3 perlfaq5 - Files and Formats ($Revision: 1.14 $, $Date: 2002/04/07 18:33:45 $)
7 This section deals with I/O and the "f" issues: filehandles, flushing,
10 =head2 How do I flush/unbuffer an output filehandle? Why must I do this?
12 The C standard I/O library (stdio) normally buffers characters sent to
13 devices. This is done for efficiency reasons so that there isn't a
14 system call for each byte. Any time you use print() or write() in
15 Perl, you go though this buffering. syswrite() circumvents stdio and
18 In most stdio implementations, the type of output buffering and the size of
19 the buffer varies according to the type of device. Disk files are block
20 buffered, often with a buffer size of more than 2k. Pipes and sockets
21 are often buffered with a buffer size between 1/2 and 2k. Serial devices
22 (e.g. modems, terminals) are normally line-buffered, and stdio sends
23 the entire line when it gets the newline.
25 Perl does not support truly unbuffered output (except insofar as you can
26 C<syswrite(OUT, $char, 1)>). What it does instead support is "command
27 buffering", in which a physical write is performed after every output
28 command. This isn't as hard on your system as unbuffering, but does
29 get the output where you want it when you want it.
31 If you expect characters to get to your device when you print them there,
32 you'll want to autoflush its handle.
33 Use select() and the C<$|> variable to control autoflushing
34 (see L<perlvar/$|> and L<perlfunc/select>):
36 $old_fh = select(OUTPUT_HANDLE);
40 Or using the traditional idiom:
42 select((select(OUTPUT_HANDLE), $| = 1)[0]);
44 Or if don't mind slowly loading several thousand lines of module code
45 just because you're afraid of the C<$|> variable:
48 open(DEV, "+</dev/tty"); # ceci n'est pas une pipe
51 or the newer IO::* modules:
54 open(DEV, ">/dev/printer"); # but is this?
59 use IO::Socket; # this one is kinda a pipe?
60 $sock = IO::Socket::INET->new(PeerAddr => 'www.perl.com',
61 PeerPort => 'http(80)',
63 die "$!" unless $sock;
66 print $sock "GET / HTTP/1.0" . "\015\012" x 2;
67 $document = join('', <$sock>);
68 print "DOC IS: $document\n";
70 Note the bizarrely hard coded carriage return and newline in their octal
71 equivalents. This is the ONLY way (currently) to assure a proper flush
72 on all platforms, including Macintosh. That's the way things work in
73 network programming: you really should specify the exact bit pattern
74 on the network line terminator. In practice, C<"\n\n"> often works,
75 but this is not portable.
77 See L<perlfaq9> for other examples of fetching URLs over the web.
79 =head2 How do I change one line in a file/delete a line in a file/insert a line in the middle of a file/append to the beginning of a file?
81 Use the Tie::File module, which is included in the standard
82 distribution since Perl 5.8.0.
84 =head2 How do I count the number of lines in a file?
86 One fairly efficient way is to count newlines in the file. The
87 following program uses a feature of tr///, as documented in L<perlop>.
88 If your text file doesn't end with a newline, then it's not really a
89 proper text file, so this may report one fewer line than you expect.
92 open(FILE, $filename) or die "Can't open `$filename': $!";
93 while (sysread FILE, $buffer, 4096) {
94 $lines += ($buffer =~ tr/\n//);
98 This assumes no funny games with newline translations.
100 =head2 How do I make a temporary file name?
102 Use the File::Temp module, see L<File::Temp> for more information.
104 use File::Temp qw/ tempfile tempdir /;
106 $dir = tempdir( CLEANUP => 1 );
107 ($fh, $filename) = tempfile( DIR => $dir );
109 # or if you don't need to know the filename
111 $fh = tempfile( DIR => $dir );
113 The File::Temp has been a standard module since Perl 5.6.1. If you
114 don't have a modern enough Perl installed, use the C<new_tmpfile>
115 class method from the IO::File module to get a filehandle opened for
116 reading and writing. Use it if you don't need to know the file's name:
119 $fh = IO::File->new_tmpfile()
120 or die "Unable to make new temporary file: $!";
122 If you're committed to creating a temporary file by hand, use the
123 process ID and/or the current time-value. If you need to have many
124 temporary files in one process, use a counter:
128 my $temp_dir = -d '/tmp' ? '/tmp' : $ENV{TMPDIR} || $ENV{TEMP};
129 my $base_name = sprintf("%s/%d-%d-0000", $temp_dir, $$, time());
133 until (defined(fileno(FH)) || $count++ > 100) {
134 $base_name =~ s/-(\d+)$/"-" . (1 + $1)/e;
135 sysopen(FH, $base_name, O_WRONLY|O_EXCL|O_CREAT);
137 if (defined(fileno(FH))
138 return (*FH, $base_name);
145 =head2 How can I manipulate fixed-record-length files?
