This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
CommitLineData
2
d92eb7b0 3perlfaq4 - Data Manipulation (\$Revision: 1.49 \$, \$Date: 1999/05/23 20:37:49 \$)
68dc0745 4
6
a6dd486b 7The section of the FAQ answers questions related to the manipulation
68dc0745 8of data as numbers, dates, strings, arrays, hashes, and miscellaneous
9data issues.
10
12
46fc3d4c 13=head2 Why am I getting long decimals (eg, 19.9499999999999) instead of the numbers I should be getting (eg, 19.95)?
14
5a964f20 15The infinite set that a mathematician thinks of as the real numbers can
a6dd486b 16only be approximated on a computer, since the computer only has a finite
5a964f20
TC
17number of bits to store an infinite number of, um, numbers.
18
46fc3d4c 19Internally, your computer represents floating-point numbers in binary.
92c2ed05
GS
20Floating-point numbers read in from a file or appearing as literals
21in your program are converted from their decimal floating-point
a6dd486b 22representation (eg, 19.95) to an internal binary representation.
46fc3d4c 23
24However, 19.95 can't be precisely represented as a binary
25floating-point number, just like 1/3 can't be exactly represented as a
26decimal floating-point number. The computer's binary representation
27of 19.95, therefore, isn't exactly 19.95.
28
29When a floating-point number gets printed, the binary floating-point
30representation is converted back to decimal. These decimal numbers
31are displayed in either the format you specify with printf(), or the
a6dd486b 32current output format for numbers. (See L<perlvar/"\$#"> if you use
46fc3d4c 33print. C<\$#> has a different default value in Perl5 than it did in
87275199 34Perl4. Changing C<\$#> yourself is deprecated.)
46fc3d4c 35
36This affects B<all> computer languages that represent decimal
37floating-point numbers in binary, not just Perl. Perl provides
38arbitrary-precision decimal numbers with the Math::BigFloat module
39(part of the standard Perl distribution), but mathematical operations
40are consequently slower.
41
80ba158a 42If precision is important, such as when dealing with money, it's good
1affb2ee
MS
43to work with integers and then divide at the last possible moment.
44For example, work in pennies (1995) instead of dollars and cents
6b927632 45(19.95) and divide by 100 at the end.
1affb2ee 46
46fc3d4c 47To get rid of the superfluous digits, just use a format (eg,
48C<printf("%.2f", 19.95)>) to get the required precision.
65acb1b1 49See L<perlop/"Floating-point Arithmetic">.
46fc3d4c 50
68dc0745 51=head2 Why isn't my octal data interpreted correctly?
52
53Perl only understands octal and hex numbers as such when they occur
33ce146f
PP
56If they are read in from somewhere and assigned, no automatic
57conversion takes place. You must explicitly use oct() or hex() if you
58want the values converted to decimal. oct() interprets
68dc0745 59both hex ("0x350") numbers and octal ones ("0350" or even without the
61with or without a leading "0x", like "0x255", "3A", "ff", or "deadbeef".
33ce146f
PP
62The inverse mapping from decimal to octal can be done with either the
63"%o" or "%O" sprintf() formats. To get from decimal to hex try either
64the "%x" or the "%X" formats to sprintf().
68dc0745 65
66This problem shows up most often when people try using chmod(), mkdir(),
33ce146f
PP
68permissions in octal.
68dc0745 69
33ce146f 70 chmod(644, \$file); # WRONG
68dc0745 71 chmod(0644, \$file); # right
72
33ce146f
PP
73Note the mistake in the first line was specifying the decimal literal
74644, rather than the intended octal literal 0644. The problem can
75be seen with:
76
434f7166 77 printf("%#o",644); # prints 01204
33ce146f
PP
78
79Surely you had not intended C<chmod(01204, \$file);> - did you? If you
80want to use numeric literals as arguments to chmod() et al. then please
81try to express them as octal constants, that is with a leading zero and
82with the following digits restricted to the set 0..7.
83
65acb1b1 84=head2 Does Perl have a round() function? What about ceil() and floor()? Trig functions?
68dc0745 85
92c2ed05
GS
86Remember that int() merely truncates toward 0. For rounding to a
87certain number of digits, sprintf() or printf() is usually the easiest
88route.
89
90 printf("%.3f", 3.1415926535); # prints 3.142
68dc0745 91
87275199 92The POSIX module (part of the standard Perl distribution) implements
68dc0745 93ceil(), floor(), and a number of other mathematical and trigonometric
94functions.
