, savepv(tmpbuf), strlen(tmpbuf));
1226 =item C<SAVEDESTRUCTOR(DESTRUCTORFUNC_NOCONTEXT_t f, void *p)>
1228 At the end of I<pseudo-block> the function C<f> is called with the
1231 =item C<SAVEDESTRUCTOR_X(DESTRUCTORFUNC_t f, void *p)>
1233 At the end of I<pseudo-block> the function C<f> is called with the
1234 implicit context argument (if any), and C<p>.
1236 =item C<SAVESTACK_POS()>
1238 The current offset on the Perl internal stack (cf. C<SP>) is restored
1239 at the end of I<pseudo-block>.
1243 The following API list contains functions, thus one needs to
1244 provide pointers to the modifiable data explicitly (either C pointers,
1245 or Perlish C<GV *>s). Where the above macros take C<int>, a similar
1246 function takes C<int *>.
1250 =item C<SV* save_scalar(GV *gv)>
1252 Equivalent to Perl code C<local $gv>.
1254 =item C<AV* save_ary(GV *gv)>
1256 =item C<HV* save_hash(GV *gv)>
1258 Similar to C<save_scalar>, but localize C<@gv> and C<%gv>.
1260 =item C<void save_item(SV *item)>
1262 Duplicates the current value of C<SV>, on the exit from the current
1263 C<ENTER>/C<LEAVE> I<pseudo-block> will restore the value of C<SV>
1264 using the stored value.
1266 =item C<void save_list(SV **sarg, I32 maxsarg)>
1268 A variant of C<save_item> which takes multiple arguments via an array
1269 C<sarg> of C<SV*> of length C<maxsarg>.
1271 =item C<SV* save_svref(SV **sptr)>
1273 Similar to C<save_scalar>, but will reinstate an C<SV *>.
1275 =item C<void save_aptr(AV **aptr)>
1277 =item C<void save_hptr(HV **hptr)>
1279 Similar to C<save_svref>, but localize C<AV *> and C<HV *>.
1283 The C<Alias> module implements localization of the basic types within the
1284 I<caller's scope>. People who are interested in how to localize things in
1285 the containing scope should take a look there too.
1289 =head2 XSUBs and the Argument Stack
1291 The XSUB mechanism is a simple way for Perl programs to access C subroutines.
1292 An XSUB routine will have a stack that contains the arguments from the Perl
1293 program, and a way to map from the Perl data structures to a C equivalent.
1295 The stack arguments are accessible through the C<ST(n)> macro, which returns
1296 the C<n>'th stack argument. Argument 0 is the first argument passed in the
1297 Perl subroutine call. These arguments are C<SV*>, and can be used anywhere
1300 Most of the time, output from the C routine can be handled through use of
1301 the RETVAL and OUTPUT directives. However, there are some cases where the
1302 argument stack is not already long enough to handle all the return values.
1303 An example is the POSIX tzname() call, which takes no arguments, but returns
1304 two, the local time zone's standard and summer time abbreviations.
1306 To handle this situation, the PPCODE directive is used and the stack is
1307 extended using the macro:
1311 where C<SP> is the macro that represents the local copy of the stack pointer,
1312 and C<num> is the number of elements the stack should be extended by.
1314 Now that there is room on the stack, values can be pushed on it using C<PUSHs>
1315 macro. The values pushed will often need to be "mortal" (See L</Reference Counts and Mortality>).
1317 PUSHs(sv_2mortal(newSViv(an_integer)))
1318 PUSHs(sv_2mortal(newSVpv("Some String",0)))
1319 PUSHs(sv_2mortal(newSVnv(3.141592)))
1321 And now the Perl program calling C<tzname>, the two values will be assigned
1324 ($standard_abbrev, $summer_abbrev) = POSIX::tzname;
1326 An alternate (and possibly simpler) method to pushing values on the stack is
1331 This macro automatically adjust the stack for you, if needed. Thus, you
1332 do not need to call C<EXTEND> to extend the stack.
1334 Despite their suggestions in earlier versions of this document the macros
1335 C<PUSHi>, C<PUSHn> and C<PUSHp> are I<not> suited to XSUBs which return
1336 multiple results, see L</Putting a C value on Perl stack>.
1338 For more information, consult L<perlxs> and L<perlxstut>.
1340 =head2 Calling Perl Routines from within C Programs
1342 There are four routines that can be used to call a Perl subroutine from
1343 within a C program. These four are:
1345 I32 call_sv(SV*, I32);
1346 I32 call_pv(const char*, I32);
1347 I32 call_method(const char*, I32);
1348 I32 call_argv(const char*, I32, register char**);
1350 The routine most often used is C<call_sv>. The C<SV*> argument
1351 contains either the name of the Perl subroutine to be called, or a
1352 reference to the subroutine. The second argument consists of flags
1353 that control the context in which the subroutine is called, whether
1354 or not the subroutine is being passed arguments, how errors should be
1355 trapped, and how to treat return values.
1357 All four routines return the number of arguments that the subroutine returned
1360 These routines used to be called C<perl_call_sv>, etc., before Perl v5.6.0,
1361 but those names are now deprecated; macros of the same name are provided for
1364 When using any of these routines (except C<call_argv>), the programmer
1365 must manipulate the Perl stack. These include the following macros and
1380 For a detailed description of calling conventions from C to Perl,
1381 consult L<perlcall>.
