This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
'v' unpacks unsigned shorts.
[perl5.git] / pod / perlpacktut.pod
CommitLineData
34babc16
JH
1=head1 NAME
2
3perlpacktut - tutorial on C<pack> and C<unpack>
4
5=head1 DESCRIPTION
6
7C<pack> and C<unpack> are two functions for transforming data according
8to a user-defined template, between the guarded way Perl stores values
47b6252e 9and some well-defined representation as might be required in the
34babc16
JH
10environment of a Perl program. Unfortunately, they're also two of
11the most misunderstood and most often overlooked functions that Perl
12provides. This tutorial will demystify them for you.
13
14
15=head1 The Basic Principle
16
17Most programming languages don't shelter the memory where variables are
18stored. In C, for instance, you can take the address of some variable,
19and the C<sizeof> operator tells you how many bytes are allocated to
20the variable. Using the address and the size, you may access the storage
21to your heart's content.
22
23In Perl, you just can't access memory at random, but the structural and
24representational conversion provided by C<pack> and C<unpack> is an
25excellent alternative. The C<pack> function converts values to a byte
26sequence containing representations according to a given specification,
27the so-called "template" argument. C<unpack> is the reverse process,
28deriving some values from the contents of a string of bytes. (Be cautioned,
29however, that not all that has been packed together can be neatly unpacked -
30a very common experience as seasoned travellers are likely to confirm.)
31
32Why, you may ask, would you need a chunk of memory containing some values
33in binary representation? One good reason is input and output accessing
34some file, a device, or a network connection, whereby this binary
35representation is either forced on you or will give you some benefit
36in processing. Another cause is passing data to some system call that
37is not available as a Perl function: C<syscall> requires you to provide
38parameters stored in the way it happens in a C program. Even text processing
39(as shown in the next section) may be simplified with judicious usage
40of these two functions.
41
42To see how (un)packing works, we'll start with a simple template
43code where the conversion is in low gear: between the contents of a byte
44sequence and a string of hexadecimal digits. Let's use C<unpack>, since
47b6252e 45this is likely to remind you of a dump program, or some desperate last
34babc16
JH
46message unfortunate programs are wont to throw at you before they expire
47into the wild blue yonder. Assuming that the variable C<$mem> holds a
48sequence of bytes that we'd like to inspect without assuming anything
49about its meaning, we can write
50
51 my( $hex ) = unpack( 'H*', $mem );
52 print "$hex\n";
53
54whereupon we might see something like this, with each pair of hex digits
55corresponding to a byte:
56
57 41204d414e204120504c414e20412043414e414c2050414e414d41
58
47b6252e 59What was in this chunk of memory? Numbers, characters, or a mixture of
34babc16
JH
60both? Assuming that we're on a computer where ASCII (or some similar)
61encoding is used: hexadecimal values in the range C<0x40> - C<0x5A>
47f22e19 62indicate an uppercase letter, and C<0x20> encodes a space. So we might
34babc16
JH
63assume it is a piece of text, which some are able to read like a tabloid;
64but others will have to get hold of an ASCII table and relive that
65firstgrader feeling. Not caring too much about which way to read this,
66we note that C<unpack> with the template code C<H> converts the contents
67of a sequence of bytes into the customary hexadecimal notation. Since
68"a sequence of" is a pretty vague indication of quantity, C<H> has been
69defined to convert just a single hexadecimal digit unless it is followed
70by a repeat count. An asterisk for the repeat count means to use whatever
71remains.
72
73The inverse operation - packing byte contents from a string of hexadecimal
74digits - is just as easily written. For instance:
75
76 my $s = pack( 'H2' x 10, map { "3$_" } ( 0..9 ) );
77 print "$s\n";
78
79Since we feed a list of ten 2-digit hexadecimal strings to C<pack>, the
80pack template should contain ten pack codes. If this is run on a computer
81with ASCII character coding, it will print C<0123456789>.
82
83
84=head1 Packing Text
85
86Let's suppose you've got to read in a data file like this:
87
88 Date |Description | Income|Expenditure
89 01/24/2001 Ahmed's Camel Emporium 1147.99
90 01/28/2001 Flea spray 24.99
91 01/29/2001 Camel rides to tourists 235.00
92
93How do we do it? You might think first to use C<split>; however, since
94C<split> collapses blank fields, you'll never know whether a record was
95income or expenditure. Oops. Well, you could always use C<substr>:
96
97 while (<>) {
98 my $date = substr($_, 0, 11);
99 my $desc = substr($_, 12, 27);
100 my $income = substr($_, 40, 7);
101 my $expend = substr($_, 52, 7);
102 ...
103 }
104
105It's not really a barrel of laughs, is it? In fact, it's worse than it
106may seem; the eagle-eyed may notice that the first field should only be
10710 characters wide, and the error has propagated right through the other
108numbers - which we've had to count by hand. So it's error-prone as well
109as horribly unfriendly.
110
111Or maybe we could use regular expressions:
7207e29d 112
34babc16
JH
113 while (<>) {
114 my($date, $desc, $income, $expend) =
115 m|(\d\d/\d\d/\d{4}) (.{27}) (.{7})(.*)|;
116 ...
117 }
118
119Urgh. Well, it's a bit better, but - well, would you want to maintain
120that?
121
122Hey, isn't Perl supposed to make this sort of thing easy? Well, it does,
123if you use the right tools. C<pack> and C<unpack> are designed to help
124you out when dealing with fixed-width data like the above. Let's have a
125look at a solution with C<unpack>:
126
127 while (<>) {
128 my($date, $desc, $income, $expend) = unpack("A10xA27xA7A*", $_);
129 ...
130 }
131
132That looks a bit nicer; but we've got to take apart that weird template.
133Where did I pull that out of?
134
135OK, let's have a look at some of our data again; in fact, we'll include
136the headers, and a handy ruler so we can keep track of where we are.