95
92c2ed05
GS
96 use POSIX;
97 \$ceil = ceil(3.5); # 4
98 \$floor = floor(3.5); # 3
99
a6dd486b 100In 5.000 to 5.003 perls, trigonometry was done in the Math::Complex
87275199 101module. With 5.004, the Math::Trig module (part of the standard Perl
46fc3d4c 102distribution) implements the trigonometric functions. Internally it
103uses the Math::Complex module and some functions can break out from
104the real axis into the complex plane, for example the inverse sine of
1052.
68dc0745 106
107Rounding in financial applications can have serious implications, and
108the rounding method used should be specified precisely. In these
109cases, it probably pays not to trust whichever system rounding is
110being used by Perl, but to instead implement the rounding function you
111need yourself.
112
65acb1b1
TC
113To see why, notice how you'll still have an issue on half-way-point
114alternation:
115
116 for (\$i = 0; \$i < 1.01; \$i += 0.05) { printf "%.1f ",\$i}
117
118 0.0 0.1 0.1 0.2 0.2 0.2 0.3 0.3 0.4 0.4 0.5 0.5 0.6 0.7 0.7
119 0.8 0.8 0.9 0.9 1.0 1.0
120
121Don't blame Perl. It's the same as in C. IEEE says we have to do this.
122Perl numbers whose absolute values are integers under 2**31 (on 32 bit
123machines) will work pretty much like mathematical integers. Other numbers
124are not guaranteed.
125
68dc0745 126=head2 How do I convert bits into ints?
127
92c2ed05 128To turn a string of 1s and 0s like C<10110110> into a scalar containing
d92eb7b0 129its binary value, use the pack() and unpack() functions (documented in
87275199 130L<perlfunc/"pack"> and L<perlfunc/"unpack">):
68dc0745 131
d92eb7b0
GS
132 \$decimal = unpack('c', pack('B8', '10110110'));
133
134This packs the string C<10110110> into an eight bit binary structure.
87275199 135This is then unpacked as a character, which returns its ordinal value.
d92eb7b0
GS
136
137This does the same thing:
138
139 \$decimal = ord(pack('B8', '10110110'));
68dc0745 140
141Here's an example of going the other way:
142
d92eb7b0 143 \$binary_string = unpack('B*', "\x29");
68dc0745 144
65acb1b1
TC
145=head2 Why doesn't & work the way I want it to?
146
147The behavior of binary arithmetic operators depends on whether they're
148used on numbers or strings. The operators treat a string as a series
149of bits and work with that (the string C<"3"> is the bit pattern
150C<00110011>). The operators work with the binary form of a number
151(the number C<3> is treated as the bit pattern C<00000011>).
152
153So, saying C<11 & 3> performs the "and" operation on numbers (yielding
154C<1>). Saying C<"11" & "3"> performs the "and" operation on strings
155(yielding C<"1">).
156
157Most problems with C<&> and C<|> arise because the programmer thinks
158they have a number but really it's a string. The rest arise because
159the programmer says:
160
161 if ("\020\020" & "\101\101") {
162 # ...
163 }
164
165but a string consisting of two null bytes (the result of C<"\020\020"
166& "\101\101">) is not a false value in Perl. You need:
167
168 if ( ("\020\020" & "\101\101") !~ /[^\000]/) {
169 # ...
170 }
171
68dc0745 172=head2 How do I multiply matrices?
173
174Use the Math::Matrix or Math::MatrixReal modules (available from CPAN)
175or the PDL extension (also available from CPAN).
176
177=head2 How do I perform an operation on a series of integers?
178
179To call a function on each element in an array, and collect the
180results, use:
181
182 @results = map { my_func(\$_) } @array;
183
184For example:
185
186 @triple = map { 3 * \$_ } @single;
187
188To call a function on each element of an array, but ignore the
189results:
190
191 foreach \$iterator (@array) {
65acb1b1 192 some_func(\$iterator);
68dc0745 193 }
194
195To call a function on each integer in a (small) range, you B<can> use:
196
65acb1b1 197 @results = map { some_func(\$_) } (5 .. 25);
68dc0745 198
199but you should be aware that the C<..> operator creates an array of
200all integers in the range. This can take a lot of memory for large
202
203 @results = ();
204 for (\$i=5; \$i < 500_005; \$i++) {
65acb1b1 205 push(@results, some_func(\$i));
68dc0745 206 }
207
87275199
GS
208This situation has been fixed in Perl5.005. Use of C<..> in a C<for>
209loop will iterate over the range, without creating the entire range.