1383 =head2 Memory Allocation
1385 All memory meant to be used with the Perl API functions should be manipulated
1386 using the macros described in this section. The macros provide the necessary
1387 transparency between differences in the actual malloc implementation that is
1390 It is suggested that you enable the version of malloc that is distributed
1391 with Perl. It keeps pools of various sizes of unallocated memory in
1392 order to satisfy allocation requests more quickly. However, on some
1393 platforms, it may cause spurious malloc or free errors.
1395 New(x, pointer, number, type);
1396 Newc(x, pointer, number, type, cast);
1397 Newz(x, pointer, number, type);
1399 These three macros are used to initially allocate memory.
1401 The first argument C<x> was a "magic cookie" that was used to keep track
1402 of who called the macro, to help when debugging memory problems. However,
1403 the current code makes no use of this feature (most Perl developers now
1404 use run-time memory checkers), so this argument can be any number.
1406 The second argument C<pointer> should be the name of a variable that will
1407 point to the newly allocated memory.
1409 The third and fourth arguments C<number> and C<type> specify how many of
1410 the specified type of data structure should be allocated. The argument
1411 C<type> is passed to C<sizeof>. The final argument to C<Newc>, C<cast>,
1412 should be used if the C<pointer> argument is different from the C<type>
1415 Unlike the C<New> and C<Newc> macros, the C<Newz> macro calls C<memzero>
1416 to zero out all the newly allocated memory.
1418 Renew(pointer, number, type);
1419 Renewc(pointer, number, type, cast);
1422 These three macros are used to change a memory buffer size or to free a
1423 piece of memory no longer needed. The arguments to C<Renew> and C<Renewc>
1424 match those of C<New> and C<Newc> with the exception of not needing the
1425 "magic cookie" argument.
1427 Move(source, dest, number, type);
1428 Copy(source, dest, number, type);
1429 Zero(dest, number, type);
1431 These three macros are used to move, copy, or zero out previously allocated
1432 memory. The C<source> and C<dest> arguments point to the source and
1433 destination starting points. Perl will move, copy, or zero out C<number>
1434 instances of the size of the C<type> data structure (using the C<sizeof>
1439 The most recent development releases of Perl has been experimenting with
1440 removing Perl's dependency on the "normal" standard I/O suite and allowing
1441 other stdio implementations to be used. This involves creating a new
1442 abstraction layer that then calls whichever implementation of stdio Perl
1443 was compiled with. All XSUBs should now use the functions in the PerlIO
1444 abstraction layer and not make any assumptions about what kind of stdio
1447 For a complete description of the PerlIO abstraction, consult L<perlapio>.
1449 =head2 Putting a C value on Perl stack
1451 A lot of opcodes (this is an elementary operation in the internal perl
1452 stack machine) put an SV* on the stack. However, as an optimization
1453 the corresponding SV is (usually) not recreated each time. The opcodes
1454 reuse specially assigned SVs (I<target>s) which are (as a corollary)
1455 not constantly freed/created.
1457 Each of the targets is created only once (but see
1458 L<Scratchpads and recursion> below), and when an opcode needs to put
1459 an integer, a double, or a string on stack, it just sets the
1460 corresponding parts of its I<target> and puts the I<target> on stack.
1462 The macro to put this target on stack is C<PUSHTARG>, and it is
1463 directly used in some opcodes, as well as indirectly in zillions of
1464 others, which use it via C<(X)PUSH[pni]>.
1466 Because the target is reused, you must be careful when pushing multiple
1467 values on the stack. The following code will not do what you think:
1472 This translates as "set C<TARG> to 10, push a pointer to C<TARG> onto
1473 the stack; set C<TARG> to 20, push a pointer to C<TARG> onto the stack".
1474 At the end of the operation, the stack does not contain the values 10
1475 and 20, but actually contains two pointers to C<TARG>, which we have set
1476 to 20. If you need to push multiple different values, use C<XPUSHs>,
1477 which bypasses C<TARG>.
1479 On a related note, if you do use C<(X)PUSH[npi]>, then you're going to
1480 need a C<dTARG> in your variable declarations so that the C<*PUSH*>
1481 macros can make use of the local variable C<TARG>.
1485 The question remains on when the SVs which are I<target>s for opcodes
1486 are created. The answer is that they are created when the current unit --
1487 a subroutine or a file (for opcodes for statements outside of
1488 subroutines) -- is compiled. During this time a special anonymous Perl
1489 array is created, which is called a scratchpad for the current
1492 A scratchpad keeps SVs which are lexicals for the current unit and are
1493 targets for opcodes. One can deduce that an SV lives on a scratchpad
1494 by looking on its flags: lexicals have C<SVs_PADMY> set, and
1495 I<target>s have C<SVs_PADTMP> set.
1497 The correspondence between OPs and I<target>s is not 1-to-1. Different
1498 OPs in the compile tree of the unit can use the same target, if this
1499 would not conflict with the expected life of the temporary.
1501 =head2 Scratchpads and recursion
1503 In fact it is not 100% true that a compiled unit contains a pointer to
1504 the scratchpad AV. In fact it contains a pointer to an AV of
1505 (initially) one element, and this element is the scratchpad AV. Why do
1506 we need an extra level of indirection?
1508 The answer is B<recursion>, and maybe B<threads>. Both
1509 these can create several execution pointers going into the same
1510 subroutine. For the subroutine-child not write over the temporaries
1511 for the subroutine-parent (lifespan of which covers the call to the
1512 child), the parent and the child should have different
1513 scratchpads. (I<And> the lexicals should be separate anyway!)