137
138 1 2 3 4 5
139 1234567890123456789012345678901234567890123456789012345678
140 Date |Description | Income|Expenditure
141 01/28/2001 Flea spray 24.99
142 01/29/2001 Camel rides to tourists 235.00
143
47b6252e 144From this, we can see that the date column stretches from column 1 to
34babc16
JH
145column 10 - ten characters wide. The C<pack>-ese for "character" is
146C<A>, and ten of them are C<A10>. So if we just wanted to extract the
147dates, we could say this:
148
149 my($date) = unpack("A10", $_);
150
151OK, what's next? Between the date and the description is a blank column;
152we want to skip over that. The C<x> template means "skip forward", so we
153want one of those. Next, we have another batch of characters, from 12 to
15438. That's 27 more characters, hence C<A27>. (Don't make the fencepost
155error - there are 27 characters between 12 and 38, not 26. Count 'em!)
156
157Now we skip another character and pick up the next 7 characters:
158
159 my($date,$description,$income) = unpack("A10xA27xA7", $_);
160
161Now comes the clever bit. Lines in our ledger which are just income and
162not expenditure might end at column 46. Hence, we don't want to tell our
163C<unpack> pattern that we B<need> to find another 12 characters; we'll
164just say "if there's anything left, take it". As you might guess from
165regular expressions, that's what the C<*> means: "use everything
166remaining".
167
168=over 3
169
170=item *
171
172Be warned, though, that unlike regular expressions, if the C<unpack>
173template doesn't match the incoming data, Perl will scream and die.
174
175=back
176
177
178Hence, putting it all together:
179
49704364 180 my($date,$description,$income,$expend) = unpack("A10xA27xA7xA*", $_);
34babc16
JH
181
182Now, that's our data parsed. I suppose what we might want to do now is
183total up our income and expenditure, and add another line to the end of
184our ledger - in the same format - saying how much we've brought in and
185how much we've spent:
186
187 while (<>) {
188 my($date, $desc, $income, $expend) = unpack("A10xA27xA7xA*", $_);
189 $tot_income += $income;
190 $tot_expend += $expend;
191 }
192
193 $tot_income = sprintf("%.2f", $tot_income); # Get them into
194 $tot_expend = sprintf("%.2f", $tot_expend); # "financial" format
195
196 $date = POSIX::strftime("%m/%d/%Y", localtime);
197
198 # OK, let's go:
199
200 print pack("A10xA27xA7xA*", $date, "Totals", $tot_income, $tot_expend);
201
202Oh, hmm. That didn't quite work. Let's see what happened:
203
204 01/24/2001 Ahmed's Camel Emporium 1147.99
205 01/28/2001 Flea spray 24.99
206 01/29/2001 Camel rides to tourists 1235.00
207 03/23/2001Totals 1235.001172.98
208
209OK, it's a start, but what happened to the spaces? We put C<x>, didn't
210we? Shouldn't it skip forward? Let's look at what L<perlfunc/pack> says:
211
212 x A null byte.
213
214Urgh. No wonder. There's a big difference between "a null byte",
215character zero, and "a space", character 32. Perl's put something
216between the date and the description - but unfortunately, we can't see
217it!
218
219What we actually need to do is expand the width of the fields. The C<A>
220format pads any non-existent characters with spaces, so we can use the
221additional spaces to line up our fields, like this:
222
223 print pack("A11 A28 A8 A*", $date, "Totals", $tot_income, $tot_expend);
224
225(Note that you can put spaces in the template to make it more readable,
226but they don't translate to spaces in the output.) Here's what we got
227this time:
228
229 01/24/2001 Ahmed's Camel Emporium 1147.99
230 01/28/2001 Flea spray 24.99
231 01/29/2001 Camel rides to tourists 1235.00
232 03/23/2001 Totals 1235.00 1172.98
233
234That's a bit better, but we still have that last column which needs to
235be moved further over. There's an easy way to fix this up:
236unfortunately, we can't get C<pack> to right-justify our fields, but we
237can get C<sprintf> to do it:
238
239 $tot_income = sprintf("%.2f", $tot_income);
240 $tot_expend = sprintf("%12.2f", $tot_expend);
241 $date = POSIX::strftime("%m/%d/%Y", localtime);
242 print pack("A11 A28 A8 A*", $date, "Totals", $tot_income, $tot_expend);
243
244This time we get the right answer:
245
246 01/28/2001 Flea spray 24.99
247 01/29/2001 Camel rides to tourists 1235.00
248 03/23/2001 Totals 1235.00 1172.98
249
250So that's how we consume and produce fixed-width data. Let's recap what
251we've seen of C<pack> and C<unpack> so far:
252
253=over 3
254
255=item *
256
257Use C<pack> to go from several pieces of data to one fixed-width
258version; use C<unpack> to turn a fixed-width-format string into several
259pieces of data.
260
261=item *
262
263The pack format C<A> means "any character"; if you're C<pack>ing and
264you've run out of things to pack, C<pack> will fill the rest up with
265spaces.
266
267=item *
268
269C<x> means "skip a byte" when C<unpack>ing; when C<pack>ing, it means
270"introduce a null byte" - that's probably not what you mean if you're
271dealing with plain text.
272
273=item *
274
275You can follow the formats with numbers to say how many characters
276should be affected by that format: C<A12> means "take 12 characters";
277C<x6> means "skip 6 bytes" or "character 0, 6 times".
278
279=item *
280
281Instead of a number, you can use C<*> to mean "consume everything else
282left".
283
284B<Warning>: when packing multiple pieces of data, C<*> only means
285"consume all of the current piece of data". That's to say
286
287 pack("A*A*", $one, $two)
288
289packs all of C<$one> into the first C<A*> and then all of C<$two> into
290the second. This is a general principle: each format character
291corresponds to one piece of data to be C<pack>ed.
292
293=back
294
295
296
297=head1 Packing Numbers
298
299So much for textual data. Let's get onto the meaty stuff that C<pack>
300and C<unpack> are best at: handling binary formats for numbers. There is,
301of course, not just one binary format - life would be too simple - but
302Perl will do all the finicky labor for you.