210
211 for my \$i (5 .. 500_005) {
212 push(@results, some_func(\$i));
213 }
214
215will not create a list of 500,000 integers.
216
68dc0745 217=head2 How can I output Roman numerals?
218
219Get the http://www.perl.com/CPAN/modules/by-module/Roman module.
220
221=head2 Why aren't my random numbers random?
222
65acb1b1
TC
223If you're using a version of Perl before 5.004, you must call C<srand>
224once at the start of your program to seed the random number generator.
2255.004 and later automatically call C<srand> at the beginning. Don't
226call C<srand> more than once--you make your numbers less random, rather
227than more.
92c2ed05 228
65acb1b1
TC
229Computers are good at being predictable and bad at being random
230(despite appearances caused by bugs in your programs :-).
a6dd486b
JB
231http://www.perl.com/CPAN/doc/FMTEYEWTK/random , courtesy of Tom
65acb1b1
TC
233attempts to generate random numbers by deterministic means is, of
234course, living in a state of sin.''
235
236If you want numbers that are more random than C<rand> with C<srand>
237provides, you should also check out the Math::TrulyRandom module from
238CPAN. It uses the imperfections in your system's timer to generate
239random numbers, but this takes quite a while. If you want a better
92c2ed05 240pseudorandom generator than comes with your operating system, look at
65acb1b1 241``Numerical Recipes in C'' at http://www.nr.com/ .
68dc0745 242
244
245=head2 How do I find the week-of-the-year/day-of-the-year?
246
247The day of the year is in the array returned by localtime() (see
248L<perlfunc/"localtime">):
249
250 \$day_of_year = (localtime(time()))[7];
251
d92eb7b0
GS
252=head2 How do I find the current century or millennium?
253
254Use the following simple functions:
255
256 sub get_century {
257 return int((((localtime(shift || time))[5] + 1999))/100);
258 }
259 sub get_millennium {
260 return 1+int((((localtime(shift || time))[5] + 1899))/1000);
261 }
262
263On some systems, you'll find that the POSIX module's strftime() function
264has been extended in a non-standard way to use a C<%C> format, which they
265sometimes claim is the "century". It isn't, because on most such systems,
266this is only the first two digits of the four-digit year, and thus cannot
267be used to reliably determine the current century or millennium.
268
92c2ed05 269=head2 How can I compare two dates and find the difference?
68dc0745 270
92c2ed05
GS
271If you're storing your dates as epoch seconds then simply subtract one
272from the other. If you've got a structured date (distinct year, day,
d92eb7b0
GS
273month, hour, minute, seconds values), then for reasons of accessibility,
274simplicity, and efficiency, merely use either timelocal or timegm (from
275the Time::Local module in the standard distribution) to reduce structured
276dates to epoch seconds. However, if you don't know the precise format of
277your dates, then you should probably use either of the Date::Manip and
278Date::Calc modules from CPAN before you go hacking up your own parsing
279routine to handle arbitrary date formats.
68dc0745 280
281=head2 How can I take a string and turn it into epoch seconds?
282
283If it's a regular enough string that it always has the same format,
92c2ed05
GS
284you can split it up and pass the parts to C<timelocal> in the standard
285Time::Local module. Otherwise, you should look into the Date::Calc
286and Date::Manip modules from CPAN.
68dc0745 287
288=head2 How can I find the Julian Day?
289
2a2bf5f4
JH
290Use the Time::JulianDay module (part of the Time-modules bundle
291available from CPAN.)
d92eb7b0 292
89435c96
MS
293Before you immerse yourself too deeply in this, be sure to verify that
294it is the I<Julian> Day you really want. Are you interested in a way
295of getting serial days so that you just can tell how many days they
296are apart or so that you can do also other date arithmetic? If you
d92eb7b0 297are interested in performing date arithmetic, this can be done using
2a2bf5f4 298modules Date::Manip or Date::Calc.
89435c96
MS
299
300There is too many details and much confusion on this issue to cover in
301this FAQ, but the term is applied (correctly) to a calendar now
302supplanted by the Gregorian Calendar, with the Julian Calendar failing
303to adjust properly for leap years on centennial years (among other
304annoyances). The term is also used (incorrectly) to mean: [1] days in
305the Gregorian Calendar; and [2] days since a particular starting time
306or `epoch', usually 1970 in the Unix world and 1980 in the
307MS-DOS/Windows world. If you find that it is not the first meaning
308that you really want, then check out the Date::Manip and Date::Calc
309modules. (Thanks to David Cassell for most of this text.)
be94a901 310
65acb1b1
TC
311=head2 How do I find yesterday's date?
