Debian Codegraph @

upstream/debian-unstable/man1/perlfaq5.1

raw
1 .\" -*- mode: troff; coding: utf-8 -*-
2 .\" Automatically generated by Pod::Man 5.0102 (Pod::Simple 3.45)
3 .\"
4 .\" Standard preamble:
5 .\" ========================================================================
6 .de Sp \" Vertical space (when we can't use .PP)
7 .if t .sp .5v
8 .if n .sp
9 ..
10 .de Vb \" Begin verbatim text
11 .ft CW
12 .nf
13 .ne \\$1
14 ..
15 .de Ve \" End verbatim text
16 .ft R
17 .fi
18 ..
19 .\" \*(C` and \*(C' are quotes in nroff, nothing in troff, for use with C<>.
20 .ie n \{\
21 . ds C` ""
22 . ds C' ""
23 'br\}
24 .el\{\
25 . ds C`
26 . ds C'
27 'br\}
28 .\"
29 .\" Escape single quotes in literal strings from groff's Unicode transform.
30 .ie \n(.g .ds Aq \(aq
31 .el .ds Aq '
32 .\"
33 .\" If the F register is >0, we'll generate index entries on stderr for
34 .\" titles (.TH), headers (.SH), subsections (.SS), items (.Ip), and index
35 .\" entries marked with X<> in POD. Of course, you'll have to process the
36 .\" output yourself in some meaningful fashion.
37 .\"
38 .\" Avoid warning from groff about undefined register 'F'.
39 .de IX
40 ..
41 .nr rF 0
42 .if \n(.g .if rF .nr rF 1
43 .if (\n(rF:(\n(.g==0)) \{\
44 . if \nF \{\
45 . de IX
46 . tm Index:\\$1\t\\n%\t"\\$2"
47 ..
48 . if !\nF==2 \{\
49 . nr % 0
50 . nr F 2
51 . \}
52 . \}
53 .\}
54 .rr rF
55 .\" ========================================================================
56 .\"
57 .IX Title "PERLFAQ5 1"
58 .TH PERLFAQ5 1 2025-11-16 "perl v5.40.1" "Perl Programmers Reference Guide"
59 .\" For nroff, turn off justification. Always turn off hyphenation; it makes
60 .\" way too many mistakes in technical documents.
61 .if n .ad l
62 .nh
63 .SH NAME
64 perlfaq5 \- Files and Formats
65 .SH VERSION
66 .IX Header "VERSION"
67 version 5.20240218
68 .SH DESCRIPTION
69 .IX Header "DESCRIPTION"
70 This section deals with I/O and the "f" issues: filehandles, flushing,
71 formats, and footers.
72 .SS "How do I flush/unbuffer an output filehandle? Why must I do this?"
73 .IX Xref "flush buffer unbuffer autoflush"
74 .IX Subsection "How do I flush/unbuffer an output filehandle? Why must I do this?"
75 (contributed by brian d foy)
76 .PP
77 You might like to read Mark Jason Dominus's "Suffering From Buffering"
78 at <http://perl.plover.com/FAQs/Buffering.html> .
79 .PP
80 Perl normally buffers output so it doesn't make a system call for every
81 bit of output. By saving up output, it makes fewer expensive system calls.
82 For instance, in this little bit of code, you want to print a dot to the
83 screen for every line you process to watch the progress of your program.
84 Instead of seeing a dot for every line, Perl buffers the output and you
85 have a long wait before you see a row of 50 dots all at once:
86 .PP
87 .Vb 4
88 \& # long wait, then row of dots all at once
89 \& while( <> ) {
90 \& print ".";
91 \& print "\en" unless ++$count % 50;
92 \&
93 \& #... expensive line processing operations
94 \& }
95 .Ve
96 .PP
97 To get around this, you have to unbuffer the output filehandle, in this
98 case, \f(CW\*(C`STDOUT\*(C'\fR. You can set the special variable \f(CW$|\fR to a true value
99 (mnemonic: making your filehandles "piping hot"):
100 .PP
101 .Vb 1
102 \& $|++;
103 \&
104 \& # dot shown immediately
105 \& while( <> ) {
106 \& print ".";
107 \& print "\en" unless ++$count % 50;
108 \&
109 \& #... expensive line processing operations
110 \& }
111 .Ve
112 .PP
113 The \f(CW$|\fR is one of the per-filehandle special variables, so each
114 filehandle has its own copy of its value. If you want to merge
115 standard output and standard error for instance, you have to unbuffer
116 each (although STDERR might be unbuffered by default):
117 .PP
118 .Vb 7
119 \& {
120 \& my $previous_default = select(STDOUT); # save previous default
121 \& $|++; # autoflush STDOUT
122 \& select(STDERR);
123 \& $|++; # autoflush STDERR, to be sure
124 \& select($previous_default); # restore previous default
125 \& }
126 \&
127 \& # now should alternate . and +
128 \& while( 1 ) {
129 \& sleep 1;
130 \& print STDOUT ".";
131 \& print STDERR "+";
132 \& print STDOUT "\en" unless ++$count % 25;
133 \& }
134 .Ve
135 .PP
136 Besides the \f(CW$|\fR special variable, you can use \f(CW\*(C`binmode\*(C'\fR to give
137 your filehandle a \f(CW\*(C`:unix\*(C'\fR layer, which is unbuffered:
138 .PP
139 .Vb 1
140 \& binmode( STDOUT, ":unix" );
141 \&
142 \& while( 1 ) {
143 \& sleep 1;
144 \& print ".";
145 \& print "\en" unless ++$count % 50;
146 \& }
147 .Ve
148 .PP
149 For more information on output layers, see the entries for \f(CW\*(C`binmode\*(C'\fR
150 and open in perlfunc, and the PerlIO module documentation.
151 .PP
152 If you are using IO::Handle or one of its subclasses, you can
153 call the \f(CW\*(C`autoflush\*(C'\fR method to change the settings of the
154 filehandle:
155 .PP
156 .Vb 3
157 \& use IO::Handle;
158 \& open my( $io_fh ), ">", "output.txt";
159 \& $io_fh\->autoflush(1);
160 .Ve
161 .PP
162 The IO::Handle objects also have a \f(CW\*(C`flush\*(C'\fR method. You can flush
163 the buffer any time you want without auto-buffering
164 .PP
165 .Vb 1
166 \& $io_fh\->flush;
167 .Ve
168 .SS "How do I change, delete, or insert a line in a file, or append to the beginning of a file?"
169 .IX Xref "file, editing"
170 .IX Subsection "How do I change, delete, or insert a line in a file, or append to the beginning of a file?"
171 (contributed by brian d foy)
172 .PP
173 The basic idea of inserting, changing, or deleting a line from a text
174 file involves reading and printing the file to the point you want to
175 make the change, making the change, then reading and printing the rest
176 of the file. Perl doesn't provide random access to lines (especially
177 since the record input separator, \f(CW$/\fR, is mutable), although modules
178 such as Tie::File can fake it.
179 .PP
180 A Perl program to do these tasks takes the basic form of opening a
181 file, printing its lines, then closing the file:
182 .PP
183 .Vb 2
184 \& open my $in, \*(Aq<\*(Aq, $file or die "Can\*(Aqt read old file: $!";
185 \& open my $out, \*(Aq>\*(Aq, "$file.new" or die "Can\*(Aqt write new file: $!";
186 \&
187 \& while( <$in> ) {
188 \& print $out $_;
189 \& }
190 \&
191 \& close $out;
192 .Ve
193 .PP
194 Within that basic form, add the parts that you need to insert, change,
195 or delete lines.
196 .PP
197 To prepend lines to the beginning, print those lines before you enter
198 the loop that prints the existing lines.
199 .PP
200 .Vb 2
201 \& open my $in, \*(Aq<\*(Aq, $file or die "Can\*(Aqt read old file: $!";
202 \& open my $out, \*(Aq>\*(Aq, "$file.new" or die "Can\*(Aqt write new file: $!";
203 \&
204 \& print $out "# Add this line to the top\en"; # <\-\-\- HERE\*(AqS THE MAGIC
205 \&
206 \& while( <$in> ) {
207 \& print $out $_;
208 \& }
209 \&
210 \& close $out;
211 .Ve
212 .PP
213 To change existing lines, insert the code to modify the lines inside
214 the \f(CW\*(C`while\*(C'\fR loop. In this case, the code finds all lowercased
215 versions of "perl" and uppercases them. The happens for every line, so
216 be sure that you're supposed to do that on every line!
217 .PP
218 .Vb 2
219 \& open my $in, \*(Aq<\*(Aq, $file or die "Can\*(Aqt read old file: $!";
220 \& open my $out, \*(Aq>\*(Aq, "$file.new" or die "Can\*(Aqt write new file: $!";
221 \&
222 \& print $out "# Add this line to the top\en";
223 \&
224 \& while( <$in> ) {
225 \& s/\eb(perl)\eb/Perl/g;
226 \& print $out $_;
227 \& }
228 \&
229 \& close $out;
230 .Ve
231 .PP
232 To change only a particular line, the input line number, \f(CW$.\fR, is
233 useful. First read and print the lines up to the one you want to
234 change. Next, read the single line you want to change, change it, and
235 print it. After that, read the rest of the lines and print those:
236 .PP
237 .Vb 4
238 \& while( <$in> ) { # print the lines before the change
239 \& print $out $_;
240 \& last if $. == 4; # line number before change
241 \& }
242 \&
243 \& my $line = <$in>;
244 \& $line =~ s/\eb(perl)\eb/Perl/g;
245 \& print $out $line;
246 \&
247 \& while( <$in> ) { # print the rest of the lines
248 \& print $out $_;
249 \& }
250 .Ve
251 .PP
252 To skip lines, use the looping controls. The \f(CW\*(C`next\*(C'\fR in this example
253 skips comment lines, and the \f(CW\*(C`last\*(C'\fR stops all processing once it
254 encounters either \f(CW\*(C`_\|_END_\|_\*(C'\fR or \f(CW\*(C`_\|_DATA_\|_\*(C'\fR.