147 The most efficient way is using pack() and unpack(). This is faster than
148 using substr() when taking many, many strings. It is slower for just a few.
150 Here is a sample chunk of code to break up and put back together again
151 some fixed-format input lines, in this case from the output of a normal,
155 # 15158 p5 T 0:00 perl /home/tchrist/scripts/now-what
156 $PS_T = 'A6 A4 A7 A5 A*';
160 ($pid, $tt, $stat, $time, $command) = unpack($PS_T, $_);
161 for $var (qw!pid tt stat time command!) {
162 print "$var: <$$var>\n";
164 print 'line=', pack($PS_T, $pid, $tt, $stat, $time, $command),
168 We've used C<$$var> in a way that forbidden by C<use strict 'refs'>.
169 That is, we've promoted a string to a scalar variable reference using
170 symbolic references. This is okay in small programs, but doesn't scale
171 well. It also only works on global variables, not lexicals.
173 =head2 How can I make a filehandle local to a subroutine? How do I pass filehandles between subroutines? How do I make an array of filehandles?
175 The fastest, simplest, and most direct way is to localize the typeglob
176 of the filehandle in question:
180 Typeglobs are fast (especially compared with the alternatives) and
181 reasonably easy to use, but they also have one subtle drawback. If you
182 had, for example, a function named TmpHandle(), or a variable named
183 %TmpHandle, you just hid it from yourself.
187 open(HostFile, "</etc/hosts") or die "no /etc/hosts: $!";
188 local $_; # <- VERY IMPORTANT
190 print if /\b127\.(0\.0\.)?1\b/;
192 # *HostFile automatically closes/disappears here
195 Here's how to use typeglobs in a loop to open and store a bunch of
196 filehandles. We'll use as values of the hash an ordered
197 pair to make it easy to sort the hash in insertion order.
199 @names = qw(motd termcap passwd hosts);
201 foreach $filename (@names) {
203 open(FH, "/etc/$filename") || die "$filename: $!";
204 $file{$filename} = [ $i++, *FH ];
207 # Using the filehandles in the array
208 foreach $name (sort { $file{$a}[0] <=> $file{$b}[0] } keys %file) {
209 my $fh = $file{$name}[1];
211 print "$name $. $line";
214 For passing filehandles to functions, the easiest way is to
215 preface them with a star, as in func(*STDIN).
216 See L<perlfaq7/"Passing Filehandles"> for details.
218 If you want to create many anonymous handles, you should check out the
219 Symbol, FileHandle, or IO::Handle (etc.) modules. Here's the equivalent
220 code with Symbol::gensym, which is reasonably light-weight:
222 foreach $filename (@names) {
225 open($fh, "/etc/$filename") || die "open /etc/$filename: $!";
226 $file{$filename} = [ $i++, $fh ];
229 Here's using the semi-object-oriented FileHandle module, which certainly
234 foreach $filename (@names) {
235 my $fh = FileHandle->new("/etc/$filename") or die "$filename: $!";
236 $file{$filename} = [ $i++, $fh ];
239 Please understand that whether the filehandle happens to be a (probably
240 localized) typeglob or an anonymous handle from one of the modules
241 in no way affects the bizarre rules for managing indirect handles.
242 See the next question.
244 =head2 How can I use a filehandle indirectly?
246 An indirect filehandle is using something other than a symbol
247 in a place that a filehandle is expected. Here are ways
248 to get indirect filehandles:
250 $fh = SOME_FH; # bareword is strict-subs hostile
251 $fh = "SOME_FH"; # strict-refs hostile; same package only
252 $fh = *SOME_FH; # typeglob
253 $fh = \*SOME_FH; # ref to typeglob (bless-able)
254 $fh = *SOME_FH{IO}; # blessed IO::Handle from *SOME_FH typeglob
256 Or, you can use the C<new> method from the FileHandle or IO modules to
257 create an anonymous filehandle, store that in a scalar variable,
258 and use it as though it were a normal filehandle.