312
313The C<time()> function returns the current time in seconds since the
d92eb7b0 314epoch. Take twenty-four hours off that:
65acb1b1
TC
315
316 \$yesterday = time() - ( 24 * 60 * 60 );
317
318Then you can pass this to C<localtime()> and get the individual year,
319month, day, hour, minute, seconds values.
320
d92eb7b0
GS
321Note very carefully that the code above assumes that your days are
322twenty-four hours each. For most people, there are two days a year
323when they aren't: the switch to and from summer time throws this off.
324A solution to this issue is offered by Russ Allbery.
325
326 sub yesterday {
327 my \$now = defined \$_[0] ? \$_[0] : time;
328 my \$then = \$now - 60 * 60 * 24;
329 my \$ndst = (localtime \$now)[8] > 0;
330 my \$tdst = (localtime \$then)[8] > 0;
331 \$then - (\$tdst - \$ndst) * 60 * 60;
332 }
333 # Should give you "this time yesterday" in seconds since epoch relative to
334 # the first argument or the current time if no argument is given and
335 # suitable for passing to localtime or whatever else you need to do with
336 # it. \$ndst is whether we're currently in daylight savings time; \$tdst is
337 # whether the point 24 hours ago was in daylight savings time. If \$tdst
338 # and \$ndst are the same, a boundary wasn't crossed, and the correction
339 # will subtract 0. If \$tdst is 1 and \$ndst is 0, subtract an hour more
340 # from yesterday's time since we gained an extra hour while going off
341 # daylight savings time. If \$tdst is 0 and \$ndst is 1, subtract a
342 # negative hour (add an hour) to yesterday's time since we lost an hour.
343 #
344 # All of this is because during those days when one switches off or onto
345 # DST, a "day" isn't 24 hours long; it's either 23 or 25.
346 #
347 # The explicit settings of \$ndst and \$tdst are necessary because localtime
348 # only says it returns the system tm struct, and the system tm struct at
87275199 349 # least on Solaris doesn't guarantee any particular positive value (like,
d92eb7b0
GS
350 # say, 1) for isdst, just a positive value. And that value can
351 # potentially be negative, if DST information isn't available (this sub
352 # just treats those cases like no DST).
353 #
354 # Note that between 2am and 3am on the day after the time zone switches
355 # off daylight savings time, the exact hour of "yesterday" corresponding
356 # to the current hour is not clearly defined. Note also that if used
357 # between 2am and 3am the day after the change to daylight savings time,
358 # the result will be between 3am and 4am of the previous day; it's
359 # arguable whether this is correct.
360 #
361 # This sub does not attempt to deal with leap seconds (most things don't).
362 #
363 # Copyright relinquished 1999 by Russ Allbery <rra@stanford.edu>
364 # This code is in the public domain
365
87275199 366=head2 Does Perl have a Year 2000 problem? Is Perl Y2K compliant?
68dc0745 367
65acb1b1
TC
368Short answer: No, Perl does not have a Year 2000 problem. Yes, Perl is
369Y2K compliant (whatever that means). The programmers you've hired to
370use it, however, probably are not.
371
372Long answer: The question belies a true understanding of the issue.
373Perl is just as Y2K compliant as your pencil--no more, and no less.
374Can you use your pencil to write a non-Y2K-compliant memo? Of course
375you can. Is that the pencil's fault? Of course it isn't.
92c2ed05 376
87275199 377The date and time functions supplied with Perl (gmtime and localtime)
65acb1b1
TC
378supply adequate information to determine the year well beyond 2000
379(2038 is when trouble strikes for 32-bit machines). The year returned
90fdbbb7 380by these functions when used in a list context is the year minus 1900.
65acb1b1
TC
381For years between 1910 and 1999 this I<happens> to be a 2-digit decimal
382number. To avoid the year 2000 problem simply do not treat the year as
383a 2-digit number. It isn't.
68dc0745 384
5a964f20 385When gmtime() and localtime() are used in scalar context they return
68dc0745 386a timestamp string that contains a fully-expanded year. For example,
387C<\$timestamp = gmtime(1005613200)> sets \$timestamp to "Tue Nov 13 01:00:00
3882001". There's no year 2000 problem here.
389
5a964f20
TC
390That doesn't mean that Perl can't be used to create non-Y2K compliant
391programs. It can. But so can your pencil. It's the fault of the user,
392not the language. At the risk of inflaming the NRA: ``Perl doesn't
393break Y2K, people do.'' See http://language.perl.com/news/y2k.html for
394a longer exposition.