255 .PP
256 .Vb 5
257 \& while( <$in> ) {
258 \& next if /^\es+#/; # skip comment lines
259 \& last if /^_\|_(END|DATA)_\|_$/; # stop at end of code marker
260 \& print $out $_;
261 \& }
262 .Ve
263 .PP
264 Do the same sort of thing to delete a particular line by using \f(CW\*(C`next\*(C'\fR
265 to skip the lines you don't want to show up in the output. This
266 example skips every fifth line:
267 .PP
268 .Vb 4
269 \& while( <$in> ) {
270 \& next unless $. % 5;
271 \& print $out $_;
272 \& }
273 .Ve
274 .PP
275 If, for some odd reason, you really want to see the whole file at once
276 rather than processing line-by-line, you can slurp it in (as long as
277 you can fit the whole thing in memory!):
278 .PP
279 .Vb 2
280 \& open my $in, \*(Aq<\*(Aq, $file or die "Can\*(Aqt read old file: $!"
281 \& open my $out, \*(Aq>\*(Aq, "$file.new" or die "Can\*(Aqt write new file: $!";
282 \&
283 \& my $content = do { local $/; <$in> }; # slurp!
284 \&
285 \& # do your magic here
286 \&
287 \& print $out $content;
288 .Ve
289 .PP
290 Modules such as Path::Tiny and Tie::File can help with that
291 too. If you can, however, avoid reading the entire file at once. Perl
292 won't give that memory back to the operating system until the process
293 finishes.
294 .PP
295 You can also use Perl one-liners to modify a file in-place. The
296 following changes all 'Fred' to 'Barney' in \fIinFile.txt\fR, overwriting
297 the file with the new contents. With the \f(CW\*(C`\-p\*(C'\fR switch, Perl wraps a
298 \&\f(CW\*(C`while\*(C'\fR loop around the code you specify with \f(CW\*(C`\-e\*(C'\fR, and \f(CW\*(C`\-i\*(C'\fR turns
299 on in-place editing. The current line is in \f(CW$_\fR. With \f(CW\*(C`\-p\*(C'\fR, Perl
300 automatically prints the value of \f(CW$_\fR at the end of the loop. See
301 perlrun for more details.
302 .PP
303 .Vb 1
304 \& perl \-pi \-e \*(Aqs/Fred/Barney/\*(Aq inFile.txt
305 .Ve
306 .PP
307 To make a backup of \f(CW\*(C`inFile.txt\*(C'\fR, give \f(CW\*(C`\-i\*(C'\fR a file extension to add:
308 .PP
309 .Vb 1
310 \& perl \-pi.bak \-e \*(Aqs/Fred/Barney/\*(Aq inFile.txt
311 .Ve
312 .PP
313 To change only the fifth line, you can add a test checking \f(CW$.\fR, the
314 input line number, then only perform the operation when the test
315 passes:
316 .PP
317 .Vb 1
318 \& perl \-pi \-e \*(Aqs/Fred/Barney/ if $. == 5\*(Aq inFile.txt
319 .Ve
320 .PP
321 To add lines before a certain line, you can add a line (or lines!)
322 before Perl prints \f(CW$_\fR:
323 .PP
324 .Vb 1
325 \& perl \-pi \-e \*(Aqprint "Put before third line\en" if $. == 3\*(Aq inFile.txt
326 .Ve
327 .PP
328 You can even add a line to the beginning of a file, since the current
329 line prints at the end of the loop:
330 .PP
331 .Vb 1
332 \& perl \-pi \-e \*(Aqprint "Put before first line\en" if $. == 1\*(Aq inFile.txt
333 .Ve
334 .PP
335 To insert a line after one already in the file, use the \f(CW\*(C`\-n\*(C'\fR switch.
336 It's just like \f(CW\*(C`\-p\*(C'\fR except that it doesn't print \f(CW$_\fR at the end of
337 the loop, so you have to do that yourself. In this case, print \f(CW$_\fR
338 first, then print the line that you want to add.
339 .PP
340 .Vb 1
341 \& perl \-ni \-e \*(Aqprint; print "Put after fifth line\en" if $. == 5\*(Aq inFile.txt
342 .Ve
343 .PP
344 To delete lines, only print the ones that you want.
345 .PP
346 .Vb 1
347 \& perl \-ni \-e \*(Aqprint if /d/\*(Aq inFile.txt
348 .Ve
349 .SS "How do I count the number of lines in a file?"
350 .IX Xref "file, counting lines lines line"
351 .IX Subsection "How do I count the number of lines in a file?"
352 (contributed by brian d foy)
353 .PP
354 Conceptually, the easiest way to count the lines in a file is to
355 simply read them and count them:
356 .PP
357 .Vb 2
358 \& my $count = 0;
359 \& while( <$fh> ) { $count++; }
360 .Ve
361 .PP
362 You don't really have to count them yourself, though, since Perl
363 already does that with the \f(CW$.\fR variable, which is the current line
364 number from the last filehandle read:
365 .PP
366 .Vb 2
367 \& 1 while( <$fh> );
368 \& my $count = $.;
369 .Ve
370 .PP
371 If you want to use \f(CW$.\fR, you can reduce it to a simple one-liner,
372 like one of these:
373 .PP
374 .Vb 1
375 \& % perl \-lne \*(Aq} print $.; {\*(Aq file
376 \&
377 \& % perl \-lne \*(AqEND { print $. }\*(Aq file
378 .Ve
379 .PP
380 Those can be rather inefficient though. If they aren't fast enough for
381 you, you might just read chunks of data and count the number of
382 newlines:
383 .PP
384 .Vb 6
385 \& my $lines = 0;
386 \& open my($fh), \*(Aq<:raw\*(Aq, $filename or die "Can\*(Aqt open $filename: $!";
387 \& while( sysread $fh, $buffer, 4096 ) {
388 \& $lines += ( $buffer =~ tr/\en// );
389 \& }
390 \& close $fh;
391 .Ve
392 .PP
393 However, that doesn't work if the line ending isn't a newline. You
394 might change that \f(CW\*(C`tr///\*(C'\fR to a \f(CW\*(C`s///\*(C'\fR so you can count the number of
395 times the input record separator, \f(CW$/\fR, shows up:
396 .PP
397 .Vb 6
398 \& my $lines = 0;
399 \& open my($fh), \*(Aq<:raw\*(Aq, $filename or die "Can\*(Aqt open $filename: $!";
400 \& while( sysread $fh, $buffer, 4096 ) {
401 \& $lines += ( $buffer =~ s|$/||g; );
402 \& }
403 \& close $fh;
404 .Ve
405 .PP
406 If you don't mind shelling out, the \f(CW\*(C`wc\*(C'\fR command is usually the
407 fastest, even with the extra interprocess overhead. Ensure that you
408 have an untainted filename though:
409 .PP
410 .Vb 1
411 \& #!perl \-T
412 \&
413 \& $ENV{PATH} = undef;
414 \&
415 \& my $lines;
416 \& if( $filename =~ /^([0\-9a\-z_.]+)\ez/ ) {
417 \& $lines = \`/usr/bin/wc \-l $1\`
418 \& chomp $lines;
419 \& }
420 .Ve
421 .SS "How do I delete the last N lines from a file?"
422 .IX Xref "lines file"
423 .IX Subsection "How do I delete the last N lines from a file?"
424 (contributed by brian d foy)
425 .PP
426 The easiest conceptual solution is to count the lines in the
427 file then start at the beginning and print the number of lines
428 (minus the last N) to a new file.
429 .PP
430 Most often, the real question is how you can delete the last N lines
431 without making more than one pass over the file, or how to do it
432 without a lot of copying. The easy concept is the hard reality when
433 you might have millions of lines in your file.
434 .PP
435 One trick is to use File::ReadBackwards, which starts at the end of
436 the file. That module provides an object that wraps the real filehandle
437 to make it easy for you to move around the file. Once you get to the
438 spot you need, you can get the actual filehandle and work with it as
439 normal. In this case, you get the file position at the end of the last
440 line you want to keep and truncate the file to that point:
441 .PP
442 .Vb 1
443 \& use File::ReadBackwards;
444 \&
445 \& my $filename = \*(Aqtest.txt\*(Aq;
446 \& my $Lines_to_truncate = 2;
447 \&
448 \& my $bw = File::ReadBackwards\->new( $filename )
449 \& or die "Could not read backwards in [$filename]: $!";
450 \&
451 \& my $lines_from_end = 0;
452 \& until( $bw\->eof or $lines_from_end == $Lines_to_truncate ) {
453 \& print "Got: ", $bw\->readline;
454 \& $lines_from_end++;
455 \& }
456 \&
457 \& truncate( $filename, $bw\->tell );
458 .Ve
459 .PP
460 The File::ReadBackwards module also has the advantage of setting
461 the input record separator to a regular expression.
462 .PP
463 You can also use the Tie::File module which lets you access
464 the lines through a tied array. You can use normal array operations
465 to modify your file, including setting the last index and using
466 \&\f(CW\*(C`splice\*(C'\fR.
467 .ie n .SS "How can I use Perl's ""\-i"" option from within a program?"
468 .el .SS "How can I use Perl's \f(CW\-i\fP option from within a program?"
469 .IX Xref "-i in-place"
470 .IX Subsection "How can I use Perl's -i option from within a program?"
471 \&\f(CW\*(C`\-i\*(C'\fR sets the value of Perl's \f(CW$^I\fR variable, which in turn affects
472 the behavior of \f(CW\*(C`<>\*(C'\fR; see perlrun for more details. By
473 modifying the appropriate variables directly, you can get the same
474 behavior within a larger program. For example:
475 .PP
476 .Vb 10
477 \& # ...
478 \& {
479 \& local($^I, @ARGV) = (\*(Aq.orig\*(Aq, glob("*.c"));
480 \& while (<>) {
481 \& if ($. == 1) {
482 \& print "This line should appear at the top of each file\en";
483 \& }
484 \& s/\eb(p)earl\eb/${1}erl/i; # Correct typos, preserving case
485 \& print;
486 \& close ARGV if eof; # Reset $.
487 \& }
488 \& }
489 \& # $^I and @ARGV return to their old values here
490 .Ve
491 .PP
492 This block modifies all the \f(CW\*(C`.c\*(C'\fR files in the current directory,
493 leaving a backup of the original data from each file in a new
494 \&\f(CW\*(C`.c.orig\*(C'\fR file.
495 .SS "How can I copy a file?"
496 .IX Xref "copy file, copy File::Copy"
497 .IX Subsection "How can I copy a file?"