261 $fh = FileHandle->new();
263 use IO::Handle; # 5.004 or higher
264 $fh = IO::Handle->new();
266 Then use any of those as you would a normal filehandle. Anywhere that
267 Perl is expecting a filehandle, an indirect filehandle may be used
268 instead. An indirect filehandle is just a scalar variable that contains
269 a filehandle. Functions like C<print>, C<open>, C<seek>, or
270 the C<< <FH> >> diamond operator will accept either a read filehandle
271 or a scalar variable containing one:
273 ($ifh, $ofh, $efh) = (*STDIN, *STDOUT, *STDERR);
274 print $ofh "Type it: ";
276 print $efh "What was that: $got";
278 If you're passing a filehandle to a function, you can write
279 the function in two ways:
283 print $fh "Sending to indirect filehandle\n";
286 Or it can localize a typeglob and use the filehandle directly:
290 print FH "Sending to localized filehandle\n";
293 Both styles work with either objects or typeglobs of real filehandles.
294 (They might also work with strings under some circumstances, but this
300 In the examples above, we assigned the filehandle to a scalar variable
301 before using it. That is because only simple scalar variables, not
302 expressions or subscripts of hashes or arrays, can be used with
303 built-ins like C<print>, C<printf>, or the diamond operator. Using
304 something other than a simple scalar variable as a filehandle is
305 illegal and won't even compile:
307 @fd = (*STDIN, *STDOUT, *STDERR);
308 print $fd[1] "Type it: "; # WRONG
309 $got = <$fd[0]> # WRONG
310 print $fd[2] "What was that: $got"; # WRONG
312 With C<print> and C<printf>, you get around this by using a block and
313 an expression where you would place the filehandle:
315 print { $fd[1] } "funny stuff\n";
316 printf { $fd[1] } "Pity the poor %x.\n", 3_735_928_559;
317 # Pity the poor deadbeef.
319 That block is a proper block like any other, so you can put more
320 complicated code there. This sends the message out to one of two places:
323 print { $ok ? $fd[1] : $fd[2] } "cat stat $ok\n";
324 print { $fd[ 1+ ($ok || 0) ] } "cat stat $ok\n";
326 This approach of treating C<print> and C<printf> like object methods
327 calls doesn't work for the diamond operator. That's because it's a
328 real operator, not just a function with a comma-less argument. Assuming
329 you've been storing typeglobs in your structure as we did above, you
330 can use the built-in function named C<readline> to reads a record just
331 as C<< <> >> does. Given the initialization shown above for @fd, this
332 would work, but only because readline() require a typeglob. It doesn't
333 work with objects or strings, which might be a bug we haven't fixed yet.
335 $got = readline($fd[0]);
337 Let it be noted that the flakiness of indirect filehandles is not
338 related to whether they're strings, typeglobs, objects, or anything else.
339 It's the syntax of the fundamental operators. Playing the object
340 game doesn't help you at all here.
342 =head2 How can I set up a footer format to be used with write()?
344 There's no builtin way to do this, but L<perlform> has a couple of
345 techniques to make it possible for the intrepid hacker.
347 =head2 How can I write() into a string?
349 See L<perlform/"Accessing Formatting Internals"> for an swrite() function.
351 =head2 How can I output my numbers with commas added?
353 This one from Benjamin Goldberg will do it for you:
355 s/(^[-+]?\d+?(?=(?>(?:\d{3})+)(?!\d))|\G\d{3}(?=\d))/$1,/g;
357 or written verbosely:
360 ^[-+]? # beginning of number.
361 \d{1,3}? # first digits before first comma
362 (?= # followed by, (but not included in the match) :
363 (?>(?:\d{3})+) # some positive multiple of three digits.
364 (?!\d) # an *exact* multiple, not x * 3 + 1 or whatever.
367 \G\d{3} # after the last group, get three digits
368 (?=\d) # but they have to have more digits after them.
371 =head2 How can I translate tildes (~) in a filename?
373 Use the <> (glob()) operator, documented in L<perlfunc>. Older
374 versions of Perl require that you have a shell installed that groks
375 tildes. Recent perl versions have this feature built in. The
376 File::KGlob module (available from CPAN) gives more portable glob
379 Within Perl, you may use this directly:
382 ^ ~ # find a leading tilde
384 [^/] # a non-slash character
385 * # repeated 0 or more times (0 means me)
390 : ( $ENV{HOME} || $ENV{LOGDIR} )
393 =head2 How come when I open a file read-write it wipes it out?
395 Because you're using something like this, which truncates the file and
396 I<then> gives you read-write access:
398 open(FH, "+> /path/name"); # WRONG (almost always)
400 Whoops. You should instead use this, which will fail if the file
403 open(FH, "+< /path/name"); # open for update
405 Using ">" always clobbers or creates. Using "<" never does
406 either. The "+" doesn't change this.