303
304
305=head2 Integers
306
307Packing and unpacking numbers implies conversion to and from some
308I<specific> binary representation. Leaving floating point numbers
309aside for the moment, the salient properties of any such representation
310are:
311
312=over 4
313
314=item *
315
316the number of bytes used for storing the integer,
317
318=item *
319
320whether the contents are interpreted as a signed or unsigned number,
321
322=item *
323
324the byte ordering: whether the first byte is the least or most
325significant byte (or: little-endian or big-endian, respectively).
326
327=back
328
329So, for instance, to pack 20302 to a signed 16 bit integer in your
330computer's representation you write
331
332 my $ps = pack( 's', 20302 );
333
334Again, the result is a string, now containing 2 bytes. If you print
335this string (which is, generally, not recommended) you might see
336C<ON> or C<NO> (depending on your system's byte ordering) - or something
337entirely different if your computer doesn't use ASCII character encoding.
338Unpacking C<$ps> with the same template returns the original integer value:
339
340 my( $s ) = unpack( 's', $ps );
341
342This is true for all numeric template codes. But don't expect miracles:
47b6252e 343if the packed value exceeds the allotted byte capacity, high order bits
34babc16
JH
344are silently discarded, and unpack certainly won't be able to pull them
345back out of some magic hat. And, when you pack using a signed template
346code such as C<s>, an excess value may result in the sign bit
347getting set, and unpacking this will smartly return a negative value.
348
34916 bits won't get you too far with integers, but there is C<l> and C<L>
350for signed and unsigned 32-bit integers. And if this is not enough and
351your system supports 64 bit integers you can push the limits much closer
352to infinity with pack codes C<q> and C<Q>. A notable exception is provided
353by pack codes C<i> and C<I> for signed and unsigned integers of the
354"local custom" variety: Such an integer will take up as many bytes as
355a local C compiler returns for C<sizeof(int)>, but it'll use I<at least>
35632 bits.
357
358Each of the integer pack codes C<sSlLqQ> results in a fixed number of bytes,
359no matter where you execute your program. This may be useful for some
360applications, but it does not provide for a portable way to pass data
361structures between Perl and C programs (bound to happen when you call
362XS extensions or the Perl function C<syscall>), or when you read or
363write binary files. What you'll need in this case are template codes that
364depend on what your local C compiler compiles when you code C<short> or
365C<unsigned long>, for instance. These codes and their corresponding
366byte lengths are shown in the table below. Since the C standard leaves
367much leeway with respect to the relative sizes of these data types, actual
368values may vary, and that's why the values are given as expressions in
369C and Perl. (If you'd like to use values from C<%Config> in your program
370you have to import it with C<use Config>.)
371
372 signed unsigned byte length in C byte length in Perl
f8b4d74f
WL
373 s! S! sizeof(short) $Config{shortsize}
374 i! I! sizeof(int) $Config{intsize}
375 l! L! sizeof(long) $Config{longsize}
d832b8f6 376 q! Q! sizeof(long long) $Config{longlongsize}
34babc16
JH
377
378The C<i!> and C<I!> codes aren't different from C<i> and C<I>; they are
379tolerated for completeness' sake.
380
381
382=head2 Unpacking a Stack Frame
383
384Requesting a particular byte ordering may be necessary when you work with
47f22e19 385binary data coming from some specific architecture whereas your program could
34babc16
JH
386run on a totally different system. As an example, assume you have 24 bytes
387containing a stack frame as it happens on an Intel 8086:
388
389 +---------+ +----+----+ +---------+
390 TOS: | IP | TOS+4:| FL | FH | FLAGS TOS+14:| SI |
391 +---------+ +----+----+ +---------+
392 | CS | | AL | AH | AX | DI |
393 +---------+ +----+----+ +---------+
394 | BL | BH | BX | BP |
395 +----+----+ +---------+
396 | CL | CH | CX | DS |
397 +----+----+ +---------+
398 | DL | DH | DX | ES |
399 +----+----+ +---------+
400
401First, we note that this time-honored 16-bit CPU uses little-endian order,
402and that's why the low order byte is stored at the lower address. To
9dc383df 403unpack such a (unsigned) short we'll have to use code C<v>. A repeat
34babc16
JH
404count unpacks all 12 shorts:
405
406 my( $ip, $cs, $flags, $ax, $bx, $cd, $dx, $si, $di, $bp, $ds, $es ) =
407 unpack( 'v12', $frame );
408
409Alternatively, we could have used C<C> to unpack the individually
410accessible byte registers FL, FH, AL, AH, etc.:
411
412 my( $fl, $fh, $al, $ah, $bl, $bh, $cl, $ch, $dl, $dh ) =
413 unpack( 'C10', substr( $frame, 4, 10 ) );
414
415It would be nice if we could do this in one fell swoop: unpack a short,
416back up a little, and then unpack 2 bytes. Since Perl I<is> nice, it
417proffers the template code C<X> to back up one byte. Putting this all
418together, we may now write:
419
420 my( $ip, $cs,
421 $flags,$fl,$fh,
422 $ax,$al,$ah, $bx,$bl,$bh, $cx,$cl,$ch, $dx,$dl,$dh,
423 $si, $di, $bp, $ds, $es ) =
424 unpack( 'v2' . ('vXXCC' x 5) . 'v5', $frame );
425
49704364
LW
426(The clumsy construction of the template can be avoided - just read on!)
427
47f22e19 428We've taken some pains to construct the template so that it matches
34babc16
JH
429the contents of our frame buffer. Otherwise we'd either get undefined values,
430or C<unpack> could not unpack all. If C<pack> runs out of items, it will
47f22e19
WL
431supply null strings (which are coerced into zeroes whenever the pack code
432says so).