498 (contributed by brian d foy)
499 .PP
500 Use the File::Copy module. It comes with Perl and can do a
501 true copy across file systems, and it does its magic in
502 a portable fashion.
503 .PP
504 .Vb 1
505 \& use File::Copy;
506 \&
507 \& copy( $original, $new_copy ) or die "Copy failed: $!";
508 .Ve
509 .PP
510 If you can't use File::Copy, you'll have to do the work yourself:
511 open the original file, open the destination file, then print
512 to the destination file as you read the original. You also have to
513 remember to copy the permissions, owner, and group to the new file.
514 .SS "How do I make a temporary file name?"
515 .IX Xref "file, temporary"
516 .IX Subsection "How do I make a temporary file name?"
517 If you don't need to know the name of the file, you can use \f(CWopen()\fR
518 with \f(CW\*(C`undef\*(C'\fR in place of the file name. In Perl 5.8 or later, the
519 \&\f(CWopen()\fR function creates an anonymous temporary file:
520 .PP
521 .Vb 1
522 \& open my $tmp, \*(Aq+>\*(Aq, undef or die $!;
523 .Ve
524 .PP
525 Otherwise, you can use the File::Temp module.
526 .PP
527 .Vb 1
528 \& use File::Temp qw/ tempfile tempdir /;
529 \&
530 \& my $dir = tempdir( CLEANUP => 1 );
531 \& ($fh, $filename) = tempfile( DIR => $dir );
532 \&
533 \& # or if you don\*(Aqt need to know the filename
534 \&
535 \& my $fh = tempfile( DIR => $dir );
536 .Ve
537 .PP
538 The File::Temp has been a standard module since Perl 5.6.1. If you
539 don't have a modern enough Perl installed, use the \f(CW\*(C`new_tmpfile\*(C'\fR
540 class method from the IO::File module to get a filehandle opened for
541 reading and writing. Use it if you don't need to know the file's name:
542 .PP
543 .Vb 3
544 \& use IO::File;
545 \& my $fh = IO::File\->new_tmpfile()
546 \& or die "Unable to make new temporary file: $!";
547 .Ve
548 .PP
549 If you're committed to creating a temporary file by hand, use the
550 process ID and/or the current time-value. If you need to have many
551 temporary files in one process, use a counter:
552 .PP
553 .Vb 6
554 \& BEGIN {
555 \& use Fcntl;
556 \& use File::Spec;
557 \& my $temp_dir = File::Spec\->tmpdir();
558 \& my $file_base = sprintf "%d\-%d\-0000", $$, time;
559 \& my $base_name = File::Spec\->catfile($temp_dir, $file_base);
560 \&
561 \& sub temp_file {
562 \& my $fh;
563 \& my $count = 0;
564 \& until( defined(fileno($fh)) || $count++ > 100 ) {
565 \& $base_name =~ s/\-(\ed+)$/"\-" . (1 + $1)/e;
566 \& # O_EXCL is required for security reasons.
567 \& sysopen $fh, $base_name, O_WRONLY|O_EXCL|O_CREAT;
568 \& }
569 \&
570 \& if( defined fileno($fh) ) {
571 \& return ($fh, $base_name);
572 \& }
573 \& else {
574 \& return ();
575 \& }
576 \& }
577 \& }
578 .Ve
579 .SS "How can I manipulate fixed-record-length files?"
580 .IX Xref "fixed-length file, fixed-length records"
581 .IX Subsection "How can I manipulate fixed-record-length files?"
582 The most efficient way is using \fBpack()\fR and
583 \&\fBunpack()\fR. This is faster than using
584 \&\fBsubstr()\fR when taking many, many strings. It is
585 slower for just a few.
586 .PP
587 Here is a sample chunk of code to break up and put back together again
588 some fixed-format input lines, in this case from the output of a normal,
589 Berkeley-style ps:
590 .PP
591 .Vb 10
592 \& # sample input line:
593 \& # 15158 p5 T 0:00 perl /home/tchrist/scripts/now\-what
594 \& my $PS_T = \*(AqA6 A4 A7 A5 A*\*(Aq;
595 \& open my $ps, \*(Aq\-|\*(Aq, \*(Aqps\*(Aq;
596 \& print scalar <$ps>;
597 \& my @fields = qw( pid tt stat time command );
598 \& while (<$ps>) {
599 \& my %process;
600 \& @process{@fields} = unpack($PS_T, $_);
601 \& for my $field ( @fields ) {
602 \& print "$field: <$process{$field}>\en";
603 \& }
604 \& print \*(Aqline=\*(Aq, pack($PS_T, @process{@fields} ), "\en";
605 \& }
606 .Ve
607 .PP
608 We've used a hash slice in order to easily handle the fields of each row.
609 Storing the keys in an array makes it easy to operate on them as a
610 group or loop over them with \f(CW\*(C`for\*(C'\fR. It also avoids polluting the program
611 with global variables and using symbolic references.
612 .SS "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?"
613 .IX Xref "filehandle, local filehandle, passing filehandle, reference"
614 .IX Subsection "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?"
615 As of perl5.6, \fBopen()\fR autovivifies file and directory handles
616 as references if you pass it an uninitialized scalar variable.
617 You can then pass these references just like any other scalar,
618 and use them in the place of named handles.
619 .PP
620 .Vb 1
621 \& open my $fh, $file_name;
622 \&
623 \& open local $fh, $file_name;
624 \&
625 \& print $fh "Hello World!\en";
626 \&
627 \& process_file( $fh );
628 .Ve
629 .PP
630 If you like, you can store these filehandles in an array or a hash.
631 If you access them directly, they aren't simple scalars and you
632 need to give \f(CW\*(C`print\*(C'\fR a little help by placing the filehandle
633 reference in braces. Perl can only figure it out on its own when
634 the filehandle reference is a simple scalar.
635 .PP
636 .Vb 1
637 \& my @fhs = ( $fh1, $fh2, $fh3 );
638 \&
639 \& for( $i = 0; $i <= $#fhs; $i++ ) {
640 \& print {$fhs[$i]} "just another Perl answer, \en";
641 \& }
642 .Ve
643 .PP
644 Before perl5.6, you had to deal with various typeglob idioms
645 which you may see in older code.
646 .PP
647 .Vb 3
648 \& open FILE, "> $filename";
649 \& process_typeglob( *FILE );
650 \& process_reference( \e*FILE );
651 \&
652 \& sub process_typeglob { local *FH = shift; print FH "Typeglob!" }
653 \& sub process_reference { local $fh = shift; print $fh "Reference!" }
654 .Ve
655 .PP
656 If you want to create many anonymous handles, you should
657 check out the Symbol or IO::Handle modules.
658 .SS "How can I use a filehandle indirectly?"
659 .IX Xref "filehandle, indirect"
660 .IX Subsection "How can I use a filehandle indirectly?"
661 An indirect filehandle is the use of something other than a symbol
662 in a place that a filehandle is expected. Here are ways
663 to get indirect filehandles:
664 .PP
665 .Vb 5
666 \& $fh = SOME_FH; # bareword is strict\-subs hostile
667 \& $fh = "SOME_FH"; # strict\-refs hostile; same package only
668 \& $fh = *SOME_FH; # typeglob
669 \& $fh = \e*SOME_FH; # ref to typeglob (bless\-able)
670 \& $fh = *SOME_FH{IO}; # blessed IO::Handle from *SOME_FH typeglob
671 .Ve
672 .PP
673 Or, you can use the \f(CW\*(C`new\*(C'\fR method from one of the IO::* modules to
674 create an anonymous filehandle and store that in a scalar variable.
675 .PP
676 .Vb 2
677 \& use IO::Handle; # 5.004 or higher
678 \& my $fh = IO::Handle\->new();
679 .Ve
680 .PP
681 Then use any of those as you would a normal filehandle. Anywhere that
682 Perl is expecting a filehandle, an indirect filehandle may be used
683 instead. An indirect filehandle is just a scalar variable that contains
684 a filehandle. Functions like \f(CW\*(C`print\*(C'\fR, \f(CW\*(C`open\*(C'\fR, \f(CW\*(C`seek\*(C'\fR, or
685 the \f(CW\*(C`<FH>\*(C'\fR diamond operator will accept either a named filehandle
686 or a scalar variable containing one:
687 .PP
688 .Vb 4
689 \& ($ifh, $ofh, $efh) = (*STDIN, *STDOUT, *STDERR);
690 \& print $ofh "Type it: ";
691 \& my $got = <$ifh>
692 \& print $efh "What was that: $got";
693 .Ve
694 .PP
695 If you're passing a filehandle to a function, you can write
696 the function in two ways:
697 .PP
698 .Vb 4
699 \& sub accept_fh {
700 \& my $fh = shift;
701 \& print $fh "Sending to indirect filehandle\en";
702 \& }
703 .Ve
704 .PP
705 Or it can localize a typeglob and use the filehandle directly:
706 .PP
707 .Vb 4
708 \& sub accept_fh {
709 \& local *FH = shift;
710 \& print FH "Sending to localized filehandle\en";
711 \& }
712 .Ve
713 .PP
714 Both styles work with either objects or typeglobs of real filehandles.
715 (They might also work with strings under some circumstances, but this
716 is risky.)
717 .PP
718 .Vb 2
719 \& accept_fh(*STDOUT);
720 \& accept_fh($handle);
721 .Ve
722 .PP
723 In the examples above, we assigned the filehandle to a scalar variable
724 before using it. That is because only simple scalar variables, not
725 expressions or subscripts of hashes or arrays, can be used with
726 built-ins like \f(CW\*(C`print\*(C'\fR, \f(CW\*(C`printf\*(C'\fR, or the diamond operator. Using
727 something other than a simple scalar variable as a filehandle is
728 illegal and won't even compile:
729 .PP
730 .Vb 4
731 \& my @fd = (*STDIN, *STDOUT, *STDERR);
732 \& print $fd[1] "Type it: "; # WRONG
733 \& my $got = <$fd[0]> # WRONG
734 \& print $fd[2] "What was that: $got"; # WRONG
735 .Ve
736 .PP
737 With \f(CW\*(C`print\*(C'\fR and \f(CW\*(C`printf\*(C'\fR, you get around this by using a block and
738 an expression where you would place the filehandle:
739 .PP
740 .Vb 3
741 \& print { $fd[1] } "funny stuff\en";
742 \& printf { $fd[1] } "Pity the poor %x.\en", 3_735_928_559;
743 \& # Pity the poor deadbeef.