1515 So each subroutine is born with an array of scratchpads (of length 1).
1516 On each entry to the subroutine it is checked that the current
1517 depth of the recursion is not more than the length of this array, and
1518 if it is, new scratchpad is created and pushed into the array.
1520 The I<target>s on this scratchpad are C<undef>s, but they are already
1521 marked with correct flags.
1523 =head1 Compiled code
1527 Here we describe the internal form your code is converted to by
1528 Perl. Start with a simple example:
1532 This is converted to a tree similar to this one:
1540 (but slightly more complicated). This tree reflects the way Perl
1541 parsed your code, but has nothing to do with the execution order.
1542 There is an additional "thread" going through the nodes of the tree
1543 which shows the order of execution of the nodes. In our simplified
1544 example above it looks like:
1546 $b ---> $c ---> + ---> $a ---> assign-to
1548 But with the actual compile tree for C<$a = $b + $c> it is different:
1549 some nodes I<optimized away>. As a corollary, though the actual tree
1550 contains more nodes than our simplified example, the execution order
1551 is the same as in our example.
1553 =head2 Examining the tree
1555 If you have your perl compiled for debugging (usually done with C<-D
1556 optimize=-g> on C<Configure> command line), you may examine the
1557 compiled tree by specifying C<-Dx> on the Perl command line. The
1558 output takes several lines per node, and for C<$b+$c> it looks like
1563 FLAGS = (SCALAR,KIDS)
1565 TYPE = null ===> (4)
1567 FLAGS = (SCALAR,KIDS)
1569 3 TYPE = gvsv ===> 4
1575 TYPE = null ===> (5)
1577 FLAGS = (SCALAR,KIDS)
1579 4 TYPE = gvsv ===> 5
1585 This tree has 5 nodes (one per C<TYPE> specifier), only 3 of them are
1586 not optimized away (one per number in the left column). The immediate
1587 children of the given node correspond to C<{}> pairs on the same level
1588 of indentation, thus this listing corresponds to the tree:
1596 The execution order is indicated by C<===E<gt>> marks, thus it is C<3
1597 4 5 6> (node C<6> is not included into above listing), i.e.,
1598 C<gvsv gvsv add whatever>.
1600 Each of these nodes represents an op, a fundamental operation inside the
1601 Perl core. The code which implements each operation can be found in the
1602 F<pp*.c> files; the function which implements the op with type C<gvsv>
1603 is C<pp_gvsv>, and so on. As the tree above shows, different ops have
1604 different numbers of children: C<add> is a binary operator, as one would
1605 expect, and so has two children. To accommodate the various different
1606 numbers of children, there are various types of op data structure, and
1607 they link together in different ways.
1609 The simplest type of op structure is C<OP>: this has no children. Unary
1610 operators, C<UNOP>s, have one child, and this is pointed to by the
1611 C<op_first> field. Binary operators (C<BINOP>s) have not only an
1612 C<op_first> field but also an C<op_last> field. The most complex type of
1613 op is a C<LISTOP>, which has any number of children. In this case, the
1614 first child is pointed to by C<op_first> and the last child by
1615 C<op_last>. The children in between can be found by iteratively
1616 following the C<op_sibling> pointer from the first child to the last.
1618 There are also two other op types: a C<PMOP> holds a regular expression,
1619 and has no children, and a C<LOOP> may or may not have children. If the
1620 C<op_children> field is non-zero, it behaves like a C<LISTOP>. To
1621 complicate matters, if a C<UNOP> is actually a C<null> op after
1622 optimization (see L</Compile pass 2: context propagation>) it will still
1623 have children in accordance with its former type.
1625 =head2 Compile pass 1: check routines
1627 The tree is created by the compiler while I<yacc> code feeds it
1628 the constructions it recognizes. Since I<yacc> works bottom-up, so does
1629 the first pass of perl compilation.
1631 What makes this pass interesting for perl developers is that some
1632 optimization may be performed on this pass. This is optimization by
1633 so-called "check routines". The correspondence between node names
1634 and corresponding check routines is described in F<opcode.pl> (do not
1635 forget to run C<make regen_headers> if you modify this file).
1637 A check routine is called when the node is fully constructed except
1638 for the execution-order thread. Since at this time there are no
1639 back-links to the currently constructed node, one can do most any
1640 operation to the top-level node, including freeing it and/or creating
1641 new nodes above/below it.
1643 The check routine returns the node which should be inserted into the
1644 tree (if the top-level node was not modified, check routine returns
1647 By convention, check routines have names C<ck_*>. They are usually
1648 called from C<new*OP> subroutines (or C<convert>) (which in turn are
1649 called from F<perly.y>).
1651 =head2 Compile pass 1a: constant folding
1653 Immediately after the check routine is called the returned node is
1654 checked for being compile-time executable. If it is (the value is
1655 judged to be constant) it is immediately executed, and a I<constant>
1656 node with the "return value" of the corresponding subtree is
1657 substituted instead. The subtree is deleted.
1659 If constant folding was not performed, the execution-order thread is
1662 =head2 Compile pass 2: context propagation
1664 When a context for a part of compile tree is known, it is propagated
1665 down through the tree. At this time the context can have 5 values
1666 (instead of 2 for runtime context): void, boolean, scalar, list, and
1667 lvalue. In contrast with the pass 1 this pass is processed from top
1668 to bottom: a node's context determines the context for its children.