34babc16
JH
433
434
435=head2 How to Eat an Egg on a Net
436
437The pack code for big-endian (high order byte at the lowest address) is
438C<n> for 16 bit and C<N> for 32 bit integers. You use these codes
439if you know that your data comes from a compliant architecture, but,
440surprisingly enough, you should also use these pack codes if you
441exchange binary data, across the network, with some system that you
442know next to nothing about. The simple reason is that this
443order has been chosen as the I<network order>, and all standard-fearing
444programs ought to follow this convention. (This is, of course, a stern
445backing for one of the Lilliputian parties and may well influence the
446political development there.) So, if the protocol expects you to send
447a message by sending the length first, followed by just so many bytes,
448you could write:
449
450 my $buf = pack( 'N', length( $msg ) ) . $msg;
451
452or even:
453
454 my $buf = pack( 'NA*', length( $msg ), $msg );
455
456and pass C<$buf> to your send routine. Some protocols demand that the
457count should include the length of the count itself: then just add 4
458to the data length. (But make sure to read L<"Lengths and Widths"> before
459you really code this!)
460
461
462
463=head2 Floating point Numbers
464
465For packing floating point numbers you have the choice between the
466pack codes C<f> and C<d> which pack into (or unpack from) single-precision or
467double-precision representation as it is provided by your system. (There
468is no such thing as a network representation for reals, so if you want
469to send your real numbers across computer boundaries, you'd better stick
470to ASCII representation, unless you're absolutely sure what's on the other
471end of the line.)
472
473
474
475=head1 Exotic Templates
476
477
478=head2 Bit Strings
479
480Bits are the atoms in the memory world. Access to individual bits may
481have to be used either as a last resort or because it is the most
482convenient way to handle your data. Bit string (un)packing converts
483between strings containing a series of C<0> and C<1> characters and
484a sequence of bytes each containing a group of 8 bits. This is almost
485as simple as it sounds, except that there are two ways the contents of
486a byte may be written as a bit string. Let's have a look at an annotated
487byte:
488
489 7 6 5 4 3 2 1 0
490 +-----------------+
491 | 1 0 0 0 1 1 0 0 |
492 +-----------------+
493 MSB LSB
494
495It's egg-eating all over again: Some think that as a bit string this should
496be written "10001100" i.e. beginning with the most significant bit, others
497insist on "00110001". Well, Perl isn't biased, so that's why we have two bit
498string codes:
499
500 $byte = pack( 'B8', '10001100' ); # start with MSB
501 $byte = pack( 'b8', '00110001' ); # start with LSB
502
503It is not possible to pack or unpack bit fields - just integral bytes.
504C<pack> always starts at the next byte boundary and "rounds up" to the
505next multiple of 8 by adding zero bits as required. (If you do want bit
506fields, there is L<perlfunc/vec>. Or you could implement bit field
507handling at the character string level, using split, substr, and
508concatenation on unpacked bit strings.)
509
510To illustrate unpacking for bit strings, we'll decompose a simple
511status register (a "-" stands for a "reserved" bit):
512
513 +-----------------+-----------------+
514 | S Z - A - P - C | - - - - O D I T |
515 +-----------------+-----------------+
516 MSB LSB MSB LSB
517
518Converting these two bytes to a string can be done with the unpack
519template C<'b16'>. To obtain the individual bit values from the bit
47f22e19 520string we use C<split> with the "empty" separator pattern which dissects
34babc16
JH
521into individual characters. Bit values from the "reserved" positions are
522simply assigned to C<undef>, a convenient notation for "I don't care where
523this goes".
524
49704364 525 ($carry, undef, $parity, undef, $auxcarry, undef, $zero, $sign,
34babc16 526 $trace, $interrupt, $direction, $overflow) =
47f22e19 527 split( //, unpack( 'b16', $status ) );
34babc16
JH
528
529We could have used an unpack template C<'b12'> just as well, since the
530last 4 bits can be ignored anyway.
531
532
533=head2 Uuencoding
534
535Another odd-man-out in the template alphabet is C<u>, which packs an
536"uuencoded string". ("uu" is short for Unix-to-Unix.) Chances are that
537you won't ever need this encoding technique which was invented to overcome
538the shortcomings of old-fashioned transmission mediums that do not support
539other than simple ASCII data. The essential recipe is simple: Take three
540bytes, or 24 bits. Split them into 4 six-packs, adding a space (0x20) to
541each. Repeat until all of the data is blended. Fold groups of 4 bytes into
542lines no longer than 60 and garnish them in front with the original byte count
543(incremented by 0x20) and a C<"\n"> at the end. - The C<pack> chef will
544prepare this for you, a la minute, when you select pack code C<u> on the menu:
545
546 my $uubuf = pack( 'u', $bindat );
547
548A repeat count after C<u> sets the number of bytes to put into an
549uuencoded line, which is the maximum of 45 by default, but could be
550set to some (smaller) integer multiple of three. C<unpack> simply ignores
551the repeat count.
552
553
554=head2 Doing Sums
555
556An even stranger template code is C<%>E<lt>I<number>E<gt>. First, because
557it's used as a prefix to some other template code. Second, because it
558cannot be used in C<pack> at all, and third, in C<unpack>, doesn't return the
559data as defined by the template code it precedes. Instead it'll give you an
560integer of I<number> bits that is computed from the data value by
561doing sums. For numeric unpack codes, no big feat is achieved:
562
563 my $buf = pack( 'iii', 100, 20, 3 );
564 print unpack( '%32i3', $buf ), "\n"; # prints 123
565
566For string values, C<%> returns the sum of the byte values saving
567you the trouble of a sum loop with C<substr> and C<ord>:
568
569 print unpack( '%32A*', "\x01\x10" ), "\n"; # prints 17
570
571Although the C<%> code is documented as returning a "checksum":
572don't put your trust in such values! Even when applied to a small number
573of bytes, they won't guarantee a noticeable Hamming distance.