395
397
398=head2 How do I validate input?
399
400The answer to this question is usually a regular expression, perhaps
5a964f20 401with auxiliary logic. See the more specific questions (numbers, mail
403
404=head2 How do I unescape a string?
405
92c2ed05
GS
406It depends just what you mean by ``escape''. URL escapes are dealt
407with in L<perlfaq9>. Shell escapes with the backslash (C<\>)
a6dd486b 408character are removed with
68dc0745 409
410 s/\\(.)/\$1/g;
411
92c2ed05 412This won't expand C<"\n"> or C<"\t"> or any other special escapes.
68dc0745 413
414=head2 How do I remove consecutive pairs of characters?
415
92c2ed05 416To turn C<"abbcccd"> into C<"abccd">:
68dc0745 417
d92eb7b0
GS
418 s/(.)\1/\$1/g; # add /s to include newlines
419
420Here's a solution that turns "abbcccd" to "abcd":
421
422 y///cs; # y == tr, but shorter :-)
68dc0745 423
424=head2 How do I expand function calls in a string?
425
426This is documented in L<perlref>. In general, this is fraught with
427quoting and readability problems, but it is possible. To interpolate
5a964f20 428a subroutine call (in list context) into a string:
68dc0745 429
430 print "My sub returned @{[mysub(1,2,3)]} that time.\n";
431
432If you prefer scalar context, similar chicanery is also useful for
433arbitrary expressions:
434
435 print "That yields \${\(\$n + 5)} widgets\n";
436
92c2ed05
GS
437Version 5.004 of Perl had a bug that gave list context to the
438expression in C<\${...}>, but this is fixed in version 5.005.
439
440See also ``How can I expand variables in text strings?'' in this
441section of the FAQ.
46fc3d4c 442
68dc0745 443=head2 How do I find matching/nesting anything?
444
92c2ed05
GS
445This isn't something that can be done in one regular expression, no
446matter how complicated. To find something between two single
447characters, a pattern like C</x([^x]*)x/> will get the intervening
448bits in \$1. For multiple ones, then something more like
449C</alpha(.*?)omega/> would be needed. But none of these deals with
450nested patterns, nor can they. For that you'll have to write a
451parser.
452
453If you are serious about writing a parser, there are a number of
6a2af475
GS
454modules or oddities that will make your life a lot easier. There are
455the CPAN modules Parse::RecDescent, Parse::Yapp, and Text::Balanced;
83df6a1d
JH
456and the byacc program. Starting from perl 5.8 the Text::Balanced
457is part of the standard distribution.
68dc0745 458
92c2ed05
GS
459One simple destructive, inside-out approach that you might try is to
460pull out the smallest nesting parts one at a time:
5a964f20 461
d92eb7b0 462 while (s/BEGIN((?:(?!BEGIN)(?!END).)*)END//gs) {
5a964f20
TC
463 # do something with \$1
464 }
465
65acb1b1
TC
466A more complicated and sneaky approach is to make Perl's regular
467expression engine do it for you. This is courtesy Dean Inada, and
468rather has the nature of an Obfuscated Perl Contest entry, but it
469really does work:
470
471 # \$_ contains the string to parse
472 # BEGIN and END are the opening and closing markers for the
473 # nested text.
c47ff5f1 474
65acb1b1
TC
475 @( = ('(','');
476 @) = (')','');
477 (\$re=\$_)=~s/((BEGIN)|(END)|.)/\$)[!\$3]\Q\$1\E\$([!\$2]/gs;
478 @\$ = (eval{/\$re/},\$@!~/unmatched/);
479 print join("\n",@\$[0..\$#\$]) if( \$\$[-1] );
480
68dc0745 481=head2 How do I reverse a string?
482
5a964f20 483Use reverse() in scalar context, as documented in
68dc0745 484L<perlfunc/reverse>.
485
486 \$reversed = reverse \$string;
487
488=head2 How do I expand tabs in a string?
489
5a964f20 490You can do it yourself:
68dc0745 491
492 1 while \$string =~ s/\t+/' ' x (length(\$&) * 8 - length(\$`) % 8)/e;
493
87275199 494Or you can just use the Text::Tabs module (part of the standard Perl
68dc0745 495distribution).
496
497 use Text::Tabs;
498 @expanded_lines = expand(@lines_with_tabs);
499
500=head2 How do I reformat a paragraph?