744 .Ve
745 .PP
746 That block is a proper block like any other, so you can put more
747 complicated code there. This sends the message out to one of two places:
748 .PP
749 .Vb 3
750 \& my $ok = \-x "/bin/cat";
751 \& print { $ok ? $fd[1] : $fd[2] } "cat stat $ok\en";
752 \& print { $fd[ 1+ ($ok || 0) ] } "cat stat $ok\en";
753 .Ve
754 .PP
755 This approach of treating \f(CW\*(C`print\*(C'\fR and \f(CW\*(C`printf\*(C'\fR like object methods
756 calls doesn't work for the diamond operator. That's because it's a
757 real operator, not just a function with a comma-less argument. Assuming
758 you've been storing typeglobs in your structure as we did above, you
759 can use the built-in function named \f(CW\*(C`readline\*(C'\fR to read a record just
760 as \f(CW\*(C`<>\*(C'\fR does. Given the initialization shown above for \f(CW@fd\fR, this
761 would work, but only because \fBreadline()\fR requires a typeglob. It doesn't
762 work with objects or strings, which might be a bug we haven't fixed yet.
763 .PP
764 .Vb 1
765 \& $got = readline($fd[0]);
766 .Ve
767 .PP
768 Let it be noted that the flakiness of indirect filehandles is not
769 related to whether they're strings, typeglobs, objects, or anything else.
770 It's the syntax of the fundamental operators. Playing the object
771 game doesn't help you at all here.
772 .SS "How can I open a filehandle to a string?"
773 .IX Xref "string open IO::String filehandle"
774 .IX Subsection "How can I open a filehandle to a string?"
775 (contributed by Peter J. Holzer, hjp\[email protected])
776 .PP
777 Since Perl 5.8.0 a file handle referring to a string can be created by
778 calling open with a reference to that string instead of the filename.
779 This file handle can then be used to read from or write to the string:
780 .PP
781 .Vb 3
782 \& open(my $fh, \*(Aq>\*(Aq, \e$string) or die "Could not open string for writing";
783 \& print $fh "foo\en";
784 \& print $fh "bar\en"; # $string now contains "foo\enbar\en"
785 \&
786 \& open(my $fh, \*(Aq<\*(Aq, \e$string) or die "Could not open string for reading";
787 \& my $x = <$fh>; # $x now contains "foo\en"
788 .Ve
789 .PP
790 With older versions of Perl, the IO::String module provides similar
791 functionality.
792 .SS "How can I set up a footer format to be used with \fBwrite()\fP?"
793 .IX Xref "footer"
794 .IX Subsection "How can I set up a footer format to be used with write()?"
795 There's no builtin way to do this, but perlform has a couple of
796 techniques to make it possible for the intrepid hacker.
797 .SS "How can I \fBwrite()\fP into a string?"
798 .IX Xref "write, into a string"
799 .IX Subsection "How can I write() into a string?"
800 (contributed by brian d foy)
801 .PP
802 If you want to \f(CW\*(C`write\*(C'\fR into a string, you just have to <open> a
803 filehandle to a string, which Perl has been able to do since Perl 5.6:
804 .PP
805 .Vb 2
806 \& open FH, \*(Aq>\*(Aq, \emy $string;
807 \& write( FH );
808 .Ve
809 .PP
810 Since you want to be a good programmer, you probably want to use a lexical
811 filehandle, even though formats are designed to work with bareword filehandles
812 since the default format names take the filehandle name. However, you can
813 control this with some Perl special per-filehandle variables: \f(CW$^\fR, which
814 names the top-of-page format, and \f(CW$~\fR which shows the line format. You have
815 to change the default filehandle to set these variables:
816 .PP
817 .Vb 1
818 \& open my($fh), \*(Aq>\*(Aq, \emy $string;
819 \&
820 \& { # set per\-filehandle variables
821 \& my $old_fh = select( $fh );
822 \& $~ = \*(AqANIMAL\*(Aq;
823 \& $^ = \*(AqANIMAL_TOP\*(Aq;
824 \& select( $old_fh );
825 \& }
826 \&
827 \& format ANIMAL_TOP =
828 \& ID Type Name
829 \& .
830 \&
831 \& format ANIMAL =
832 \& @## @<<< @<<<<<<<<<<<<<<
833 \& $id, $type, $name
834 \& .
835 .Ve
836 .PP
837 Although write can work with lexical or package variables, whatever variables
838 you use have to scope in the format. That most likely means you'll want to
839 localize some package variables:
840 .PP
841 .Vb 4
842 \& {
843 \& local( $id, $type, $name ) = qw( 12 cat Buster );
844 \& write( $fh );
845 \& }
846 \&
847 \& print $string;
848 .Ve
849 .PP
850 There are also some tricks that you can play with \f(CW\*(C`formline\*(C'\fR and the
851 accumulator variable \f(CW$^A\fR, but you lose a lot of the value of formats
852 since \f(CW\*(C`formline\*(C'\fR won't handle paging and so on. You end up reimplementing
853 formats when you use them.
854 .SS "How can I output my numbers with commas added?"
855 .IX Xref "number, commify"
856 .IX Subsection "How can I output my numbers with commas added?"
857 (contributed by brian d foy and Benjamin Goldberg)
858 .PP
859 The CLDR::Number module (the Unicode Common Locale Data Repository)
860 knows how to format numbers based on a locale to use its
861 digit grouping and separator:
862 .PP
863 .Vb 1
864 \& use CLDR::Number;
865 \&
866 \& my $number = 1234567890;
867 \&
868 \& foreach my $locale ( qw(en\-US de\-DE en\-IN) ) {
869 \& my $cldr = CLDR::Number\->new( locale => $locale );
870 \& my $decimal = $cldr\->decimal_formatter;
871 \& printf "%\-5s %s\en", $locale, $decimal\->format( $number );
872 \& }
873 .Ve
874 .PP
875 The output shows the same number formatted according to each locale's
876 rules. The bare \f(CW\*(C`en\*(C'\fR uses commas between groups of three, the \f(CW\*(C`de\*(C'\fR
877 uses dots and groups of three, and the \f(CW\*(C`en\-IN\*(C'\fR uses the 3:2:2 rule with
878 commas:
879 .PP
880 .Vb 3
881 \& en\-US 1,234,567,890
882 \& de\-DE 1.234.567.890
883 \& en\-IN 1,23,45,67,890
884 .Ve
885 .PP
886 CLDR::Number has many more advanced features to handle just about
887 anything that you might want. Olaf Alders' article "Stop Writing Your Own
888 Commify Functions" has more examples
889 ("/www.olafalders.com/2015/09/04/stop\-writing\-your\-own\-commify\-
890 functions/" in https:).
891 .PP
892 If you know exactly what you want, a simple subroutine might suffice.
893 This example will add separators to your number between groups of three
894 digits (or whatever grouping). This uses groups of three and a comma:
895 .PP
896 .Vb 5
897 \& sub commify {
898 \& local $_ = shift;
899 \& 1 while s/^([\-+]?\ed+)(\ed{3})/$1,$2/;
900 \& return $_;
901 \& }
902 .Ve
903 .PP
904 This regex from Benjamin Goldberg does the same thing without the loop:
905 .PP
906 .Vb 1
907 \& s/(^[\-+]?\ed+?(?=(?>(?:\ed{3})+)(?!\ed))|\eG\ed{3}(?=\ed))/$1,/g;
908 .Ve
909 .PP
910 It is easier to see with comments:
911 .PP
912 .Vb 11
913 \& s/(
914 \& ^[\-+]? # beginning of number.
915 \& \ed+? # first digits before first comma
916 \& (?= # followed by, (but not included in the match) :
917 \& (?>(?:\ed{3})+) # some positive multiple of three digits.
918 \& (?!\ed) # an *exact* multiple, not x * 3 + 1 or whatever.
919 \& )
920 \& | # or:
921 \& \eG\ed{3} # after the last group, get three digits
922 \& (?=\ed) # but they have to have more digits after them.
923 \& )/$1,/xg;
924 .Ve
925 .SS "How can I translate tildes (~) in a filename?"
926 .IX Xref "tilde tilde expansion"
927 .IX Subsection "How can I translate tildes (~) in a filename?"
928 Use the <> (\f(CWglob()\fR) operator, documented in perlfunc.
929 Versions of Perl older than 5.6 require that you have a shell
930 installed that groks tildes. Later versions of Perl have this feature
931 built in. The File::KGlob module (available from CPAN) gives more
932 portable glob functionality.
933 .PP
934 Within Perl, you may use this directly:
935 .PP
936 .Vb 11
937 \& $filename =~ s{
938 \& ^ ~ # find a leading tilde
939 \& ( # save this in $1
940 \& [^/] # a non\-slash character
941 \& * # repeated 0 or more times (0 means me)
942 \& )
943 \& }{
944 \& $1
945 \& ? (getpwnam($1))[7]
946 \& : ( $ENV{HOME} || $ENV{LOGDIR} )
947 \& }ex;
948 .Ve
949 .SS "When I open a file read-write, why does it wipe it out?"
950 .IX Xref "clobber read-write clobbering truncate truncating"
951 .IX Subsection "When I open a file read-write, why does it wipe it out?"
952 Because you're using something like this, which truncates the file
953 \&\fIthen\fR gives you read-write access:
954 .PP
955 .Vb 1
956 \& open my $fh, \*(Aq+>\*(Aq, \*(Aq/path/name\*(Aq; # WRONG (almost always)
957 .Ve
958 .PP
959 Whoops. You should instead use this, which will fail if the file
960 doesn't exist:
961 .PP
962 .Vb 1
963 \& open my $fh, \*(Aq+<\*(Aq, \*(Aq/path/name\*(Aq; # open for update
964 .Ve
965 .PP
966 Using ">" always clobbers or creates. Using "<" never does
967 either. The "+" doesn't change this.