408 Here are examples of many kinds of file opens. Those using sysopen()
413 To open file for reading:
415 open(FH, "< $path") || die $!;
416 sysopen(FH, $path, O_RDONLY) || die $!;
418 To open file for writing, create new file if needed or else truncate old file:
420 open(FH, "> $path") || die $!;
421 sysopen(FH, $path, O_WRONLY|O_TRUNC|O_CREAT) || die $!;
422 sysopen(FH, $path, O_WRONLY|O_TRUNC|O_CREAT, 0666) || die $!;
424 To open file for writing, create new file, file must not exist:
426 sysopen(FH, $path, O_WRONLY|O_EXCL|O_CREAT) || die $!;
427 sysopen(FH, $path, O_WRONLY|O_EXCL|O_CREAT, 0666) || die $!;
429 To open file for appending, create if necessary:
431 open(FH, ">> $path") || die $!;
432 sysopen(FH, $path, O_WRONLY|O_APPEND|O_CREAT) || die $!;
433 sysopen(FH, $path, O_WRONLY|O_APPEND|O_CREAT, 0666) || die $!;
435 To open file for appending, file must exist:
437 sysopen(FH, $path, O_WRONLY|O_APPEND) || die $!;
439 To open file for update, file must exist:
441 open(FH, "+< $path") || die $!;
442 sysopen(FH, $path, O_RDWR) || die $!;
444 To open file for update, create file if necessary:
446 sysopen(FH, $path, O_RDWR|O_CREAT) || die $!;
447 sysopen(FH, $path, O_RDWR|O_CREAT, 0666) || die $!;
449 To open file for update, file must not exist:
451 sysopen(FH, $path, O_RDWR|O_EXCL|O_CREAT) || die $!;
452 sysopen(FH, $path, O_RDWR|O_EXCL|O_CREAT, 0666) || die $!;
454 To open a file without blocking, creating if necessary:
456 sysopen(FH, "/tmp/somefile", O_WRONLY|O_NDELAY|O_CREAT)
457 or die "can't open /tmp/somefile: $!":
459 Be warned that neither creation nor deletion of files is guaranteed to
460 be an atomic operation over NFS. That is, two processes might both
461 successfully create or unlink the same file! Therefore O_EXCL
462 isn't as exclusive as you might wish.
464 See also the new L<perlopentut> if you have it (new for 5.6).
466 =head2 Why do I sometimes get an "Argument list too long" when I use <*>?
468 The C<< <> >> operator performs a globbing operation (see above).
469 In Perl versions earlier than v5.6.0, the internal glob() operator forks
470 csh(1) to do the actual glob expansion, but
471 csh can't handle more than 127 items and so gives the error message
472 C<Argument list too long>. People who installed tcsh as csh won't
473 have this problem, but their users may be surprised by it.
475 To get around this, either upgrade to Perl v5.6.0 or later, do the glob
476 yourself with readdir() and patterns, or use a module like File::KGlob,
477 one that doesn't use the shell to do globbing.
479 =head2 Is there a leak/bug in glob()?
481 Due to the current implementation on some operating systems, when you
482 use the glob() function or its angle-bracket alias in a scalar
483 context, you may cause a memory leak and/or unpredictable behavior. It's
484 best therefore to use glob() only in list context.
486 =head2 How can I open a file with a leading ">" or trailing blanks?
488 Normally perl ignores trailing blanks in filenames, and interprets
489 certain leading characters (or a trailing "|") to mean something
492 The three argument form of open() lets you specify the mode
493 separately from the filename. The open() function treats
494 special mode characters and whitespace in the filename as
497 open FILE, "<", " file "; # filename is " file "
498 open FILE, ">", ">file"; # filename is ">file"
500 It may be a lot clearer to use sysopen(), though:
503 $badpath = "<<<something really wicked ";
504 sysopen (FH, $badpath, O_WRONLY | O_CREAT | O_TRUNC)
505 or die "can't open $badpath: $!";
507 =head2 How can I reliably rename a file?
509 If your operating system supports a proper mv(1) utility or its functional
510 equivalent, this works:
512 rename($old, $new) or system("mv", $old, $new);
514 It may be more portable to use the File::Copy module instead.
515 You just copy to the new file to the new name (checking return
516 values), then delete the old one. This isn't really the same
517 semantically as a rename(), which preserves meta-information like
518 permissions, timestamps, inode info, etc.
520 Newer versions of File::Copy export a move() function.
522 =head2 How can I lock a file?
524 Perl's builtin flock() function (see L<perlfunc> for details) will call
525 flock(2) if that exists, fcntl(2) if it doesn't (on perl version 5.004 and
526 later), and lockf(3) if neither of the two previous system calls exists.
527 On some systems, it may even use a different form of native locking.
528 Here are some gotchas with Perl's flock():
534 Produces a fatal error if none of the three system calls (or their
535 close equivalent) exists.