574
575In connection with C<b> or C<B>, C<%> simply adds bits, and this can be put
576to good use to count set bits efficiently:
577
578 my $bitcount = unpack( '%32b*', $mask );
579
580And an even parity bit can be determined like this:
581
582 my $evenparity = unpack( '%1b*', $mask );
583
584
585=head2 Unicode
586
587Unicode is a character set that can represent most characters in most of
588the world's languages, providing room for over one million different
589characters. Unicode 3.1 specifies 94,140 characters: The Basic Latin
590characters are assigned to the numbers 0 - 127. The Latin-1 Supplement with
591characters that are used in several European languages is in the next
592range, up to 255. After some more Latin extensions we find the character
47b6252e 593sets from languages using non-Roman alphabets, interspersed with a
34babc16
JH
594variety of symbol sets such as currency symbols, Zapf Dingbats or Braille.
595(You might want to visit L<www.unicode.org> for a look at some of
596them - my personal favourites are Telugu and Kannada.)
597
598The Unicode character sets associates characters with integers. Encoding
599these numbers in an equal number of bytes would more than double the
47b6252e 600requirements for storing texts written in Latin alphabets.
34babc16
JH
601The UTF-8 encoding avoids this by storing the most common (from a western
602point of view) characters in a single byte while encoding the rarer
603ones in three or more bytes.
604
605So what has this got to do with C<pack>? Well, if you want to convert
606between a Unicode number and its UTF-8 representation you can do so by
607using template code C<U>. As an example, let's produce the UTF-8
608representation of the Euro currency symbol (code number 0x20AC):
609
610 $UTF8{Euro} = pack( 'U', 0x20AC );
611
612Inspecting C<$UTF8{Euro}> shows that it contains 3 bytes: "\xe2\x82\xac". The
613round trip can be completed with C<unpack>:
614
615 $Unicode{Euro} = unpack( 'U', $UTF8{Euro} );
616
617Usually you'll want to pack or unpack UTF-8 strings:
618
619 # pack and unpack the Hebrew alphabet
620 my $alefbet = pack( 'U*', 0x05d0..0x05ea );
621 my @hebrew = unpack( 'U*', $utf );
622
623
47f22e19
WL
624=head2 Another Portable Binary Encoding
625
626The pack code C<w> has been added to support a portable binary data
627encoding scheme that goes way beyond simple integers. (Details can
628be found at L<Casbah.org>, the Scarab project.) A BER (Binary Encoded
629Representation) compressed unsigned integer stores base 128
630digits, most significant digit first, with as few digits as possible.
631Bit eight (the high bit) is set on each byte except the last. There
632is no size limit to BER encoding, but Perl won't go to extremes.
633
634 my $berbuf = pack( 'w*', 1, 128, 128+1, 128*128+127 );
635
636A hex dump of C<$berbuf>, with spaces inserted at the right places,
637shows 01 8100 8101 81807F. Since the last byte is always less than
638128, C<unpack> knows where to stop.
639
34babc16 640
49704364
LW
641=head1 Template Grouping
642
643Prior to Perl 5.8, repetitions of templates had to be made by
644C<x>-multiplication of template strings. Now there is a better way as
645we may use the pack codes C<(> and C<)> combined with a repeat count.
646The C<unpack> template from the Stack Frame example can simply
647be written like this:
648
649 unpack( 'v2 (vXXCC)5 v5', $frame )
650
651Let's explore this feature a little more. We'll begin with the equivalent of
652
653 join( '', map( substr( $_, 0, 1 ), @str ) )
654
655which returns a string consisting of the first character from each string.
656Using pack, we can write
657
658 pack( '(A)'.@str, @str )
659
660or, because a repeat count C<*> means "repeat as often as required",
661simply
662
663 pack( '(A)*', @str )
664
665(Note that the template C<A*> would only have packed C<$str[0]> in full
666length.)
ffc145e8 667
49704364
LW
668To pack dates stored as triplets ( day, month, year ) in an array C<@dates>
669into a sequence of byte, byte, short integer we can write
670
671 $pd = pack( '(CCS)*', map( @$_, @dates ) );
672
673To swap pairs of characters in a string (with even length) one could use
674several techniques. First, let's use C<x> and C<X> to skip forward and back:
675
676 $s = pack( '(A)*', unpack( '(xAXXAx)*', $s ) );
677
678We can also use C<@> to jump to an offset, with 0 being the position where
679we were when the last C<(> was encountered:
680
681 $s = pack( '(A)*', unpack( '(@1A @0A @2)*', $s ) );
682
683Finally, there is also an entirely different approach by unpacking big
684endian shorts and packing them in the reverse byte order:
685
686 $s = pack( '(v)*', unpack( '(n)*', $s );
687
688
34babc16
JH
689=head1 Lengths and Widths
690
691=head2 String Lengths
692
693In the previous section we've seen a network message that was constructed
694by prefixing the binary message length to the actual message. You'll find
695that packing a length followed by so many bytes of data is a
696frequently used recipe since appending a null byte won't work
47b6252e 697if a null byte may be part of the data. Here is an example where both
34babc16
JH
698techniques are used: after two null terminated strings with source and
699destination address, a Short Message (to a mobile phone) is sent after
700a length byte:
701
702 my $msg = pack( 'Z*Z*CA*', $src, $dst, length( $sm ), $sm );
703
704Unpacking this message can be done with the same template:
705
706 ( $src, $dst, $len, $sm ) = unpack( 'Z*Z*CA*', $msg );
707
47b6252e 708There's a subtle trap lurking in the offing: Adding another field after
34babc16
JH
709the Short Message (in variable C<$sm>) is all right when packing, but this
710cannot be unpacked naively:
711
712 # pack a message
713 my $msg = pack( 'Z*Z*CA*C', $src, $dst, length( $sm ), $sm, $prio );
7207e29d 714
34babc16
JH
715 # unpack fails - $prio remains undefined!
716 ( $src, $dst, $len, $sm, $prio ) = unpack( 'Z*Z*CA*C', $msg );
717
718The pack code C<A*> gobbles up all remaining bytes, and C<$prio> remains
719undefined! Before we let disappointment dampen the morale: Perl's got
720the trump card to make this trick too, just a little further up the sleeve.