1670 Additional context-dependent optimizations are performed at this time.
1671 Since at this moment the compile tree contains back-references (via
1672 "thread" pointers), nodes cannot be free()d now. To allow
1673 optimized-away nodes at this stage, such nodes are null()ified instead
1674 of free()ing (i.e. their type is changed to OP_NULL).
1676 =head2 Compile pass 3: peephole optimization
1678 After the compile tree for a subroutine (or for an C<eval> or a file)
1679 is created, an additional pass over the code is performed. This pass
1680 is neither top-down or bottom-up, but in the execution order (with
1681 additional complications for conditionals). These optimizations are
1682 done in the subroutine peep(). Optimizations performed at this stage
1683 are subject to the same restrictions as in the pass 2.
1685 =head2 Pluggable runops
1687 The compile tree is executed in a runops function. There are two runops
1688 functions in F<run.c>. C<Perl_runops_debug> is used with DEBUGGING and
1689 C<Perl_runops_standard> is used otherwise. For fine control over the
1690 execution of the compile tree it is possible to provide your own runops
1693 It's probably best to copy one of the existing runops functions and
1694 change it to suit your needs. Then, in the BOOT section of your XS
1697 PL_runops = my_runops;
1699 This function should be as efficient as possible to keep your programs
1700 running as fast as possible.
1702 =head1 Examining internal data structures with the C<dump> functions
1704 To aid debugging, the source file F<dump.c> contains a number of
1705 functions which produce formatted output of internal data structures.
1707 The most commonly used of these functions is C<Perl_sv_dump>; it's used
1708 for dumping SVs, AVs, HVs, and CVs. The C<Devel::Peek> module calls
1709 C<sv_dump> to produce debugging output from Perl-space, so users of that
1710 module should already be familiar with its format.
1712 C<Perl_op_dump> can be used to dump an C<OP> structure or any of its
1713 derivatives, and produces output similar to C<perl -Dx>; in fact,
1714 C<Perl_dump_eval> will dump the main root of the code being evaluated,
1715 exactly like C<-Dx>.
1717 Other useful functions are C<Perl_dump_sub>, which turns a C<GV> into an
1718 op tree, C<Perl_dump_packsubs> which calls C<Perl_dump_sub> on all the
1719 subroutines in a package like so: (Thankfully, these are all xsubs, so
1720 there is no op tree)
1722 (gdb) print Perl_dump_packsubs(PL_defstash)
1724 SUB attributes::bootstrap = (xsub 0x811fedc 0)
1726 SUB UNIVERSAL::can = (xsub 0x811f50c 0)
1728 SUB UNIVERSAL::isa = (xsub 0x811f304 0)
1730 SUB UNIVERSAL::VERSION = (xsub 0x811f7ac 0)
1732 SUB DynaLoader::boot_DynaLoader = (xsub 0x805b188 0)
1734 and C<Perl_dump_all>, which dumps all the subroutines in the stash and
1735 the op tree of the main root.
1737 =head1 How multiple interpreters and concurrency are supported
1739 =head2 Background and PERL_IMPLICIT_CONTEXT
1741 The Perl interpreter can be regarded as a closed box: it has an API
1742 for feeding it code or otherwise making it do things, but it also has
1743 functions for its own use. This smells a lot like an object, and
1744 there are ways for you to build Perl so that you can have multiple
1745 interpreters, with one interpreter represented either as a C structure,
1746 or inside a thread-specific structure. These structures contain all
1747 the context, the state of that interpreter.
1749 Two macros control the major Perl build flavors: MULTIPLICITY and
1750 USE_5005THREADS. The MULTIPLICITY build has a C structure
1751 that packages all the interpreter state, and there is a similar thread-specific
1752 data structure under USE_5005THREADS. In both cases,
1753 PERL_IMPLICIT_CONTEXT is also normally defined, and enables the
1754 support for passing in a "hidden" first argument that represents all three
1757 All this obviously requires a way for the Perl internal functions to be
1758 either subroutines taking some kind of structure as the first
1759 argument, or subroutines taking nothing as the first argument. To
1760 enable these two very different ways of building the interpreter,
1761 the Perl source (as it does in so many other situations) makes heavy
1762 use of macros and subroutine naming conventions.
1764 First problem: deciding which functions will be public API functions and
1765 which will be private. All functions whose names begin C<S_> are private
1766 (think "S" for "secret" or "static"). All other functions begin with
1767 "Perl_", but just because a function begins with "Perl_" does not mean it is
1768 part of the API. (See L</Internal Functions>.) The easiest way to be B<sure> a
1769 function is part of the API is to find its entry in L<perlapi>.
1770 If it exists in L<perlapi>, it's part of the API. If it doesn't, and you
1771 think it should be (i.e., you need it for your extension), send mail via
1772 L<perlbug> explaining why you think it should be.
1774 Second problem: there must be a syntax so that the same subroutine
1775 declarations and calls can pass a structure as their first argument,
1776 or pass nothing. To solve this, the subroutines are named and
1777 declared in a particular way. Here's a typical start of a static
1778 function used within the Perl guts:
1781 S_incline(pTHX_ char *s)
1783 STATIC becomes "static" in C, and may be #define'd to nothing in some
1784 configurations in future.