501
87275199 502Use Text::Wrap (part of the standard Perl distribution):
68dc0745 503
504 use Text::Wrap;
505 print wrap("\t", ' ', @paragraphs);
506
92c2ed05 507The paragraphs you give to Text::Wrap should not contain embedded
46fc3d4c 508newlines. Text::Wrap doesn't justify the lines (flush-right).
509
68dc0745 510=head2 How can I access/change the first N letters of a string?
511
512There are many ways. If you just want to grab a copy, use
92c2ed05 513substr():
68dc0745 514
515 \$first_byte = substr(\$a, 0, 1);
516
517If you want to modify part of a string, the simplest way is often to
518use substr() as an lvalue:
519
520 substr(\$a, 0, 3) = "Tom";
521
92c2ed05 522Although those with a pattern matching kind of thought process will
a6dd486b 523likely prefer
68dc0745 524
525 \$a =~ s/^.../Tom/;
526
527=head2 How do I change the Nth occurrence of something?
528
92c2ed05
GS
529You have to keep track of N yourself. For example, let's say you want
530to change the fifth occurrence of C<"whoever"> or C<"whomever"> into
d92eb7b0
GS
531C<"whosoever"> or C<"whomsoever">, case insensitively. These
532all assume that \$_ contains the string to be altered.
68dc0745 533
534 \$count = 0;
535 s{((whom?)ever)}{
536 ++\$count == 5 # is it the 5th?
537 ? "\${2}soever" # yes, swap
538 : \$1 # renege and leave it there
d92eb7b0 539 }ige;
68dc0745 540
5a964f20
TC
541In the more general case, you can use the C</g> modifier in a C<while>
542loop, keeping count of matches.
543
544 \$WANT = 3;
545 \$count = 0;
d92eb7b0 546 \$_ = "One fish two fish red fish blue fish";
5a964f20
TC
547 while (/(\w+)\s+fish\b/gi) {
548 if (++\$count == \$WANT) {
549 print "The third fish is a \$1 one.\n";
5a964f20
TC
550 }
551 }
552
92c2ed05 553That prints out: C<"The third fish is a red one."> You can also use a
5a964f20
TC
554repetition count and repeated pattern like this:
555
556 /(?:\w+\s+fish\s+){2}(\w+)\s+fish/i;
557
68dc0745 558=head2 How can I count the number of occurrences of a substring within a string?
559
a6dd486b 560There are a number of ways, with varying efficiency. If you want a
68dc0745 561count of a certain single character (X) within a string, you can use the
562C<tr///> function like so:
563
368c9434 564 \$string = "ThisXlineXhasXsomeXx'sXinXit";
68dc0745 565 \$count = (\$string =~ tr/X//);
d92eb7b0 566 print "There are \$count X characters in the string";
68dc0745 567
568This is fine if you are just looking for a single character. However,
569if you are trying to count multiple character substrings within a
570larger string, C<tr///> won't work. What you can do is wrap a while()
571loop around a global pattern match. For example, let's count negative
572integers:
573
574 \$string = "-9 55 48 -2 23 -76 4 14 -44";
575 while (\$string =~ /-\d+/g) { \$count++ }
576 print "There are \$count negative numbers in the string";
577
578=head2 How do I capitalize all the words on one line?
579
580To make the first letter of each word upper case:
3fe9a6f1 581
68dc0745 582 \$line =~ s/\b(\w)/\U\$1/g;
583
46fc3d4c 584This has the strange effect of turning "C<don't do it>" into "C<Don'T
a6dd486b
JB
585Do It>". Sometimes you might want this. Other times you might need a
586more thorough solution (Suggested by brian d. foy):
46fc3d4c 587
588 \$string =~ s/ (
589 (^\w) #at the beginning of the line
590 | # or
591 (\s\w) #preceded by whitespace
592 )
593 /\U\$1/xg;
594 \$string =~ /([\w']+)/\u\L\$1/g;
595
68dc0745 596To make the whole line upper case:
3fe9a6f1 597
68dc0745 598 \$line = uc(\$line);
599
600To force each word to be lower case, with the first letter upper case:
3fe9a6f1 601
68dc0745 602 \$line =~ s/(\w+)/\u\L\$1/g;
603
5a964f20
TC
604You can (and probably should) enable locale awareness of those
605characters by placing a C<use locale> pragma in your program.
92c2ed05 606See L<perllocale> for endless details on locales.
5a964f20 607
65acb1b1 608This is sometimes referred to as putting something into "title
d92eb7b0 609case", but that's not quite accurate. Consider the proper
65acb1b1
TC
610capitalization of the movie I<Dr. Strangelove or: How I Learned to
611Stop Worrying and Love the Bomb>, for example.