539 lockf(3) does not provide shared locking, and requires that the
540 filehandle be open for writing (or appending, or read/writing).
544 Some versions of flock() can't lock files over a network (e.g. on NFS file
545 systems), so you'd need to force the use of fcntl(2) when you build Perl.
546 But even this is dubious at best. See the flock entry of L<perlfunc>
547 and the F<INSTALL> file in the source distribution for information on
548 building Perl to do this.
550 Two potentially non-obvious but traditional flock semantics are that
551 it waits indefinitely until the lock is granted, and that its locks are
552 I<merely advisory>. Such discretionary locks are more flexible, but
553 offer fewer guarantees. This means that files locked with flock() may
554 be modified by programs that do not also use flock(). Cars that stop
555 for red lights get on well with each other, but not with cars that don't
556 stop for red lights. See the perlport manpage, your port's specific
557 documentation, or your system-specific local manpages for details. It's
558 best to assume traditional behavior if you're writing portable programs.
559 (If you're not, you should as always feel perfectly free to write
560 for your own system's idiosyncrasies (sometimes called "features").
561 Slavish adherence to portability concerns shouldn't get in the way of
562 your getting your job done.)
564 For more information on file locking, see also
565 L<perlopentut/"File Locking"> if you have it (new for 5.6).
569 =head2 Why can't I just open(FH, ">file.lock")?
571 A common bit of code B<NOT TO USE> is this:
573 sleep(3) while -e "file.lock"; # PLEASE DO NOT USE
574 open(LCK, "> file.lock"); # THIS BROKEN CODE
576 This is a classic race condition: you take two steps to do something
577 which must be done in one. That's why computer hardware provides an
578 atomic test-and-set instruction. In theory, this "ought" to work:
580 sysopen(FH, "file.lock", O_WRONLY|O_EXCL|O_CREAT)
581 or die "can't open file.lock: $!":
583 except that lamentably, file creation (and deletion) is not atomic
584 over NFS, so this won't work (at least, not every time) over the net.
585 Various schemes involving link() have been suggested, but
586 these tend to involve busy-wait, which is also subdesirable.
588 =head2 I still don't get locking. I just want to increment the number in the file. How can I do this?
590 Didn't anyone ever tell you web-page hit counters were useless?
591 They don't count number of hits, they're a waste of time, and they serve
592 only to stroke the writer's vanity. It's better to pick a random number;
593 they're more realistic.
595 Anyway, this is what you can do if you can't help yourself.
597 use Fcntl qw(:DEFAULT :flock);
598 sysopen(FH, "numfile", O_RDWR|O_CREAT) or die "can't open numfile: $!";
599 flock(FH, LOCK_EX) or die "can't flock numfile: $!";
601 seek(FH, 0, 0) or die "can't rewind numfile: $!";
602 truncate(FH, 0) or die "can't truncate numfile: $!";
603 (print FH $num+1, "\n") or die "can't write numfile: $!";
604 close FH or die "can't close numfile: $!";
606 Here's a much better web-page hit counter:
608 $hits = int( (time() - 850_000_000) / rand(1_000) );
610 If the count doesn't impress your friends, then the code might. :-)
612 =head2 All I want to do is append a small amount of text to the end of a file. Do I still have to use locking?
614 If you are on a system that correctly implements flock() and you use the
615 example appending code from "perldoc -f flock" everything will be OK
616 even if the OS you are on doesn't implement append mode correctly (if
617 such a system exists.) So if you are happy to restrict yourself to OSs
618 that implement flock() (and that's not really much of a restriction)
619 then that is what you should do.
621 If you know you are only going to use a system that does correctly
622 implement appending (i.e. not Win32) then you can omit the seek() from
625 If you know you are only writing code to run on an OS and filesystem that
626 does implement append mode correctly (a local filesystem on a modern
627 Unix for example), and you keep the file in block-buffered mode and you
628 write less than one buffer-full of output between each manual flushing
629 of the buffer then each bufferload is almost guaranteed to be written to
630 the end of the file in one chunk without getting intermingled with
631 anyone else's output. You can also use the syswrite() function which is
632 simply a wrapper around your systems write(2) system call.
634 There is still a small theoretical chance that a signal will interrupt
635 the system level write() operation before completion. There is also a
636 possibility that some STDIO implementations may call multiple system
637 level write()s even if the buffer was empty to start. There may be some
638 systems where this probability is reduced to zero.
640 =head2 How do I randomly update a binary file?