721Watch this:
722
723 # pack a message: ASCIIZ, ASCIIZ, length/string, byte
724 my $msg = pack( 'Z* Z* C/A* C', $src, $dst, $sm, $prio );
725
726 # unpack
727 ( $src, $dst, $sm, $prio ) = unpack( 'Z* Z* C/A* C', $msg );
728
729Combining two pack codes with a slash (C</>) associates them with a single
730value from the argument list. In C<pack>, the length of the argument is
731taken and packed according to the first code while the argument itself
732is added after being converted with the template code after the slash.
733This saves us the trouble of inserting the C<length> call, but it is
734in C<unpack> where we really score: The value of the length byte marks the
735end of the string to be taken from the buffer. Since this combination
f8b4d74f 736doesn't make sense except when the second pack code isn't C<a*>, C<A*>
34babc16
JH
737or C<Z*>, Perl won't let you.
738
739The pack code preceding C</> may be anything that's fit to represent a
740number: All the numeric binary pack codes, and even text codes such as
741C<A4> or C<Z*>:
742
743 # pack/unpack a string preceded by its length in ASCII
744 my $buf = pack( 'A4/A*', "Humpty-Dumpty" );
745 # unpack $buf: '13 Humpty-Dumpty'
746 my $txt = unpack( 'A4/A*', $buf );
747
47b6252e
NC
748C</> is not implemented in Perls before 5.6, so if your code is required to
749work on older Perls you'll need to C<unpack( 'Z* Z* C')> to get the length,
750then use it to make a new unpack string. For example
751
752 # pack a message: ASCIIZ, ASCIIZ, length, string, byte (5.005 compatible)
753 my $msg = pack( 'Z* Z* C A* C', $src, $dst, length $sm, $sm, $prio );
754
755 # unpack
756 ( undef, undef, $len) = unpack( 'Z* Z* C', $msg );
757 ($src, $dst, $sm, $prio) = unpack ( "Z* Z* x A$len C", $msg );
758
759But that second C<unpack> is rushing ahead. It isn't using a simple literal
760string for the template. So maybe we should introduce...
34babc16
JH
761
762=head2 Dynamic Templates
763
764So far, we've seen literals used as templates. If the list of pack
765items doesn't have fixed length, an expression constructing the
49704364
LW
766template is required (whenever, for some reason, C<()*> cannot be used).
767Here's an example: To store named string values in a way that can be
768conveniently parsed by a C program, we create a sequence of names and
769null terminated ASCII strings, with C<=> between the name and the value,
770followed by an additional delimiting null byte. Here's how:
34babc16 771
49704364 772 my $env = pack( '(A*A*Z*)' . keys( %Env ) . 'C',
47f22e19
WL
773 map( { ( $_, '=', $Env{$_} ) } keys( %Env ) ), 0 );
774
775Let's examine the cogs of this byte mill, one by one. There's the C<map>
776call, creating the items we intend to stuff into the C<$env> buffer:
777to each key (in C<$_>) it adds the C<=> separator and the hash entry value.
778Each triplet is packed with the template code sequence C<A*A*Z*> that
49704364 779is repeated according to the number of keys. (Yes, that's what the C<keys>
fe854a6f 780function returns in scalar context.) To get the very last null byte,
47f22e19
WL
781we add a C<0> at the end of the C<pack> list, to be packed with C<C>.
782(Attentive readers may have noticed that we could have omitted the 0.)
34babc16
JH
783
784For the reverse operation, we'll have to determine the number of items
785in the buffer before we can let C<unpack> rip it apart:
786
47b6252e 787 my $n = $env =~ tr/\0// - 1;
49704364 788 my %env = map( split( /=/, $_ ), unpack( "(Z*)$n", $env ) );
34babc16 789
47b6252e 790The C<tr> counts the null bytes. The C<unpack> call returns a list of
47f22e19 791name-value pairs each of which is taken apart in the C<map> block.
34babc16
JH
792
793
49704364
LW
794=head2 Counting Repetitions
795
796Rather than storing a sentinel at the end of a data item (or a list of items),
797we could precede the data with a count. Again, we pack keys and values of
798a hash, preceding each with an unsigned short length count, and up front
799we store the number of pairs:
800
801 my $env = pack( 'S(S/A* S/A*)*', scalar keys( %Env ), %Env );
802
803This simplifies the reverse operation as the number of repetitions can be
804unpacked with the C</> code:
805
806 my %env = unpack( 'S/(S/A* S/A*)', $env );
807
808Note that this is one of the rare cases where you cannot use the same
809template for C<pack> and C<unpack> because C<pack> can't determine
810a repeat count for a C<()>-group.
811
812
34babc16
JH
813=head1 Packing and Unpacking C Structures
814
815In previous sections we have seen how to pack numbers and character
816strings. If it were not for a couple of snags we could conclude this
817section right away with the terse remark that C structures don't
818contain anything else, and therefore you already know all there is to it.
819Sorry, no: read on, please.
820
821=head2 The Alignment Pit
822
823In the consideration of speed against memory requirements the balance
824has been tilted in favor of faster execution. This has influenced the
825way C compilers allocate memory for structures: On architectures
826where a 16-bit or 32-bit operand can be moved faster between places in
827memory, or to or from a CPU register, if it is aligned at an even or
828multiple-of-four or even at a multiple-of eight address, a C compiler
829will give you this speed benefit by stuffing extra bytes into structures.
830If you don't cross the C shoreline this is not likely to cause you any
47b6252e
NC
831grief (although you should care when you design large data structures,
832or you want your code to be portable between architectures (you do want
833that, don't you?)).