1786 A public function (i.e. part of the internal API, but not necessarily
1787 sanctioned for use in extensions) begins like this:
1790 Perl_sv_setsv(pTHX_ SV* dsv, SV* ssv)
1792 C<pTHX_> is one of a number of macros (in perl.h) that hide the
1793 details of the interpreter's context. THX stands for "thread", "this",
1794 or "thingy", as the case may be. (And no, George Lucas is not involved. :-)
1795 The first character could be 'p' for a B<p>rototype, 'a' for B<a>rgument,
1796 or 'd' for B<d>eclaration, so we have C<pTHX>, C<aTHX> and C<dTHX>, and
1799 When Perl is built without options that set PERL_IMPLICIT_CONTEXT, there is no
1800 first argument containing the interpreter's context. The trailing underscore
1801 in the pTHX_ macro indicates that the macro expansion needs a comma
1802 after the context argument because other arguments follow it. If
1803 PERL_IMPLICIT_CONTEXT is not defined, pTHX_ will be ignored, and the
1804 subroutine is not prototyped to take the extra argument. The form of the
1805 macro without the trailing underscore is used when there are no additional
1808 When a core function calls another, it must pass the context. This
1809 is normally hidden via macros. Consider C<sv_setsv>. It expands into
1810 something like this:
1812 ifdef PERL_IMPLICIT_CONTEXT
1813 define sv_setsv(a,b) Perl_sv_setsv(aTHX_ a, b)
1814 /* can't do this for vararg functions, see below */
1816 define sv_setsv Perl_sv_setsv
1819 This works well, and means that XS authors can gleefully write:
1823 and still have it work under all the modes Perl could have been
1826 This doesn't work so cleanly for varargs functions, though, as macros
1827 imply that the number of arguments is known in advance. Instead we
1828 either need to spell them out fully, passing C<aTHX_> as the first
1829 argument (the Perl core tends to do this with functions like
1830 Perl_warner), or use a context-free version.
1832 The context-free version of Perl_warner is called
1833 Perl_warner_nocontext, and does not take the extra argument. Instead
1834 it does dTHX; to get the context from thread-local storage. We
1835 C<#define warner Perl_warner_nocontext> so that extensions get source
1836 compatibility at the expense of performance. (Passing an arg is
1837 cheaper than grabbing it from thread-local storage.)
1839 You can ignore [pad]THXx when browsing the Perl headers/sources.
1840 Those are strictly for use within the core. Extensions and embedders
1841 need only be aware of [pad]THX.
1843 =head2 So what happened to dTHR?
1845 C<dTHR> was introduced in perl 5.005 to support the older thread model.
1846 The older thread model now uses the C<THX> mechanism to pass context
1847 pointers around, so C<dTHR> is not useful any more. Perl 5.6.0 and
1848 later still have it for backward source compatibility, but it is defined
1851 =head2 How do I use all this in extensions?
1853 When Perl is built with PERL_IMPLICIT_CONTEXT, extensions that call
1854 any functions in the Perl API will need to pass the initial context
1855 argument somehow. The kicker is that you will need to write it in
1856 such a way that the extension still compiles when Perl hasn't been
1857 built with PERL_IMPLICIT_CONTEXT enabled.
1859 There are three ways to do this. First, the easy but inefficient way,
1860 which is also the default, in order to maintain source compatibility
1861 with extensions: whenever XSUB.h is #included, it redefines the aTHX
1862 and aTHX_ macros to call a function that will return the context.
1863 Thus, something like:
1867 in your extension will translate to this when PERL_IMPLICIT_CONTEXT is
1870 Perl_sv_setsv(Perl_get_context(), asv, bsv);
1872 or to this otherwise:
1874 Perl_sv_setsv(asv, bsv);
1876 You have to do nothing new in your extension to get this; since
1877 the Perl library provides Perl_get_context(), it will all just
1880 The second, more efficient way is to use the following template for
1883 #define PERL_NO_GET_CONTEXT /* we want efficiency */
1888 static my_private_function(int arg1, int arg2);
1891 my_private_function(int arg1, int arg2)
1893 dTHX; /* fetch context */
1894 ... call many Perl API functions ...
1899 MODULE = Foo PACKAGE = Foo
1907 my_private_function(arg, 10);
1909 Note that the only two changes from the normal way of writing an
1910 extension is the addition of a C<#define PERL_NO_GET_CONTEXT> before
1911 including the Perl headers, followed by a C<dTHX;> declaration at
1912 the start of every function that will call the Perl API. (You'll
1913 know which functions need this, because the C compiler will complain
1914 that there's an undeclared identifier in those functions.) No changes
1915 are needed for the XSUBs themselves, because the XS() macro is
1916 correctly defined to pass in the implicit context if needed.
1918 The third, even more efficient way is to ape how it is done within
1922 #define PERL_NO_GET_CONTEXT /* we want efficiency */
1927 /* pTHX_ only needed for functions that call Perl API */
1928 static my_private_function(pTHX_ int arg1, int arg2);
1931 my_private_function(pTHX_ int arg1, int arg2)
1933 /* dTHX; not needed here, because THX is an argument */
1934 ... call Perl API functions ...
1939 MODULE = Foo PACKAGE = Foo
1947 my_private_function(aTHX_ arg, 10);
1949 This implementation never has to fetch the context using a function
1950 call, since it is always passed as an extra argument. Depending on
1951 your needs for simplicity or efficiency, you may mix the previous
1952 two approaches freely.
1954 Never add a comma after C<pTHX> yourself--always use the form of the
1955 macro with the underscore for functions that take explicit arguments,
1956 or the form without the argument for functions with no explicit arguments.