642 If you're just trying to patch a binary, in many cases something as
643 simple as this works:
645 perl -i -pe 's{window manager}{window mangler}g' /usr/bin/emacs
647 However, if you have fixed sized records, then you might do something more
650 $RECSIZE = 220; # size of record, in bytes
651 $recno = 37; # which record to update
652 open(FH, "+<somewhere") || die "can't update somewhere: $!";
653 seek(FH, $recno * $RECSIZE, 0);
654 read(FH, $record, $RECSIZE) == $RECSIZE || die "can't read record $recno: $!";
656 seek(FH, -$RECSIZE, 1);
660 Locking and error checking are left as an exercise for the reader.
661 Don't forget them or you'll be quite sorry.
663 =head2 How do I get a file's timestamp in perl?
665 If you want to retrieve the time at which the file was last
666 read, written, or had its meta-data (owner, etc) changed,
667 you use the B<-M>, B<-A>, or B<-C> file test operations as
668 documented in L<perlfunc>. These retrieve the age of the
669 file (measured against the start-time of your program) in
670 days as a floating point number. Some platforms may not have
671 all of these times. See L<perlport> for details. To
672 retrieve the "raw" time in seconds since the epoch, you
673 would call the stat function, then use localtime(),
674 gmtime(), or POSIX::strftime() to convert this into
679 $write_secs = (stat($file))[9];
680 printf "file %s updated at %s\n", $file,
681 scalar localtime($write_secs);
683 If you prefer something more legible, use the File::stat module
684 (part of the standard distribution in version 5.004 and later):
686 # error checking left as an exercise for reader.
689 $date_string = ctime(stat($file)->mtime);
690 print "file $file updated at $date_string\n";
692 The POSIX::strftime() approach has the benefit of being,
693 in theory, independent of the current locale. See L<perllocale>
696 =head2 How do I set a file's timestamp in perl?
698 You use the utime() function documented in L<perlfunc/utime>.
699 By way of example, here's a little program that copies the
700 read and write times from its first argument to all the rest
704 die "usage: cptimes timestamp_file other_files ...\n";
707 ($atime, $mtime) = (stat($timestamp))[8,9];
708 utime $atime, $mtime, @ARGV;
710 Error checking is, as usual, left as an exercise for the reader.
712 Note that utime() currently doesn't work correctly with Win95/NT
713 ports. A bug has been reported. Check it carefully before using
714 utime() on those platforms.
716 =head2 How do I print to more than one file at once?
718 If you only have to do this once, you can do this:
720 for $fh (FH1, FH2, FH3) { print $fh "whatever\n" }
722 To connect up to one filehandle to several output filehandles, it's
723 easiest to use the tee(1) program if you have it, and let it take care
726 open (FH, "| tee file1 file2 file3");
730 # make STDOUT go to three files, plus original STDOUT
731 open (STDOUT, "| tee file1 file2 file3") or die "Teeing off: $!\n";
732 print "whatever\n" or die "Writing: $!\n";
733 close(STDOUT) or die "Closing: $!\n";
735 Otherwise you'll have to write your own multiplexing print
736 function--or your own tee program--or use Tom Christiansen's,
737 at http://www.cpan.org/authors/id/TOMC/scripts/tct.gz , which is
738 written in Perl and offers much greater functionality
739 than the stock version.
741 =head2 How can I read in an entire file all at once?
743 The customary Perl approach for processing all the lines in a file is to
744 do so one line at a time:
746 open (INPUT, $file) || die "can't open $file: $!";
749 # do something with $_
751 close(INPUT) || die "can't close $file: $!";
753 This is tremendously more efficient than reading the entire file into
754 memory as an array of lines and then processing it one element at a time,
755 which is often--if not almost always--the wrong approach. Whenever
756 you see someone do this:
760 you should think long and hard about why you need everything loaded at
761 once. It's just not a scalable solution. You might also find it more
762 fun to use the standard Tie::File module, or the DB_File module's
763 $DB_RECNO bindings, which allow you to tie an array to a file so that
764 accessing an element the array actually accesses the corresponding
767 You can read the entire filehandle contents into a scalar.
771 open (INPUT, $file) || die "can't open $file: $!";
775 That temporarily undefs your record separator, and will automatically
776 close the file at block exit. If the file is already open, just use this:
778 $var = do { local $/; <INPUT> };
780 For ordinary files you can also use the read function.
782 read( INPUT, $var, -s INPUT );
784 The third argument tests the byte size of the data on the INPUT filehandle
785 and reads that many bytes into the buffer $var.
787 =head2 How can I read in a file by paragraphs?