612
68dc0745 613=head2 How can I split a [character] delimited string except when inside
614[character]? (Comma-separated files)
615
616Take the example case of trying to split a string that is comma-separated
617into its different fields. (We'll pretend you said comma-separated, not
618comma-delimited, which is different and almost never what you mean.) You
619can't use C<split(/,/)> because you shouldn't split if the comma is inside
620quotes. For example, take a data line like this:
621
622 SAR001,"","Cimetrix, Inc","Bob Smith","CAM",N,8,1,0,7,"Error, Core Dumped"
623
624Due to the restriction of the quotes, this is a fairly complex
625problem. Thankfully, we have Jeffrey Friedl, author of a highly
626recommended book on regular expressions, to handle these for us. He
627suggests (assuming your string is contained in \$text):
628
629 @new = ();
630 push(@new, \$+) while \$text =~ m{
631 "([^\"\\]*(?:\\.[^\"\\]*)*)",? # groups the phrase inside the quotes
632 | ([^,]+),?
633 | ,
634 }gx;
635 push(@new, undef) if substr(\$text,-1,1) eq ',';
636
46fc3d4c 637If you want to represent quotation marks inside a
638quotation-mark-delimited field, escape them with backslashes (eg,
46fc3d4c 640this section.
641
87275199 642Alternatively, the Text::ParseWords module (part of the standard Perl
68dc0745 643distribution) lets you say:
644
645 use Text::ParseWords;
646 @new = quotewords(",", 0, \$text);
647
a6dd486b 648There's also a Text::CSV (Comma-Separated Values) module on CPAN.
65acb1b1 649
68dc0745 650=head2 How do I strip blank space from the beginning/end of a string?
651
a6dd486b 652Although the simplest approach would seem to be
68dc0745 653
654 \$string =~ s/^\s*(.*?)\s*\$/\$1/;
655
a6dd486b 656not only is this unnecessarily slow and destructive, it also fails with
d92eb7b0 657embedded newlines. It is much faster to do this operation in two steps:
68dc0745 658
659 \$string =~ s/^\s+//;
660 \$string =~ s/\s+\$//;
661
662Or more nicely written as:
663
664 for (\$string) {
665 s/^\s+//;
666 s/\s+\$//;
667 }
668
5e3006a4 669This idiom takes advantage of the C<foreach> loop's aliasing
5a964f20
TC
670behavior to factor out common code. You can do this
671on several strings at once, or arrays, or even the
d92eb7b0 672values of a hash if you use a slice:
5a964f20
TC
673
674 # trim whitespace in the scalar, the array,
675 # and all the values in the hash
676 foreach (\$scalar, @array, @hash{keys %hash}) {
677 s/^\s+//;
678 s/\s+\$//;
679 }
680
65acb1b1
TC
682
d92eb7b0
GS
683(This answer contributed by Uri Guttman, with kibitzing from
684Bart Lateur.)
65acb1b1
TC
685
686In the following examples, C<\$pad_len> is the length to which you wish
d92eb7b0
GS
687to pad the string, C<\$text> or C<\$num> contains the string to be padded,
690know what it is in advance. And in the same way you can use an integer in
65acb1b1 692
d92eb7b0
GS
693The simplest method uses the C<sprintf> function. It can pad on the left
694or right with blanks and on the left with zeroes and it will not
695truncate the result. The C<pack> function can only pad strings on the
696right with blanks and it will truncate the result to a maximum length of
65acb1b1 698
d92eb7b0
GS
699 # Left padding a string with blanks (no truncation):
65acb1b1 701
d92eb7b0
GS
702 # Right padding a string with blanks (no truncation):
65acb1b1 704
d92eb7b0
GS
705 # Left padding a number with 0 (no truncation):
65acb1b1 707
d92eb7b0
GS
708 # Right padding a string with blanks using pack (will truncate):
65acb1b1 710
d92eb7b0
GS
711If you need to pad with a character other than blank or zero you can use
712one of the following methods. They all generate a pad string with the
713C<x> operator and combine that with C<\$text>. These methods do
714not truncate C<\$text>.
65acb1b1 715
d92eb7b0 716Left and right padding with any character, creating a new string:
65acb1b1 717
d92eb7b0
GS
65acb1b1 720
d92eb7b0 721Left and right padding with any character, modifying C<\$text> directly:
65acb1b1 722
d92eb7b0
GS
723 substr( \$text, 0, 0 ) = \$pad_char x ( \$pad_len - length( \$text ) );
724 \$text .= \$pad_char x ( \$pad_len - length( \$text ) );
65acb1b1 725
68dc0745 726=head2 How do I extract selected columns from a string?