1958 =head2 Should I do anything special if I call perl from multiple threads?
1960 If you create interpreters in one thread and then proceed to call them in
1961 another, you need to make sure perl's own Thread Local Storage (TLS) slot is
1962 initialized correctly in each of those threads.
1964 The C<perl_alloc> and C<perl_clone> API functions will automatically set
1965 the TLS slot to the interpreter they created, so that there is no need to do
1966 anything special if the interpreter is always accessed in the same thread that
1967 created it, and that thread did not create or call any other interpreters
1968 afterwards. If that is not the case, you have to set the TLS slot of the
1969 thread before calling any functions in the Perl API on that particular
1970 interpreter. This is done by calling the C<PERL_SET_CONTEXT> macro in that
1971 thread as the first thing you do:
1973 /* do this before doing anything else with some_perl */
1974 PERL_SET_CONTEXT(some_perl);
1976 ... other Perl API calls on some_perl go here ...
1978 =head2 Future Plans and PERL_IMPLICIT_SYS
1980 Just as PERL_IMPLICIT_CONTEXT provides a way to bundle up everything
1981 that the interpreter knows about itself and pass it around, so too are
1982 there plans to allow the interpreter to bundle up everything it knows
1983 about the environment it's running on. This is enabled with the
1984 PERL_IMPLICIT_SYS macro. Currently it only works with USE_ITHREADS
1985 and USE_5005THREADS on Windows (see inside iperlsys.h).
1987 This allows the ability to provide an extra pointer (called the "host"
1988 environment) for all the system calls. This makes it possible for
1989 all the system stuff to maintain their own state, broken down into
1990 seven C structures. These are thin wrappers around the usual system
1991 calls (see win32/perllib.c) for the default perl executable, but for a
1992 more ambitious host (like the one that would do fork() emulation) all
1993 the extra work needed to pretend that different interpreters are
1994 actually different "processes", would be done here.
1996 The Perl engine/interpreter and the host are orthogonal entities.
1997 There could be one or more interpreters in a process, and one or
1998 more "hosts", with free association between them.
2000 =head1 Internal Functions
2002 All of Perl's internal functions which will be exposed to the outside
2003 world are be prefixed by C<Perl_> so that they will not conflict with XS
2004 functions or functions used in a program in which Perl is embedded.
2005 Similarly, all global variables begin with C<PL_>. (By convention,
2006 static functions start with C<S_>)
2008 Inside the Perl core, you can get at the functions either with or
2009 without the C<Perl_> prefix, thanks to a bunch of defines that live in
2010 F<embed.h>. This header file is generated automatically from
2011 F<embed.pl>. F<embed.pl> also creates the prototyping header files for
2012 the internal functions, generates the documentation and a lot of other
2013 bits and pieces. It's important that when you add a new function to the
2014 core or change an existing one, you change the data in the table at the
2015 end of F<embed.pl> as well. Here's a sample entry from that table:
2017 Apd |SV** |av_fetch |AV* ar|I32 key|I32 lval
2019 The second column is the return type, the third column the name. Columns
2020 after that are the arguments. The first column is a set of flags:
2026 This function is a part of the public API.
2030 This function has a C<Perl_> prefix; ie, it is defined as C<Perl_av_fetch>
2034 This function has documentation using the C<apidoc> feature which we'll
2035 look at in a second.
2039 Other available flags are:
2045 This is a static function and is defined as C<S_whatever>, and usually
2046 called within the sources as C<whatever(...)>.
2050 This does not use C<aTHX_> and C<pTHX> to pass interpreter context. (See
2051 L<perlguts/Background and PERL_IMPLICIT_CONTEXT>.)
2055 This function never returns; C<croak>, C<exit> and friends.
2059 This function takes a variable number of arguments, C<printf> style.
2060 The argument list should end with C<...>, like this:
2062 Afprd |void |croak |const char* pat|...
2066 This function is part of the experimental development API, and may change
2067 or disappear without notice.
2071 This function should not have a compatibility macro to define, say,
2072 C<Perl_parse> to C<parse>. It must be called as C<Perl_parse>.
2076 This function is not a member of C<CPerlObj>. If you don't know
2077 what this means, don't use it.
2081 This function isn't exported out of the Perl core.
2085 If you edit F<embed.pl>, you will need to run C<make regen_headers> to
2086 force a rebuild of F<embed.h> and other auto-generated files.
2088 =head2 Formatted Printing of IVs, UVs, and NVs
2090 If you are printing IVs, UVs, or NVS instead of the stdio(3) style
2091 formatting codes like C<%d>, C<%ld>, C<%f>, you should use the
2092 following macros for portability
2097 UVxf UV in hexadecimal
2102 These will take care of 64-bit integers and long doubles.
2105 printf("IV is %"IVdf"\n", iv);
2107 The IVdf will expand to whatever is the correct format for the IVs.
2109 If you are printing addresses of pointers, use UVxf combined
2110 with PTR2UV(), do not use %lx or %p.
2112 =head2 Pointer-To-Integer and Integer-To-Pointer
2114 Because pointer size does not necessarily equal integer size,
2115 use the follow macros to do it right.
2120 INT2PTR(pointertotype, integer)
2125 SV *sv = INT2PTR(SV*, iv);
2132 =head2 Source Documentation
2134 There's an effort going on to document the internal functions and
2135 automatically produce reference manuals from them - L<perlapi> is one
2136 such manual which details all the functions which are available to XS
2137 writers. L<perlintern> is the autogenerated manual for the functions
2138 which are not part of the API and are supposedly for internal use only.