789 Use the C<$/> variable (see L<perlvar> for details). You can either
790 set it to C<""> to eliminate empty paragraphs (C<"abc\n\n\n\ndef">,
791 for instance, gets treated as two paragraphs and not three), or
792 C<"\n\n"> to accept empty paragraphs.
794 Note that a blank line must have no blanks in it. Thus
795 S<C<"fred\n \nstuff\n\n">> is one paragraph, but C<"fred\n\nstuff\n\n"> is two.
797 =head2 How can I read a single character from a file? From the keyboard?
799 You can use the builtin C<getc()> function for most filehandles, but
800 it won't (easily) work on a terminal device. For STDIN, either use
801 the Term::ReadKey module from CPAN or use the sample code in
804 If your system supports the portable operating system programming
805 interface (POSIX), you can use the following code, which you'll note
806 turns off echo processing as well.
820 use POSIX qw(:termios_h);
822 my ($term, $oterm, $echo, $noecho, $fd_stdin);
824 $fd_stdin = fileno(STDIN);
826 $term = POSIX::Termios->new();
827 $term->getattr($fd_stdin);
828 $oterm = $term->getlflag();
830 $echo = ECHO | ECHOK | ICANON;
831 $noecho = $oterm & ~$echo;
834 $term->setlflag($noecho);
835 $term->setcc(VTIME, 1);
836 $term->setattr($fd_stdin, TCSANOW);
840 $term->setlflag($oterm);
841 $term->setcc(VTIME, 0);
842 $term->setattr($fd_stdin, TCSANOW);
848 sysread(STDIN, $key, 1);
857 The Term::ReadKey module from CPAN may be easier to use. Recent versions
858 include also support for non-portable systems as well.
861 open(TTY, "</dev/tty");
862 print "Gimme a char: ";
864 $key = ReadKey 0, *TTY;
866 printf "\nYou said %s, char number %03d\n",
869 =head2 How can I tell whether there's a character waiting on a filehandle?
871 The very first thing you should do is look into getting the Term::ReadKey
872 extension from CPAN. As we mentioned earlier, it now even has limited
873 support for non-portable (read: not open systems, closed, proprietary,
874 not POSIX, not Unix, etc) systems.
876 You should also check out the Frequently Asked Questions list in
877 comp.unix.* for things like this: the answer is essentially the same.
878 It's very system dependent. Here's one solution that works on BSD
883 vec($rin, fileno(STDIN), 1) = 1;
884 return $nfd = select($rin,undef,undef,0);
887 If you want to find out how many characters are waiting, there's
888 also the FIONREAD ioctl call to be looked at. The I<h2ph> tool that
889 comes with Perl tries to convert C include files to Perl code, which
890 can be C<require>d. FIONREAD ends up defined as a function in the
891 I<sys/ioctl.ph> file:
893 require 'sys/ioctl.ph';
895 $size = pack("L", 0);
896 ioctl(FH, FIONREAD(), $size) or die "Couldn't call ioctl: $!\n";
897 $size = unpack("L", $size);
899 If I<h2ph> wasn't installed or doesn't work for you, you can
900 I<grep> the include files by hand:
902 % grep FIONREAD /usr/include/*/*
903 /usr/include/asm/ioctls.h:#define FIONREAD 0x541B
905 Or write a small C program using the editor of champions:
908 #include <sys/ioctl.h>
910 printf("%#08x\n", FIONREAD);
913 % cc -o fionread fionread.c
917 And then hard code it, leaving porting as an exercise to your successor.
919 $FIONREAD = 0x4004667f; # XXX: opsys dependent
921 $size = pack("L", 0);
922 ioctl(FH, $FIONREAD, $size) or die "Couldn't call ioctl: $!\n";
923 $size = unpack("L", $size);
925 FIONREAD requires a filehandle connected to a stream, meaning that sockets,
926 pipes, and tty devices work, but I<not> files.
928 =head2 How do I do a C<tail -f> in perl?
934 The statement C<seek(GWFILE, 0, 1)> doesn't change the current position,
935 but it does clear the end-of-file condition on the handle, so that the
936 next <GWFILE> makes Perl try again to read something.
938 If that doesn't work (it relies on features of your stdio implementation),
939 then you need something more like this:
942 for ($curpos = tell(GWFILE); <GWFILE>; $curpos = tell(GWFILE)) {
943 # search for some stuff and put it into files
946 seek(GWFILE, $curpos, 0); # seek to where we had been
949 If this still doesn't work, look into the POSIX module. POSIX defines
950 the clearerr() method, which can remove the end of file condition on a
951 filehandle. The method: read until end of file, clearerr(), read some
952 more. Lather, rinse, repeat.
954 There's also a File::Tail module from CPAN.
956 =head2 How do I dup() a filehandle in Perl?