968 .PP
969 Here are examples of many kinds of file opens. Those using \f(CW\*(C`sysopen\*(C'\fR
970 all assume that you've pulled in the constants from Fcntl:
971 .PP
972 .Vb 1
973 \& use Fcntl;
974 .Ve
975 .PP
976 To open file for reading:
977 .PP
978 .Vb 2
979 \& open my $fh, \*(Aq<\*(Aq, $path or die $!;
980 \& sysopen my $fh, $path, O_RDONLY or die $!;
981 .Ve
982 .PP
983 To open file for writing, create new file if needed or else truncate old file:
984 .PP
985 .Vb 3
986 \& open my $fh, \*(Aq>\*(Aq, $path or die $!;
987 \& sysopen my $fh, $path, O_WRONLY|O_TRUNC|O_CREAT or die $!;
988 \& sysopen my $fh, $path, O_WRONLY|O_TRUNC|O_CREAT, 0666 or die $!;
989 .Ve
990 .PP
991 To open file for writing, create new file, file must not exist:
992 .PP
993 .Vb 2
994 \& sysopen my $fh, $path, O_WRONLY|O_EXCL|O_CREAT or die $!;
995 \& sysopen my $fh, $path, O_WRONLY|O_EXCL|O_CREAT, 0666 or die $!;
996 .Ve
997 .PP
998 To open file for appending, create if necessary:
999 .PP
1000 .Vb 3
1001 \& open my $fh, \*(Aq>>\*(Aq, $path or die $!;
1002 \& sysopen my $fh, $path, O_WRONLY|O_APPEND|O_CREAT or die $!;
1003 \& sysopen my $fh, $path, O_WRONLY|O_APPEND|O_CREAT, 0666 or die $!;
1004 .Ve
1005 .PP
1006 To open file for appending, file must exist:
1007 .PP
1008 .Vb 1
1009 \& sysopen my $fh, $path, O_WRONLY|O_APPEND or die $!;
1010 .Ve
1011 .PP
1012 To open file for update, file must exist:
1013 .PP
1014 .Vb 2
1015 \& open my $fh, \*(Aq+<\*(Aq, $path or die $!;
1016 \& sysopen my $fh, $path, O_RDWR or die $!;
1017 .Ve
1018 .PP
1019 To open file for update, create file if necessary:
1020 .PP
1021 .Vb 2
1022 \& sysopen my $fh, $path, O_RDWR|O_CREAT or die $!;
1023 \& sysopen my $fh, $path, O_RDWR|O_CREAT, 0666 or die $!;
1024 .Ve
1025 .PP
1026 To open file for update, file must not exist:
1027 .PP
1028 .Vb 2
1029 \& sysopen my $fh, $path, O_RDWR|O_EXCL|O_CREAT or die $!;
1030 \& sysopen my $fh, $path, O_RDWR|O_EXCL|O_CREAT, 0666 or die $!;
1031 .Ve
1032 .PP
1033 To open a file without blocking, creating if necessary:
1034 .PP
1035 .Vb 2
1036 \& sysopen my $fh, \*(Aq/foo/somefile\*(Aq, O_WRONLY|O_NDELAY|O_CREAT
1037 \& or die "can\*(Aqt open /foo/somefile: $!":
1038 .Ve
1039 .PP
1040 Be warned that neither creation nor deletion of files is guaranteed to
1041 be an atomic operation over NFS. That is, two processes might both
1042 successfully create or unlink the same file! Therefore O_EXCL
1043 isn't as exclusive as you might wish.
1044 .PP
1045 See also perlopentut.
1046 .SS "Why do I sometimes get an ""Argument list too long"" when I use <*>?"
1047 .IX Xref "argument list too long"
1048 .IX Subsection "Why do I sometimes get an ""Argument list too long"" when I use <*>?"
1049 The \f(CW\*(C`<>\*(C'\fR operator performs a globbing operation (see above).
1050 In Perl versions earlier than v5.6.0, the internal \fBglob()\fR operator forks
1051 \&\fBcsh\fR\|(1) to do the actual glob expansion, but
1052 csh can't handle more than 127 items and so gives the error message
1053 \&\f(CW\*(C`Argument list too long\*(C'\fR. People who installed tcsh as csh won't
1054 have this problem, but their users may be surprised by it.
1055 .PP
1056 To get around this, either upgrade to Perl v5.6.0 or later, do the glob
1057 yourself with \fBreaddir()\fR and patterns, or use a module like File::Glob,
1058 one that doesn't use the shell to do globbing.
1059 .SS "How can I open a file named with a leading "">"" or trailing blanks?"
1060 .IX Xref "filename, special characters"
1061 .IX Subsection "How can I open a file named with a leading "">"" or trailing blanks?"
1062 (contributed by Brian McCauley)
1063 .PP
1064 The special two-argument form of Perl's \fBopen()\fR function ignores
1065 trailing blanks in filenames and infers the mode from certain leading
1066 characters (or a trailing "|"). In older versions of Perl this was the
1067 only version of \fBopen()\fR and so it is prevalent in old code and books.
1068 .PP
1069 Unless you have a particular reason to use the two-argument form you
1070 should use the three-argument form of \fBopen()\fR which does not treat any
1071 characters in the filename as special.
1072 .PP
1073 .Vb 2
1074 \& open my $fh, "<", " file "; # filename is " file "
1075 \& open my $fh, ">", ">file"; # filename is ">file"
1076 .Ve
1077 .SS "How can I reliably rename a file?"
1078 .IX Xref "rename mv move file, rename"
1079 .IX Subsection "How can I reliably rename a file?"
1080 If your operating system supports a proper \fBmv\fR\|(1) utility or its
1081 functional equivalent, this works:
1082 .PP
1083 .Vb 1
1084 \& rename($old, $new) or system("mv", $old, $new);
1085 .Ve
1086 .PP
1087 It may be more portable to use the File::Copy module instead.
1088 You just copy to the new file to the new name (checking return
1089 values), then delete the old one. This isn't really the same
1090 semantically as a \f(CWrename()\fR, which preserves meta-information like
1091 permissions, timestamps, inode info, etc.
1092 .SS "How can I lock a file?"
1093 .IX Xref "lock file, lock flock"
1094 .IX Subsection "How can I lock a file?"
1095 Perl's builtin \fBflock()\fR function (see perlfunc for details) will call
1096 \&\fBflock\fR\|(2) if that exists, \fBfcntl\fR\|(2) if it doesn't (on perl version 5.004 and
1097 later), and \fBlockf\fR\|(3) if neither of the two previous system calls exists.
1098 On some systems, it may even use a different form of native locking.
1099 Here are some gotchas with Perl's \fBflock()\fR:
1100 .IP 1. 4
1101 Produces a fatal error if none of the three system calls (or their
1102 close equivalent) exists.
1103 .IP 2. 4
1104 \&\fBlockf\fR\|(3) does not provide shared locking, and requires that the
1105 filehandle be open for writing (or appending, or read/writing).
1106 .IP 3. 4
1107 Some versions of \fBflock()\fR can't lock files over a network (e.g. on NFS file
1108 systems), so you'd need to force the use of \fBfcntl\fR\|(2) when you build Perl.
1109 But even this is dubious at best. See the flock entry of perlfunc
1110 and the \fIINSTALL\fR file in the source distribution for information on
1111 building Perl to do this.
1112 .Sp
1113 Two potentially non-obvious but traditional flock semantics are that
1114 it waits indefinitely until the lock is granted, and that its locks are
1115 \&\fImerely advisory\fR. Such discretionary locks are more flexible, but
1116 offer fewer guarantees. This means that files locked with \fBflock()\fR may
1117 be modified by programs that do not also use \fBflock()\fR. Cars that stop
1118 for red lights get on well with each other, but not with cars that don't
1119 stop for red lights. See the perlport manpage, your port's specific
1120 documentation, or your system-specific local manpages for details. It's
1121 best to assume traditional behavior if you're writing portable programs.
1122 (If you're not, you should as always feel perfectly free to write
1123 for your own system's idiosyncrasies (sometimes called "features").
1124 Slavish adherence to portability concerns shouldn't get in the way of
1125 your getting your job done.)
1126 .Sp
1127 For more information on file locking, see also
1128 "File Locking" in perlopentut if you have it (new for 5.6).
1129 .SS "Why can't I just open(FH, "">file.lock"")?"
1130 .IX Xref "lock, lockfile race condition"
1131 .IX Subsection "Why can't I just open(FH, "">file.lock"")?"
1132 A common bit of code \fBNOT TO USE\fR is this:
1133 .PP
1134 .Vb 2
1135 \& sleep(3) while \-e \*(Aqfile.lock\*(Aq; # PLEASE DO NOT USE
1136 \& open my $lock, \*(Aq>\*(Aq, \*(Aqfile.lock\*(Aq; # THIS BROKEN CODE
1137 .Ve
1138 .PP
1139 This is a classic race condition: you take two steps to do something
1140 which must be done in one. That's why computer hardware provides an
1141 atomic test-and-set instruction. In theory, this "ought" to work:
1142 .PP
1143 .Vb 2
1144 \& sysopen my $fh, "file.lock", O_WRONLY|O_EXCL|O_CREAT
1145 \& or die "can\*(Aqt open file.lock: $!";
1146 .Ve
1147 .PP
1148 except that lamentably, file creation (and deletion) is not atomic
1149 over NFS, so this won't work (at least, not every time) over the net.
1150 Various schemes involving \fBlink()\fR have been suggested, but
1151 these tend to involve busy-wait, which is also less than desirable.
1152 .SS "I still don't get locking. I just want to increment the number in the file. How can I do this?"
1153 .IX Xref "counter file, counter"
1154 .IX Subsection "I still don't get locking. I just want to increment the number in the file. How can I do this?"
1155 Didn't anyone ever tell you web-page hit counters were useless?
1156 They don't count number of hits, they're a waste of time, and they serve
1157 only to stroke the writer's vanity. It's better to pick a random number;
1158 they're more realistic.
1159 .PP
1160 Anyway, this is what you can do if you can't help yourself.
1161 .PP
1162 .Vb 8
1163 \& use Fcntl qw(:DEFAULT :flock);
1164 \& sysopen my $fh, "numfile", O_RDWR|O_CREAT or die "can\*(Aqt open numfile: $!";
1165 \& flock $fh, LOCK_EX or die "can\*(Aqt flock numfile: $!";
1166 \& my $num = <$fh> || 0;
1167 \& seek $fh, 0, 0 or die "can\*(Aqt rewind numfile: $!";
1168 \& truncate $fh, 0 or die "can\*(Aqt truncate numfile: $!";
1169 \& (print $fh $num+1, "\en") or die "can\*(Aqt write numfile: $!";
1170 \& close $fh or die "can\*(Aqt close numfile: $!";
1171 .Ve
1172 .PP
1173 Here's a much better web-page hit counter:
1174 .PP
1175 .Vb 1
1176 \& $hits = int( (time() \- 850_000_000) / rand(1_000) );
1177 .Ve
1178 .PP
1179 If the count doesn't impress your friends, then the code might. :\-)
1180 .SS "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?"