2140 Source documentation is created by putting POD comments into the C
2144 =for apidoc sv_setiv
2146 Copies an integer into the given SV. Does not handle 'set' magic. See
2152 Please try and supply some documentation if you add functions to the
2155 =head1 Unicode Support
2157 Perl 5.6.0 introduced Unicode support. It's important for porters and XS
2158 writers to understand this support and make sure that the code they
2159 write does not corrupt Unicode data.
2161 =head2 What B<is> Unicode, anyway?
2163 In the olden, less enlightened times, we all used to use ASCII. Most of
2164 us did, anyway. The big problem with ASCII is that it's American. Well,
2165 no, that's not actually the problem; the problem is that it's not
2166 particularly useful for people who don't use the Roman alphabet. What
2167 used to happen was that particular languages would stick their own
2168 alphabet in the upper range of the sequence, between 128 and 255. Of
2169 course, we then ended up with plenty of variants that weren't quite
2170 ASCII, and the whole point of it being a standard was lost.
2172 Worse still, if you've got a language like Chinese or
2173 Japanese that has hundreds or thousands of characters, then you really
2174 can't fit them into a mere 256, so they had to forget about ASCII
2175 altogether, and build their own systems using pairs of numbers to refer
2178 To fix this, some people formed Unicode, Inc. and
2179 produced a new character set containing all the characters you can
2180 possibly think of and more. There are several ways of representing these
2181 characters, and the one Perl uses is called UTF8. UTF8 uses
2182 a variable number of bytes to represent a character, instead of just
2183 one. You can learn more about Unicode at http://www.unicode.org/
2185 =head2 How can I recognise a UTF8 string?
2187 You can't. This is because UTF8 data is stored in bytes just like
2188 non-UTF8 data. The Unicode character 200, (C<0xC8> for you hex types)
2189 capital E with a grave accent, is represented by the two bytes
2190 C<v196.172>. Unfortunately, the non-Unicode string C<chr(196).chr(172)>
2191 has that byte sequence as well. So you can't tell just by looking - this
2192 is what makes Unicode input an interesting problem.
2194 The API function C<is_utf8_string> can help; it'll tell you if a string
2195 contains only valid UTF8 characters. However, it can't do the work for
2196 you. On a character-by-character basis, C<is_utf8_char> will tell you
2197 whether the current character in a string is valid UTF8.
2199 =head2 How does UTF8 represent Unicode characters?
2201 As mentioned above, UTF8 uses a variable number of bytes to store a
2202 character. Characters with values 1...128 are stored in one byte, just
2203 like good ol' ASCII. Character 129 is stored as C<v194.129>; this
2204 continues up to character 191, which is C<v194.191>. Now we've run out of
2205 bits (191 is binary C<10111111>) so we move on; 192 is C<v195.128>. And
2206 so it goes on, moving to three bytes at character 2048.
2208 Assuming you know you're dealing with a UTF8 string, you can find out
2209 how long the first character in it is with the C<UTF8SKIP> macro:
2211 char *utf = "\305\233\340\240\201";
2214 len = UTF8SKIP(utf); /* len is 2 here */
2216 len = UTF8SKIP(utf); /* len is 3 here */
2218 Another way to skip over characters in a UTF8 string is to use
2219 C<utf8_hop>, which takes a string and a number of characters to skip
2220 over. You're on your own about bounds checking, though, so don't use it
2223 All bytes in a multi-byte UTF8 character will have the high bit set, so
2224 you can test if you need to do something special with this character
2230 /* Must treat this as UTF8 */
2231 uv = utf8_to_uv(utf);
2233 /* OK to treat this character as a byte */
2236 You can also see in that example that we use C<utf8_to_uv> to get the
2237 value of the character; the inverse function C<uv_to_utf8> is available
2238 for putting a UV into UTF8:
2241 /* Must treat this as UTF8 */
2242 utf8 = uv_to_utf8(utf8, uv);
2244 /* OK to treat this character as a byte */
2247 You B<must> convert characters to UVs using the above functions if
2248 you're ever in a situation where you have to match UTF8 and non-UTF8
2249 characters. You may not skip over UTF8 characters in this case. If you
2250 do this, you'll lose the ability to match hi-bit non-UTF8 characters;
2251 for instance, if your UTF8 string contains C<v196.172>, and you skip
2252 that character, you can never match a C<chr(200)> in a non-UTF8 string.
2255 =head2 How does Perl store UTF8 strings?
2257 Currently, Perl deals with Unicode strings and non-Unicode strings
2258 slightly differently. If a string has been identified as being UTF-8
2259 encoded, Perl will set a flag in the SV, C<SVf_UTF8>. You can check and
2260 manipulate this flag with the following macros:
2266 This flag has an important effect on Perl's treatment of the string: if
2267 Unicode data is not properly distinguished, regular expressions,
2268 C<length>, C<substr> and other string handling operations will have
2269 undesirable results.
2271 The problem comes when you have, for instance, a string that isn't
2272 flagged is UTF8, and contains a byte sequence that could be UTF8 -
2273 especially when combining non-UTF8 and UTF8 strings.