34babc16
JH
834
835To see how this affects C<pack> and C<unpack>, we'll compare these two
836C structures:
837
838 typedef struct {
839 char c1;
840 short s;
841 char c2;
842 long l;
843 } gappy_t;
844
845 typedef struct {
846 long l;
847 short s;
848 char c1;
849 char c2;
850 } dense_t;
851
852Typically, a C compiler allocates 12 bytes to a C<gappy_t> variable, but
853requires only 8 bytes for a C<dense_t>. After investigating this further,
854we can draw memory maps, showing where the extra 4 bytes are hidden:
855
856 0 +4 +8 +12
857 +--+--+--+--+--+--+--+--+--+--+--+--+
858 |c1|xx| s |c2|xx|xx|xx| l | xx = fill byte
859 +--+--+--+--+--+--+--+--+--+--+--+--+
860 gappy_t
861
862 0 +4 +8
863 +--+--+--+--+--+--+--+--+
864 | l | h |c1|c2|
865 +--+--+--+--+--+--+--+--+
866 dense_t
867
868And that's where the first quirk strikes: C<pack> and C<unpack>
869templates have to be stuffed with C<x> codes to get those extra fill bytes.
870
871The natural question: "Why can't Perl compensate for the gaps?" warrants
872an answer. One good reason is that C compilers might provide (non-ANSI)
873extensions permitting all sorts of fancy control over the way structures
874are aligned, even at the level of an individual structure field. And, if
875this were not enough, there is an insidious thing called C<union> where
876the amount of fill bytes cannot be derived from the alignment of the next
877item alone.
878
879OK, so let's bite the bullet. Here's one way to get the alignment right
880by inserting template codes C<x>, which don't take a corresponding item
881from the list:
882
883 my $gappy = pack( 'cxs cxxx l!', $c1, $s, $c2, $l );
884
885Note the C<!> after C<l>: We want to make sure that we pack a long
47b6252e
NC
886integer as it is compiled by our C compiler. And even now, it will only
887work for the platforms where the compiler aligns things as above.
888And somebody somewhere has a platform where it doesn't.
889[Probably a Cray, where C<short>s, C<int>s and C<long>s are all 8 bytes. :-)]
34babc16
JH
890
891Counting bytes and watching alignments in lengthy structures is bound to
892be a drag. Isn't there a way we can create the template with a simple
893program? Here's a C program that does the trick:
894
895 #include <stdio.h>
896 #include <stddef.h>
897
898 typedef struct {
899 char fc1;
900 short fs;
901 char fc2;
902 long fl;
903 } gappy_t;
904
905 #define Pt(struct,field,tchar) \
906 printf( "@%d%s ", offsetof(struct,field), # tchar );
907
d832b8f6 908 int main() {
34babc16
JH
909 Pt( gappy_t, fc1, c );
910 Pt( gappy_t, fs, s! );
911 Pt( gappy_t, fc2, c );
912 Pt( gappy_t, fl, l! );
913 printf( "\n" );
914 }
915
916The output line can be used as a template in a C<pack> or C<unpack> call:
917
918 my $gappy = pack( '@0c @2s! @4c @8l!', $c1, $s, $c2, $l );
919
920Gee, yet another template code - as if we hadn't plenty. But
921C<@> saves our day by enabling us to specify the offset from the beginning
922of the pack buffer to the next item: This is just the value
923the C<offsetof> macro (defined in C<E<lt>stddef.hE<gt>>) returns when
924given a C<struct> type and one of its field names ("member-designator" in
925C standardese).
926
49704364
LW
927Neither using offsets nor adding C<x>'s to bridge the gaps is satisfactory.
928(Just imagine what happens if the structure changes.) What we really need
929is a way of saying "skip as many bytes as required to the next multiple of N".
930In fluent Templatese, you say this with C<x!N> where N is replaced by the
931appropriate value. Here's the next version of our struct packaging:
932
933 my $gappy = pack( 'c x!2 s c x!4 l!', $c1, $s, $c2, $l );
934
935That's certainly better, but we still have to know how long all the
936integers are, and portability is far away. Rather than C<2>,
937for instance, we want to say "however long a short is". But this can be
938done by enclosing the appropriate pack code in brackets: C<[s]>. So, here's
939the very best we can do:
940
941 my $gappy = pack( 'c x![s] s c x![l!] l!', $c1, $s, $c2, $l );
942
34babc16
JH
943
944=head2 Alignment, Take 2
945
946I'm afraid that we're not quite through with the alignment catch yet. The
947hydra raises another ugly head when you pack arrays of structures:
948
949 typedef struct {
950 short count;
951 char glyph;
952 } cell_t;
953
954 typedef cell_t buffer_t[BUFLEN];
955
956Where's the catch? Padding is neither required before the first field C<count>,
957nor between this and the next field C<glyph>, so why can't we simply pack
958like this:
959
960 # something goes wrong here:
961 pack( 's!a' x @buffer,
962 map{ ( $_->{count}, $_->{glyph} ) } @buffer );
963
964This packs C<3*@buffer> bytes, but it turns out that the size of
965C<buffer_t> is four times C<BUFLEN>! The moral of the story is that
966the required alignment of a structure or array is propagated to the
967next higher level where we have to consider padding I<at the end>
968of each component as well. Thus the correct template is:
969
970 pack( 's!ax' x @buffer,
971 map{ ( $_->{count}, $_->{glyph} ) } @buffer );
972
47b6252e
NC
973=head2 Alignment, Take 3
974
975And even if you take all the above into account, ANSI still lets this:
976
977 typedef struct {
978 char foo[2];
979 } foo_t;
34babc16 980
47b6252e
NC
981vary in size. The alignment constraint of the structure can be greater than
982any of its elements. [And if you think that this doesn't affect anything
983common, dismember the next cellphone that you see. Many have ARM cores, and
984the ARM structure rules make C<sizeof (foo_t)> == 4]
34babc16
JH
985
986=head2 Pointers for How to Use Them
987
988The title of this section indicates the second problem you may run into
989sooner or later when you pack C structures. If the function you intend
990to call expects a, say, C<void *> value, you I<cannot> simply take
991a reference to a Perl variable. (Although that value certainly is a
992memory address, it's not the address where the variable's contents are
993stored.)
994
995Template code C<P> promises to pack a "pointer to a fixed length string".