1181 .IX Xref "append file, append"
1182 .IX Subsection "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?"
1183 If you are on a system that correctly implements \f(CW\*(C`flock\*(C'\fR and you use
1184 the example appending code from "perldoc \-f flock" everything will be
1185 OK even if the OS you are on doesn't implement append mode correctly
1186 (if such a system exists). So if you are happy to restrict yourself to
1187 OSs that implement \f(CW\*(C`flock\*(C'\fR (and that's not really much of a
1188 restriction) then that is what you should do.
1189 .PP
1190 If you know you are only going to use a system that does correctly
1191 implement appending (i.e. not Win32) then you can omit the \f(CW\*(C`seek\*(C'\fR
1192 from the code in the previous answer.
1193 .PP
1194 If you know you are only writing code to run on an OS and filesystem
1195 that does implement append mode correctly (a local filesystem on a
1196 modern Unix for example), and you keep the file in block-buffered mode
1197 and you write less than one buffer-full of output between each manual
1198 flushing of the buffer then each bufferload is almost guaranteed to be
1199 written to the end of the file in one chunk without getting
1200 intermingled with anyone else's output. You can also use the
1201 \&\f(CW\*(C`syswrite\*(C'\fR function which is simply a wrapper around your system's
1202 \&\f(CWwrite(2)\fR system call.
1203 .PP
1204 There is still a small theoretical chance that a signal will interrupt
1205 the system-level \f(CWwrite()\fR operation before completion. There is also
1206 a possibility that some STDIO implementations may call multiple system
1207 level \f(CWwrite()\fRs even if the buffer was empty to start. There may be
1208 some systems where this probability is reduced to zero, and this is
1209 not a concern when using \f(CW\*(C`:perlio\*(C'\fR instead of your system's STDIO.
1210 .SS "How do I randomly update a binary file?"
1211 .IX Xref "file, binary patch"
1212 .IX Subsection "How do I randomly update a binary file?"
1213 If you're just trying to patch a binary, in many cases something as
1214 simple as this works:
1215 .PP
1216 .Vb 1
1217 \& perl \-i \-pe \*(Aqs{window manager}{window mangler}g\*(Aq /usr/bin/emacs
1218 .Ve
1219 .PP
1220 However, if you have fixed sized records, then you might do something more
1221 like this:
1222 .PP
1223 .Vb 9
1224 \& my $RECSIZE = 220; # size of record, in bytes
1225 \& my $recno = 37; # which record to update
1226 \& open my $fh, \*(Aq+<\*(Aq, \*(Aqsomewhere\*(Aq or die "can\*(Aqt update somewhere: $!";
1227 \& seek $fh, $recno * $RECSIZE, 0;
1228 \& read $fh, $record, $RECSIZE == $RECSIZE or die "can\*(Aqt read record $recno: $!";
1229 \& # munge the record
1230 \& seek $fh, \-$RECSIZE, 1;
1231 \& print $fh $record;
1232 \& close $fh;
1233 .Ve
1234 .PP
1235 Locking and error checking are left as an exercise for the reader.
1236 Don't forget them or you'll be quite sorry.
1237 .SS "How do I get a file's timestamp in perl?"
1238 .IX Xref "timestamp file, timestamp"
1239 .IX Subsection "How do I get a file's timestamp in perl?"
1240 If you want to retrieve the time at which the file was last read,
1241 written, or had its meta-data (owner, etc) changed, you use the \fB\-A\fR,
1242 \&\fB\-M\fR, or \fB\-C\fR file test operations as documented in perlfunc.
1243 These retrieve the age of the file (measured against the start-time of
1244 your program) in days as a floating point number. Some platforms may
1245 not have all of these times. See perlport for details. To retrieve
1246 the "raw" time in seconds since the epoch, you would call the stat
1247 function, then use \f(CWlocaltime()\fR, \f(CWgmtime()\fR, or
1248 \&\f(CWPOSIX::strftime()\fR to convert this into human-readable form.
1249 .PP
1250 Here's an example:
1251 .PP
1252 .Vb 3
1253 \& my $write_secs = (stat($file))[9];
1254 \& printf "file %s updated at %s\en", $file,
1255 \& scalar localtime($write_secs);
1256 .Ve
1257 .PP
1258 If you prefer something more legible, use the File::stat module
1259 (part of the standard distribution in version 5.004 and later):
1260 .PP
1261 .Vb 5
1262 \& # error checking left as an exercise for reader.
1263 \& use File::stat;
1264 \& use Time::localtime;
1265 \& my $date_string = ctime(stat($file)\->mtime);
1266 \& print "file $file updated at $date_string\en";
1267 .Ve
1268 .PP
1269 The \fBPOSIX::strftime()\fR approach has the benefit of being,
1270 in theory, independent of the current locale. See perllocale
1271 for details.
1272 .SS "How do I set a file's timestamp in perl?"
1273 .IX Xref "timestamp file, timestamp"
1274 .IX Subsection "How do I set a file's timestamp in perl?"
1275 You use the \fButime()\fR function documented in "utime" in perlfunc.
1276 By way of example, here's a little program that copies the
1277 read and write times from its first argument to all the rest
1278 of them.
1279 .PP
1280 .Vb 6
1281 \& if (@ARGV < 2) {
1282 \& die "usage: cptimes timestamp_file other_files ...\en";
1283 \& }
1284 \& my $timestamp = shift;
1285 \& my($atime, $mtime) = (stat($timestamp))[8,9];
1286 \& utime $atime, $mtime, @ARGV;
1287 .Ve
1288 .PP
1289 Error checking is, as usual, left as an exercise for the reader.
1290 .PP
1291 The perldoc for utime also has an example that has the same
1292 effect as \fBtouch\fR\|(1) on files that \fIalready exist\fR.
1293 .PP
1294 Certain file systems have a limited ability to store the times
1295 on a file at the expected level of precision. For example, the
1296 FAT and HPFS filesystem are unable to create dates on files with
1297 a finer granularity than two seconds. This is a limitation of
1298 the filesystems, not of \fButime()\fR.
1299 .SS "How do I print to more than one file at once?"
1300 .IX Xref "print, to multiple files"
1301 .IX Subsection "How do I print to more than one file at once?"
1302 To connect one filehandle to several output filehandles,
1303 you can use the IO::Tee or Tie::FileHandle::Multiplex modules.
1304 .PP
1305 If you only have to do this once, you can print individually
1306 to each filehandle.
1307 .PP
1308 .Vb 1
1309 \& for my $fh ($fh1, $fh2, $fh3) { print $fh "whatever\en" }
1310 .Ve
1311 .SS "How can I read in an entire file all at once?"
1312 .IX Xref "slurp file, slurping"
1313 .IX Subsection "How can I read in an entire file all at once?"
1314 The customary Perl approach for processing all the lines in a file is to
1315 do so one line at a time:
1316 .PP
1317 .Vb 6
1318 \& open my $input, \*(Aq<\*(Aq, $file or die "can\*(Aqt open $file: $!";
1319 \& while (<$input>) {
1320 \& chomp;
1321 \& # do something with $_
1322 \& }
1323 \& close $input or die "can\*(Aqt close $file: $!";
1324 .Ve
1325 .PP
1326 This is tremendously more efficient than reading the entire file into
1327 memory as an array of lines and then processing it one element at a time,
1328 which is often\-\-if not almost always\-\-the wrong approach. Whenever
1329 you see someone do this:
1330 .PP
1331 .Vb 1
1332 \& my @lines = <INPUT>;
1333 .Ve
1334 .PP
1335 You should think long and hard about why you need everything loaded at
1336 once. It's just not a scalable solution.
1337 .PP
1338 If you "mmap" the file with the File::Map module from
1339 CPAN, you can virtually load the entire file into a
1340 string without actually storing it in memory:
1341 .PP
1342 .Vb 1
1343 \& use File::Map qw(map_file);
1344 \&
1345 \& map_file my $string, $filename;
1346 .Ve
1347 .PP
1348 Once mapped, you can treat \f(CW$string\fR as you would any other string.
1349 Since you don't necessarily have to load the data, mmap-ing can be
1350 very fast and may not increase your memory footprint.
1351 .PP
1352 You might also find it more
1353 fun to use the standard Tie::File module, or the DB_File module's
1354 \&\f(CW$DB_RECNO\fR bindings, which allow you to tie an array to a file so that
1355 accessing an element of the array actually accesses the corresponding
1356 line in the file.
1357 .PP
1358 If you want to load the entire file, you can use the Path::Tiny
1359 module to do it in one simple and efficient step:
1360 .PP
1361 .Vb 1
1362 \& use Path::Tiny;
1363 \&
1364 \& my $all_of_it = path($filename)\->slurp; # entire file in scalar
1365 \& my @all_lines = path($filename)\->lines; # one line per element
1366 .Ve
1367 .PP
1368 Or you can read the entire file contents into a scalar like this:
1369 .PP
1370 .Vb 6
1371 \& my $var;
1372 \& {
1373 \& local $/;
1374 \& open my $fh, \*(Aq<\*(Aq, $file or die "can\*(Aqt open $file: $!";
1375 \& $var = <$fh>;
1376 \& }
1377 .Ve
1378 .PP
1379 That temporarily undefs your record separator, and will automatically
1380 close the file at block exit. If the file is already open, just use this:
1381 .PP
1382 .Vb 1
1383 \& my $var = do { local $/; <$fh> };
1384 .Ve
1385 .PP
1386 You can also use a localized \f(CW@ARGV\fR to eliminate the \f(CW\*(C`open\*(C'\fR:
1387 .PP
1388 .Vb 1
1389 \& my $var = do { local( @ARGV, $/ ) = $file; <> };
1390 .Ve
1391 .SS "How can I read in a file by paragraphs?"
1392 .IX Xref "file, reading by paragraphs"
1393 .IX Subsection "How can I read in a file by paragraphs?"