2275 Never forget that the C<SVf_UTF8> flag is separate to the PV value; you
2276 need be sure you don't accidentally knock it off while you're
2277 manipulating SVs. More specifically, you cannot expect to do this:
2286 nsv = newSVpvn(p, len);
2288 The C<char*> string does not tell you the whole story, and you can't
2289 copy or reconstruct an SV just by copying the string value. Check if the
2290 old SV has the UTF8 flag set, and act accordingly:
2294 nsv = newSVpvn(p, len);
2298 In fact, your C<frobnicate> function should be made aware of whether or
2299 not it's dealing with UTF8 data, so that it can handle the string
2302 =head2 How do I convert a string to UTF8?
2304 If you're mixing UTF8 and non-UTF8 strings, you might find it necessary
2305 to upgrade one of the strings to UTF8. If you've got an SV, the easiest
2308 sv_utf8_upgrade(sv);
2310 However, you must not do this, for example:
2313 sv_utf8_upgrade(left);
2315 If you do this in a binary operator, you will actually change one of the
2316 strings that came into the operator, and, while it shouldn't be noticeable
2317 by the end user, it can cause problems.
2319 Instead, C<bytes_to_utf8> will give you a UTF8-encoded B<copy> of its
2320 string argument. This is useful for having the data available for
2321 comparisons and so on, without harming the original SV. There's also
2322 C<utf8_to_bytes> to go the other way, but naturally, this will fail if
2323 the string contains any characters above 255 that can't be represented
2326 =head2 Is there anything else I need to know?
2328 Not really. Just remember these things:
2334 There's no way to tell if a string is UTF8 or not. You can tell if an SV
2335 is UTF8 by looking at is C<SvUTF8> flag. Don't forget to set the flag if
2336 something should be UTF8. Treat the flag as part of the PV, even though
2337 it's not - if you pass on the PV to somewhere, pass on the flag too.
2341 If a string is UTF8, B<always> use C<utf8_to_uv> to get at the value,
2342 unless C<!(*s & 0x80)> in which case you can use C<*s>.
2346 When writing to a UTF8 string, B<always> use C<uv_to_utf8>, unless
2347 C<uv < 0x80> in which case you can use C<*s = uv>.
2351 Mixing UTF8 and non-UTF8 strings is tricky. Use C<bytes_to_utf8> to get
2352 a new string which is UTF8 encoded. There are tricks you can use to
2353 delay deciding whether you need to use a UTF8 string until you get to a
2354 high character - C<HALF_UPGRADE> is one of those.
2358 =head1 Custom Operators
2360 Custom operator support is a new experimental feature that allows you to
2361 define your own ops. This is primarily to allow the building of
2362 interpreters for other languages in the Perl core, but it also allows
2363 optimizations through the creation of "macro-ops" (ops which perform the
2364 functions of multiple ops which are usually executed together, such as
2365 C<gvsv, gvsv, add>.)
2367 This feature is implemented as a new op type, C<OP_CUSTOM>. The Perl
2368 core does not "know" anything special about this op type, and so it will
2369 not be involved in any optimizations. This also means that you can
2370 define your custom ops to be any op structure - unary, binary, list and
2373 It's important to know what custom operators won't do for you. They
2374 won't let you add new syntax to Perl, directly. They won't even let you
2375 add new keywords, directly. In fact, they won't change the way Perl
2376 compiles a program at all. You have to do those changes yourself, after
2377 Perl has compiled the program. You do this either by manipulating the op
2378 tree using a C<CHECK> block and the C<B::Generate> module, or by adding
2379 a custom peephole optimizer with the C<optimize> module.
2381 When you do this, you replace ordinary Perl ops with custom ops by
2382 creating ops with the type C<OP_CUSTOM> and the C<pp_addr> of your own
2383 PP function. This should be defined in XS code, and should look like
2384 the PP ops in C<pp_*.c>. You are responsible for ensuring that your op
2385 takes the appropriate number of values from the stack, and you are
2386 responsible for adding stack marks if necessary.
2388 You should also "register" your op with the Perl interpreter so that it
2389 can produce sensible error and warning messages. Since it is possible to
2390 have multiple custom ops within the one "logical" op type C<OP_CUSTOM>,
2391 Perl uses the value of C<< o->op_ppaddr >> as a key into the
2392 C<PL_custom_op_descs> and C<PL_custom_op_names> hashes. This means you
2393 need to enter a name and description for your op at the appropriate
2394 place in the C<PL_custom_op_names> and C<PL_custom_op_descs> hashes.
2396 Forthcoming versions of C<B::Generate> (version 1.0 and above) should
2397 directly support the creation of custom ops by name; C<Opcodes::Custom>
2398 will provide functions which make it trivial to "register" custom ops to
2399 the Perl interpreter.
2403 Until May 1997, this document was maintained by Jeff Okamoto
2404 E<lt>okamoto@corp.hp.comE<gt>. It is now maintained as part of Perl
2405 itself by the Perl 5 Porters E<lt>perl5-porters@perl.orgE<gt>.
2407 With lots of help and suggestions from Dean Roehrich, Malcolm Beattie,
2408 Andreas Koenig, Paul Hudson, Ilya Zakharevich, Paul Marquess, Neil
2409 Bowers, Matthew Green, Tim Bunce, Spider Boardman, Ulrich Pfeifer,
2410 Stephen McCamant, and Gurusamy Sarathy.
2412 API Listing originally by Dean Roehrich E<lt>roehrich@cray.comE<gt>.
2414 Modifications to autogenerate the API listing (L<perlapi>) by Benjamin
2419 perlapi(1), perlintern(1), perlxs(1), perlembed(1)