958 If you check L<perlfunc/open>, you'll see that several of the ways
959 to call open() should do the trick. For example:
961 open(LOG, ">>/tmp/logfile");
962 open(STDERR, ">&LOG");
964 Or even with a literal numeric descriptor:
966 $fd = $ENV{MHCONTEXTFD};
967 open(MHCONTEXT, "<&=$fd"); # like fdopen(3S)
969 Note that "<&STDIN" makes a copy, but "<&=STDIN" make
970 an alias. That means if you close an aliased handle, all
971 aliases become inaccessible. This is not true with
974 Error checking, as always, has been left as an exercise for the reader.
976 =head2 How do I close a file descriptor by number?
978 This should rarely be necessary, as the Perl close() function is to be
979 used for things that Perl opened itself, even if it was a dup of a
980 numeric descriptor as with MHCONTEXT above. But if you really have
981 to, you may be able to do this:
983 require 'sys/syscall.ph';
984 $rc = syscall(&SYS_close, $fd + 0); # must force numeric
985 die "can't sysclose $fd: $!" unless $rc == -1;
987 Or, just use the fdopen(3S) feature of open():
991 open F, "<&=$fd" or die "Cannot reopen fd=$fd: $!";
995 =head2 Why can't I use "C:\temp\foo" in DOS paths? Why doesn't `C:\temp\foo.exe` work?
997 Whoops! You just put a tab and a formfeed into that filename!
998 Remember that within double quoted strings ("like\this"), the
999 backslash is an escape character. The full list of these is in
1000 L<perlop/Quote and Quote-like Operators>. Unsurprisingly, you don't
1001 have a file called "c:(tab)emp(formfeed)oo" or
1002 "c:(tab)emp(formfeed)oo.exe" on your legacy DOS filesystem.
1004 Either single-quote your strings, or (preferably) use forward slashes.
1005 Since all DOS and Windows versions since something like MS-DOS 2.0 or so
1006 have treated C</> and C<\> the same in a path, you might as well use the
1007 one that doesn't clash with Perl--or the POSIX shell, ANSI C and C++,
1008 awk, Tcl, Java, or Python, just to mention a few. POSIX paths
1009 are more portable, too.
1011 =head2 Why doesn't glob("*.*") get all the files?
1013 Because even on non-Unix ports, Perl's glob function follows standard
1014 Unix globbing semantics. You'll need C<glob("*")> to get all (non-hidden)
1015 files. This makes glob() portable even to legacy systems. Your
1016 port may include proprietary globbing functions as well. Check its
1017 documentation for details.
1019 =head2 Why does Perl let me delete read-only files? Why does C<-i> clobber protected files? Isn't this a bug in Perl?
1021 This is elaborately and painstakingly described in the
1022 F<file-dir-perms> article in the "Far More Than You Ever Wanted To
1023 Know" collection in http://www.cpan.org/olddoc/FMTEYEWTK.tgz .
1025 The executive summary: learn how your filesystem works. The
1026 permissions on a file say what can happen to the data in that file.
1027 The permissions on a directory say what can happen to the list of
1028 files in that directory. If you delete a file, you're removing its
1029 name from the directory (so the operation depends on the permissions
1030 of the directory, not of the file). If you try to write to the file,
1031 the permissions of the file govern whether you're allowed to.
1033 =head2 How do I select a random line from a file?
1035 Here's an algorithm from the Camel Book:
1038 rand($.) < 1 && ($line = $_) while <>;
1040 This has a significant advantage in space over reading the whole
1041 file in. A simple proof by induction is available upon
1042 request if you doubt the algorithm's correctness.
1044 =head2 Why do I get weird spaces when I print an array of lines?
1050 joins together the elements of C<@lines> with a space between them.
1051 If C<@lines> were C<("little", "fluffy", "clouds")> then the above
1052 statement would print
1054 little fluffy clouds
1056 but if each element of C<@lines> was a line of text, ending a newline
1057 character C<("little\n", "fluffy\n", "clouds\n")> then it would print:
1063 If your array contains lines, just print them:
1067 =head1 AUTHOR AND COPYRIGHT
1069 Copyright (c) 1997-2002 Tom Christiansen and Nathan Torkington.
1070 All rights reserved.
1072 This documentation is free; you can redistribute it and/or modify it
1073 under the same terms as Perl itself.
1075 Irrespective of its distribution, all code examples here are in the public
1076 domain. You are permitted and encouraged to use this code and any
1077 derivatives thereof in your own programs for fun or for profit as you
1078 see fit. A simple comment in the code giving credit to the FAQ would
1079 be courteous but is not required.