1394 Use the \f(CW$/\fR variable (see perlvar for details). You can either
1395 set it to \f(CW""\fR to eliminate empty paragraphs (\f(CW"abc\en\en\en\endef"\fR,
1396 for instance, gets treated as two paragraphs and not three), or
1397 \&\f(CW"\en\en"\fR to accept empty paragraphs.
1398 .PP
1399 Note that a blank line must have no blanks in it. Thus
1400 \&\f(CW"fred\en\ \enstuff\en\en"\fR is one paragraph, but \f(CW"fred\en\enstuff\en\en"\fR is two.
1401 .SS "How can I read a single character from a file? From the keyboard?"
1402 .IX Xref "getc file, reading one character at a time"
1403 .IX Subsection "How can I read a single character from a file? From the keyboard?"
1404 You can use the builtin \f(CWgetc()\fR function for most filehandles, but
1405 it won't (easily) work on a terminal device. For STDIN, either use
1406 the Term::ReadKey module from CPAN or use the sample code in
1407 "getc" in perlfunc.
1408 .PP
1409 If your system supports the portable operating system programming
1410 interface (POSIX), you can use the following code, which you'll note
1411 turns off echo processing as well.
1412 .PP
1413 .Vb 9
1414 \& #!/usr/bin/perl \-w
1415 \& use strict;
1416 \& $| = 1;
1417 \& for (1..4) {
1418 \& print "gimme: ";
1419 \& my $got = getone();
1420 \& print "\-\-> $got\en";
1421 \& }
1422 \& exit;
1423 \&
1424 \& BEGIN {
1425 \& use POSIX qw(:termios_h);
1426 \&
1427 \& my ($term, $oterm, $echo, $noecho, $fd_stdin);
1428 \&
1429 \& my $fd_stdin = fileno(STDIN);
1430 \&
1431 \& $term = POSIX::Termios\->new();
1432 \& $term\->getattr($fd_stdin);
1433 \& $oterm = $term\->getlflag();
1434 \&
1435 \& $echo = ECHO | ECHOK | ICANON;
1436 \& $noecho = $oterm & ~$echo;
1437 \&
1438 \& sub cbreak {
1439 \& $term\->setlflag($noecho);
1440 \& $term\->setcc(VTIME, 1);
1441 \& $term\->setattr($fd_stdin, TCSANOW);
1442 \& }
1443 \&
1444 \& sub cooked {
1445 \& $term\->setlflag($oterm);
1446 \& $term\->setcc(VTIME, 0);
1447 \& $term\->setattr($fd_stdin, TCSANOW);
1448 \& }
1449 \&
1450 \& sub getone {
1451 \& my $key = \*(Aq\*(Aq;
1452 \& cbreak();
1453 \& sysread(STDIN, $key, 1);
1454 \& cooked();
1455 \& return $key;
1456 \& }
1457 \& }
1458 \&
1459 \& END { cooked() }
1460 .Ve
1461 .PP
1462 The Term::ReadKey module from CPAN may be easier to use. Recent versions
1463 include also support for non-portable systems as well.
1464 .PP
1465 .Vb 8
1466 \& use Term::ReadKey;
1467 \& open my $tty, \*(Aq<\*(Aq, \*(Aq/dev/tty\*(Aq;
1468 \& print "Gimme a char: ";
1469 \& ReadMode "raw";
1470 \& my $key = ReadKey 0, $tty;
1471 \& ReadMode "normal";
1472 \& printf "\enYou said %s, char number %03d\en",
1473 \& $key, ord $key;
1474 .Ve
1475 .SS "How can I tell whether there's a character waiting on a filehandle?"
1476 .IX Subsection "How can I tell whether there's a character waiting on a filehandle?"
1477 The very first thing you should do is look into getting the Term::ReadKey
1478 extension from CPAN. As we mentioned earlier, it now even has limited
1479 support for non-portable (read: not open systems, closed, proprietary,
1480 not POSIX, not Unix, etc.) systems.
1481 .PP
1482 You should also check out the Frequently Asked Questions list in
1483 comp.unix.* for things like this: the answer is essentially the same.
1484 It's very system-dependent. Here's one solution that works on BSD
1485 systems:
1486 .PP
1487 .Vb 5
1488 \& sub key_ready {
1489 \& my($rin, $nfd);
1490 \& vec($rin, fileno(STDIN), 1) = 1;
1491 \& return $nfd = select($rin,undef,undef,0);
1492 \& }
1493 .Ve
1494 .PP
1495 If you want to find out how many characters are waiting, there's
1496 also the FIONREAD ioctl call to be looked at. The \fIh2ph\fR tool that
1497 comes with Perl tries to convert C include files to Perl code, which
1498 can be \f(CW\*(C`require\*(C'\fRd. FIONREAD ends up defined as a function in the
1499 \&\fIsys/ioctl.ph\fR file:
1500 .PP
1501 .Vb 1
1502 \& require \*(Aq./sys/ioctl.ph\*(Aq;
1503 \&
1504 \& $size = pack("L", 0);
1505 \& ioctl(FH, FIONREAD(), $size) or die "Couldn\*(Aqt call ioctl: $!\en";
1506 \& $size = unpack("L", $size);
1507 .Ve
1508 .PP
1509 If \fIh2ph\fR wasn't installed or doesn't work for you, you can
1510 \&\fIgrep\fR the include files by hand:
1511 .PP
1512 .Vb 2
1513 \& % grep FIONREAD /usr/include/*/*
1514 \& /usr/include/asm/ioctls.h:#define FIONREAD 0x541B
1515 .Ve
1516 .PP
1517 Or write a small C program using the editor of champions:
1518 .PP
1519 .Vb 9
1520 \& % cat > fionread.c
1521 \& #include <sys/ioctl.h>
1522 \& main() {
1523 \& printf("%#08x\en", FIONREAD);
1524 \& }
1525 \& ^D
1526 \& % cc \-o fionread fionread.c
1527 \& % ./fionread
1528 \& 0x4004667f
1529 .Ve
1530 .PP
1531 And then hard-code it, leaving porting as an exercise to your successor.
1532 .PP
1533 .Vb 1
1534 \& $FIONREAD = 0x4004667f; # XXX: opsys dependent
1535 \&
1536 \& $size = pack("L", 0);
1537 \& ioctl(FH, $FIONREAD, $size) or die "Couldn\*(Aqt call ioctl: $!\en";
1538 \& $size = unpack("L", $size);
1539 .Ve
1540 .PP
1541 FIONREAD requires a filehandle connected to a stream, meaning that sockets,
1542 pipes, and tty devices work, but \fInot\fR files.
1543 .ie n .SS "How do I do a ""tail \-f"" in perl?"
1544 .el .SS "How do I do a \f(CWtail \-f\fP in perl?"
1545 .IX Xref "tail IO::Handle File::Tail clearerr"
1546 .IX Subsection "How do I do a tail -f in perl?"
1547 First try
1548 .PP
1549 .Vb 1
1550 \& seek($gw_fh, 0, 1);
1551 .Ve
1552 .PP
1553 The statement \f(CW\*(C`seek($gw_fh, 0, 1)\*(C'\fR doesn't change the current position,
1554 but it does clear the end-of-file condition on the handle, so that the
1555 next \f(CW\*(C`<$gw_fh>\*(C'\fR makes Perl try again to read something.
1556 .PP
1557 If that doesn't work (it relies on features of your stdio implementation),
1558 then you need something more like this:
1559 .PP
1560 .Vb 7
1561 \& for (;;) {
1562 \& for ($curpos = tell($gw_fh); <$gw_fh>; $curpos =tell($gw_fh)) {
1563 \& # search for some stuff and put it into files
1564 \& }
1565 \& # sleep for a while
1566 \& seek($gw_fh, $curpos, 0); # seek to where we had been
1567 \& }
1568 .Ve
1569 .PP
1570 If this still doesn't work, look into the \f(CW\*(C`clearerr\*(C'\fR method
1571 from IO::Handle, which resets the error and end-of-file states
1572 on the handle.
1573 .PP
1574 There's also a File::Tail module from CPAN.
1575 .SS "How do I \fBdup()\fP a filehandle in Perl?"
1576 .IX Xref "dup"
1577 .IX Subsection "How do I dup() a filehandle in Perl?"
1578 If you check "open" in perlfunc, you'll see that several of the ways
1579 to call \fBopen()\fR should do the trick. For example:
1580 .PP
1581 .Vb 2
1582 \& open my $log, \*(Aq>>\*(Aq, \*(Aq/foo/logfile\*(Aq;
1583 \& open STDERR, \*(Aq>&\*(Aq, $log;
1584 .Ve
1585 .PP
1586 Or even with a literal numeric descriptor:
1587 .PP
1588 .Vb 2
1589 \& my $fd = $ENV{MHCONTEXTFD};
1590 \& open $mhcontext, "<&=$fd"; # like fdopen(3S)
1591 .Ve
1592 .PP
1593 Note that "<&STDIN" makes a copy, but "<&=STDIN" makes
1594 an alias. That means if you close an aliased handle, all
1595 aliases become inaccessible. This is not true with
1596 a copied one.
1597 .PP
1598 Error checking, as always, has been left as an exercise for the reader.
1599 .SS "How do I close a file descriptor by number?"
1600 .IX Xref "file, closing file descriptors POSIX close"
1601 .IX Subsection "How do I close a file descriptor by number?"
1602 If, for some reason, you have a file descriptor instead of a
1603 filehandle (perhaps you used \f(CW\*(C`POSIX::open\*(C'\fR), you can use the
1604 \&\f(CWclose()\fR function from the POSIX module:
1605 .PP
1606 .Vb 1
1607 \& use POSIX ();
1608 \&
1609 \& POSIX::close( $fd );
1610 .Ve
1611 .PP
1612 This should rarely be necessary, as the Perl \f(CWclose()\fR function is to be
1613 used for things that Perl opened itself, even if it was a dup of a
1614 numeric descriptor as with \f(CW\*(C`MHCONTEXT\*(C'\fR above. But if you really have
1615 to, you may be able to do this:
1616 .PP
1617 .Vb 3
1618 \& require \*(Aq./sys/syscall.ph\*(Aq;
1619 \& my $rc = syscall(SYS_close(), $fd + 0); # must force numeric
1620 \& die "can\*(Aqt sysclose $fd: $!" unless $rc == \-1;
1621 .Ve
1622 .PP
1623 Or, just use the fdopen(3S) feature of \f(CWopen()\fR:
1624 .PP
1625 .Vb 4
1626 \& {
1627 \& open my $fh, "<&=$fd" or die "Cannot reopen fd=$fd: $!";
1628 \& close $fh;
1629 \& }
1630 .Ve
1631 .SS "Why can't I use ""C:\etemp\efoo"" in DOS paths? Why doesn't `C:\etemp\efoo.exe` work?"