996Isn't this what we want? Let's try:
997
998 # allocate some storage and pack a pointer to it
999 my $memory = "\x00" x $size;
1000 my $memptr = pack( 'P', $memory );
1001
1002But wait: doesn't C<pack> just return a sequence of bytes? How can we pass this
1003string of bytes to some C code expecting a pointer which is, after all,
1004nothing but a number? The answer is simple: We have to obtain the numeric
1005address from the bytes returned by C<pack>.
1006
1007 my $ptr = unpack( 'L!', $memptr );
1008
1009Obviously this assumes that it is possible to typecast a pointer
1010to an unsigned long and vice versa, which frequently works but should not
1011be taken as a universal law. - Now that we have this pointer the next question
1012is: How can we put it to good use? We need a call to some C function
1013where a pointer is expected. The read(2) system call comes to mind:
1014
1015 ssize_t read(int fd, void *buf, size_t count);
1016
1017After reading L<perlfunc> explaining how to use C<syscall> we can write
1018this Perl function copying a file to standard output:
1019
1020 require 'syscall.ph';
1021 sub cat($){
1022 my $path = shift();
1023 my $size = -s $path;
1024 my $memory = "\x00" x $size; # allocate some memory
1025 my $ptr = unpack( 'L', pack( 'P', $memory ) );
1026 open( F, $path ) || die( "$path: cannot open ($!)\n" );
1027 my $fd = fileno(F);
1028 my $res = syscall( &SYS_read, fileno(F), $ptr, $size );
1029 print $memory;
1030 close( F );
1031 }
1032
1033This is neither a specimen of simplicity nor a paragon of portability but
1034it illustrates the point: We are able to sneak behind the scenes and
1035access Perl's otherwise well-guarded memory! (Important note: Perl's
1036C<syscall> does I<not> require you to construct pointers in this roundabout
1037way. You simply pass a string variable, and Perl forwards the address.)
1038
1039How does C<unpack> with C<P> work? Imagine some pointer in the buffer
1040about to be unpacked: If it isn't the null pointer (which will smartly
1041produce the C<undef> value) we have a start address - but then what?
1042Perl has no way of knowing how long this "fixed length string" is, so
1043it's up to you to specify the actual size as an explicit length after C<P>.
1044
1045 my $mem = "abcdefghijklmn";
1046 print unpack( 'P5', pack( 'P', $mem ) ); # prints "abcde"
1047
1048As a consequence, C<pack> ignores any number or C<*> after C<P>.
1049
1050
1051Now that we have seen C<P> at work, we might as well give C<p> a whirl.
1052Why do we need a second template code for packing pointers at all? The
1053answer lies behind the simple fact that an C<unpack> with C<p> promises
1054a null-terminated string starting at the address taken from the buffer,
1055and that implies a length for the data item to be returned:
1056
1057 my $buf = pack( 'p', "abc\x00efhijklmn" );
1058 print unpack( 'p', $buf ); # prints "abc"
1059
1060
1061
1062Albeit this is apt to be confusing: As a consequence of the length being
1063implied by the string's length, a number after pack code C<p> is a repeat
1064count, not a length as after C<P>.
1065
1066
1067Using C<pack(..., $x)> with C<P> or C<p> to get the address where C<$x> is
1068actually stored must be used with circumspection. Perl's internal machinery
1069considers the relation between a variable and that address as its very own
1070private matter and doesn't really care that we have obtained a copy. Therefore:
1071
1072=over 4
1073
1074=item *
1075
1076Do not use C<pack> with C<p> or C<P> to obtain the address of variable
1077that's bound to go out of scope (and thereby freeing its memory) before you
1078are done with using the memory at that address.
1079
1080=item *
1081
1082Be very careful with Perl operations that change the value of the
1083variable. Appending something to the variable, for instance, might require
1084reallocation of its storage, leaving you with a pointer into no-man's land.
1085
1086=item *
1087
1088Don't think that you can get the address of a Perl variable
1089when it is stored as an integer or double number! C<pack('P', $x)> will
1090force the variable's internal representation to string, just as if you
1091had written something like C<$x .= ''>.
1092
1093=back
1094
1095It's safe, however, to P- or p-pack a string literal, because Perl simply
1096allocates an anonymous variable.
1097
1098
1099
1100=head1 Pack Recipes
1101
1102Here are a collection of (possibly) useful canned recipes for C<pack>
1103and C<unpack>:
1104
1105 # Convert IP address for socket functions
1106 pack( "C4", split /\./, "123.4.5.6" );
1107
1108 # Count the bits in a chunk of memory (e.g. a select vector)
1109 unpack( '%32b*', $mask );
1110
1111 # Determine the endianness of your system
1112 $is_little_endian = unpack( 'c', pack( 's', 1 ) );
1113 $is_big_endian = unpack( 'xc', pack( 's', 1 ) );
1114
1115 # Determine the number of bits in a native integer
1116 $bits = unpack( '%32I!', ~0 );
1117
1118 # Prepare argument for the nanosleep system call
1119 my $timespec = pack( 'L!L!', $secs, $nanosecs );
1120
f8b4d74f
WL
1121For a simple memory dump we unpack some bytes into just as
1122many pairs of hex digits, and use C<map> to handle the traditional
1123spacing - 16 bytes to a line:
1124
34babc16 1125 my $i;
49704364
LW
1126 print map( ++$i % 16 ? "$_ " : "$_\n",
1127 unpack( 'H2' x length( $mem ), $mem ) ),
f8b4d74f 1128 length( $mem ) % 16 ? "\n" : '';
34babc16
JH
1129
1130
47f22e19
WL
1131=head1 Funnies Section
1132
1133 # Pulling digits out of nowhere...
1134 print unpack( 'C', pack( 'x' ) ),
1135 unpack( '%B*', pack( 'A' ) ),
1136 unpack( 'H', pack( 'A' ) ),
1137 unpack( 'A', unpack( 'C', pack( 'A' ) ) ), "\n";
1138
1139 # One for the road ;-)
1140 my $advice = pack( 'all u can in a van' );
1141
1142
34babc16
JH
1143=head1 Authors
1144
1145Simon Cozens and Wolfgang Laun.
1146