1632 .IX Xref "filename, DOS issues"
1633 .IX Subsection "Why can't I use ""C:tempfoo"" in DOS paths? Why doesn't `C:tempfoo.exe` work?"
1634 Whoops! You just put a tab and a formfeed into that filename!
1635 Remember that within double quoted strings ("like\ethis"), the
1636 backslash is an escape character. The full list of these is in
1637 "Quote and Quote-like Operators" in perlop. Unsurprisingly, you don't
1638 have a file called "c:(tab)emp(formfeed)oo" or
1639 "c:(tab)emp(formfeed)oo.exe" on your legacy DOS filesystem.
1640 .PP
1641 Either single-quote your strings, or (preferably) use forward slashes.
1642 Since all DOS and Windows versions since something like MS-DOS 2.0 or so
1643 have treated \f(CW\*(C`/\*(C'\fR and \f(CW\*(C`\e\*(C'\fR the same in a path, you might as well use the
1644 one that doesn't clash with Perl\-\-or the POSIX shell, ANSI C and C++,
1645 awk, Tcl, Java, or Python, just to mention a few. POSIX paths
1646 are more portable, too.
1647 .SS "Why doesn't glob(""*.*"") get all the files?"
1648 .IX Xref "glob"
1649 .IX Subsection "Why doesn't glob(""*.*"") get all the files?"
1650 Because even on non-Unix ports, Perl's glob function follows standard
1651 Unix globbing semantics. You'll need \f(CWglob("*")\fR to get all (non-hidden)
1652 files. This makes \fBglob()\fR portable even to legacy systems. Your
1653 port may include proprietary globbing functions as well. Check its
1654 documentation for details.
1655 .ie n .SS "Why does Perl let me delete read-only files? Why does ""\-i"" clobber protected files? Isn't this a bug in Perl?"
1656 .el .SS "Why does Perl let me delete read-only files? Why does \f(CW\-i\fP clobber protected files? Isn't this a bug in Perl?"
1657 .IX Subsection "Why does Perl let me delete read-only files? Why does -i clobber protected files? Isn't this a bug in Perl?"
1658 This is elaborately and painstakingly described in the
1659 \&\fIfile-dir-perms\fR article in the "Far More Than You Ever Wanted To
1660 Know" collection in <http://www.cpan.org/misc/olddoc/FMTEYEWTK.tgz> .
1661 .PP
1662 The executive summary: learn how your filesystem works. The
1663 permissions on a file say what can happen to the data in that file.
1664 The permissions on a directory say what can happen to the list of
1665 files in that directory. If you delete a file, you're removing its
1666 name from the directory (so the operation depends on the permissions
1667 of the directory, not of the file). If you try to write to the file,
1668 the permissions of the file govern whether you're allowed to.
1669 .SS "How do I select a random line from a file?"
1670 .IX Xref "file, selecting a random line"
1671 .IX Subsection "How do I select a random line from a file?"
1672 Short of loading the file into a database or pre-indexing the lines in
1673 the file, there are a couple of things that you can do.
1674 .PP
1675 Here's a reservoir-sampling algorithm from the Camel Book:
1676 .PP
1677 .Vb 2
1678 \& srand;
1679 \& rand($.) < 1 && ($line = $_) while <>;
1680 .Ve
1681 .PP
1682 This has a significant advantage in space over reading the whole file
1683 in. You can find a proof of this method in \fIThe Art of Computer
1684 Programming\fR, Volume 2, Section 3.4.2, by Donald E. Knuth.
1685 .PP
1686 You can use the File::Random module which provides a function
1687 for that algorithm:
1688 .PP
1689 .Vb 2
1690 \& use File::Random qw/random_line/;
1691 \& my $line = random_line($filename);
1692 .Ve
1693 .PP
1694 Another way is to use the Tie::File module, which treats the entire
1695 file as an array. Simply access a random array element.
1696 .SS "Why do I get weird spaces when I print an array of lines?"
1697 .IX Subsection "Why do I get weird spaces when I print an array of lines?"
1698 (contributed by brian d foy)
1699 .PP
1700 If you are seeing spaces between the elements of your array when
1701 you print the array, you are probably interpolating the array in
1702 double quotes:
1703 .PP
1704 .Vb 2
1705 \& my @animals = qw(camel llama alpaca vicuna);
1706 \& print "animals are: @animals\en";
1707 .Ve
1708 .PP
1709 It's the double quotes, not the \f(CW\*(C`print\*(C'\fR, doing this. Whenever you
1710 interpolate an array in a double quote context, Perl joins the
1711 elements with spaces (or whatever is in \f(CW$"\fR, which is a space by
1712 default):
1713 .PP
1714 .Vb 1
1715 \& animals are: camel llama alpaca vicuna
1716 .Ve
1717 .PP
1718 This is different than printing the array without the interpolation:
1719 .PP
1720 .Vb 2
1721 \& my @animals = qw(camel llama alpaca vicuna);
1722 \& print "animals are: ", @animals, "\en";
1723 .Ve
1724 .PP
1725 Now the output doesn't have the spaces between the elements because
1726 the elements of \f(CW@animals\fR simply become part of the list to
1727 \&\f(CW\*(C`print\*(C'\fR:
1728 .PP
1729 .Vb 1
1730 \& animals are: camelllamaalpacavicuna
1731 .Ve
1732 .PP
1733 You might notice this when each of the elements of \f(CW@array\fR end with
1734 a newline. You expect to print one element per line, but notice that
1735 every line after the first is indented:
1736 .PP
1737 .Vb 3
1738 \& this is a line
1739 \& this is another line
1740 \& this is the third line
1741 .Ve
1742 .PP
1743 That extra space comes from the interpolation of the array. If you
1744 don't want to put anything between your array elements, don't use the
1745 array in double quotes. You can send it to print without them:
1746 .PP
1747 .Vb 1
1748 \& print @lines;
1749 .Ve
1750 .SS "How do I traverse a directory tree?"
1751 .IX Subsection "How do I traverse a directory tree?"
1752 (contributed by brian d foy)
1753 .PP
1754 The File::Find module, which comes with Perl, does all of the hard
1755 work to traverse a directory structure. You simply
1756 call the \f(CW\*(C`find\*(C'\fR subroutine with a callback subroutine and the
1757 directories you want to traverse:
1758 .PP
1759 .Vb 1
1760 \& use File::Find;
1761 \&
1762 \& find( \e&wanted, @directories );
1763 \&
1764 \& sub wanted {
1765 \& # full path in $File::Find::name
1766 \& # just filename in $_
1767 \& ... do whatever you want to do ...
1768 \& }
1769 .Ve
1770 .PP
1771 The File::Find::Closures, which you can download from CPAN, provides
1772 many ready-to-use subroutines that you can use with File::Find.
1773 .PP
1774 The File::Finder, which you can download from CPAN, can help you
1775 create the callback subroutine using something closer to the syntax of
1776 the \f(CW\*(C`find\*(C'\fR command-line utility:
1777 .PP
1778 .Vb 2
1779 \& use File::Find;
1780 \& use File::Finder;
1781 \&
1782 \& my $deep_dirs = File::Finder\->depth\->type(\*(Aqd\*(Aq)\->ls\->exec(\*(Aqrmdir\*(Aq,\*(Aq{}\*(Aq);
1783 \&
1784 \& find( $deep_dirs\->as_options, @places );
1785 .Ve
1786 .PP
1787 The File::Find::Rule module, which you can download from CPAN, has
1788 a similar interface, but does the traversal for you too:
1789 .PP
1790 .Vb 1
1791 \& use File::Find::Rule;
1792 \&
1793 \& my @files = File::Find::Rule\->file()
1794 \& \->name( \*(Aq*.pm\*(Aq )
1795 \& \->in( @INC );
1796 .Ve
1797 .SS "How do I delete a directory tree?"
1798 .IX Subsection "How do I delete a directory tree?"
1799 (contributed by brian d foy)
1800 .PP
1801 If you have an empty directory, you can use Perl's built-in \f(CW\*(C`rmdir\*(C'\fR.
1802 If the directory is not empty (so, with files or subdirectories), you
1803 either have to empty it yourself (a lot of work) or use a module to
1804 help you.
1805 .PP
1806 The File::Path module, which comes with Perl, has a \f(CW\*(C`remove_tree\*(C'\fR
1807 which can take care of all of the hard work for you:
1808 .PP
1809 .Vb 1
1810 \& use File::Path qw(remove_tree);
1811 \&
1812 \& remove_tree( @directories );
1813 .Ve
1814 .PP
1815 The File::Path module also has a legacy interface to the older
1816 \&\f(CW\*(C`rmtree\*(C'\fR subroutine.
1817 .SS "How do I copy an entire directory?"
1818 .IX Subsection "How do I copy an entire directory?"
1819 (contributed by Shlomi Fish)
1820 .PP
1821 To do the equivalent of \f(CW\*(C`cp \-R\*(C'\fR (i.e. copy an entire directory tree
1822 recursively) in portable Perl, you'll either need to write something yourself
1823 or find a good CPAN module such as File::Copy::Recursive.
1824 .SH "AUTHOR AND COPYRIGHT"
1825 .IX Header "AUTHOR AND COPYRIGHT"
1826 Copyright (c) 1997\-2010 Tom Christiansen, Nathan Torkington, and
1827 other authors as noted. All rights reserved.
1828 .PP
1829 This documentation is free; you can redistribute it and/or modify it
1830 under the same terms as Perl itself.
1831 .PP
1832 Irrespective of its distribution, all code examples here are in the public
1833 domain. You are permitted and encouraged to use this code and any
1834 derivatives thereof in your own programs for fun or for profit as you
1835 see fit. A simple comment in the code giving credit to the FAQ would
1836 be courteous but is not required.