727
728Use substr() or unpack(), both documented in L<perlfunc>.
5a964f20
TC
729If you prefer thinking in terms of columns instead of widths,
730you can use this kind of thing:
731
732 # determine the unpack format needed to split Linux ps output
733 # arguments are cut columns
734 my \$fmt = cut2fmt(8, 14, 20, 26, 30, 34, 41, 47, 59, 63, 67, 72);
735
736 sub cut2fmt {
737 my(@positions) = @_;
738 my \$template = '';
739 my \$lastpos = 1;
740 for my \$place (@positions) {
741 \$template .= "A" . (\$place - \$lastpos) . " ";
742 \$lastpos = \$place;
743 }
744 \$template .= "A*";
745 return \$template;
746 }
68dc0745 747
748=head2 How do I find the soundex value of a string?
749
87275199 750Use the standard Text::Soundex module distributed with Perl.
a6dd486b 751Before you do so, you may want to determine whether `soundex' is in
d92eb7b0
GS
752fact what you think it is. Knuth's soundex algorithm compresses words
753into a small space, and so it does not necessarily distinguish between
754two words which you might want to appear separately. For example, the
755last names `Knuth' and `Kant' are both mapped to the soundex code K530.
756If Text::Soundex does not do what you are looking for, you might want
757to consider the String::Approx module available at CPAN.
68dc0745 758
759=head2 How can I expand variables in text strings?
760
761Let's assume that you have a string like:
762
763 \$text = 'this has a \$foo in it and a \$bar';
5a964f20
TC
764
765If those were both global variables, then this would
766suffice:
767
65acb1b1 768 \$text =~ s/\\$(\w+)/\${\$1}/g; # no /e needed
68dc0745 769
5a964f20
TC
770But since they are probably lexicals, or at least, they could
771be, you'd have to do this:
68dc0745 772
773 \$text =~ s/(\\$\w+)/\$1/eeg;
65acb1b1 774 die if \$@; # needed /ee, not /e
68dc0745 775
5a964f20
TC
776It's probably better in the general case to treat those
777variables as entries in some special hash. For example:
778
779 %user_defs = (
780 foo => 23,
781 bar => 19,
782 );
783 \$text =~ s/\\$(\w+)/\$user_defs{\$1}/g;
68dc0745 784
92c2ed05 785See also ``How do I expand function calls in a string?'' in this section
46fc3d4c 786of the FAQ.
787
68dc0745 788=head2 What's wrong with always quoting "\$vars"?
789
a6dd486b
JB
790The problem is that those double-quotes force stringification--
791coercing numbers and references into strings--even when you
792don't want them to be strings. Think of it this way: double-quote
65acb1b1
TC
793expansion is used to produce new strings. If you already
794have a string, why do you need more?
68dc0745 795
796If you get used to writing odd things like these:
797
799 \$new = "\$old"; # BAD
801
802You'll be in trouble. Those should (in 99.8% of the cases) be
803the simpler and more direct:
804
805 print \$var;
806 \$new = \$old;
807 somefunc(\$var);
808
809Otherwise, besides slowing you down, you're going to break code when
810the thing in the scalar is actually neither a string nor a number, but
811a reference:
812
813 func(\@array);
814 sub func {
815 my \$aref = shift;
816 my \$oref = "\$aref"; # WRONG
817 }
818
819You can also get into subtle problems on those few operations in Perl
820that actually do care about the difference between a string and a
821number, such as the magical C<++> autoincrement operator or the
822syscall() function.
823
5a964f20
TC
824Stringification also destroys arrays.
825
826 @lines = `command`;
827 print "@lines"; # WRONG - extra blanks
828 print @lines; # right
829
c47ff5f1 830=head2 Why don't my <<HERE documents work?
68dc0745 831
832Check for these three things:
833
834=over 4
835
836=item 1. There must be no space after the << part.
837
838=item 2. There (probably) should be a semicolon at the end.
839
840=item 3. You can't (easily) have any space in front of the tag.
841
842=back
843
5a964f20
TC
844If you want to indent the text in the here document, you
845can do this:
846
847 # all in one
848 (\$VAR = <<HERE_TARGET) =~ s/^\s+//gm;
850 goes here
851 HERE_TARGET
852
853But the HERE_TARGET must still be flush against the margin.
854If you want that indented also, you'll have to quote
855in the indentation.
856
857 (\$quote = <<' FINIS') =~ s/^\s+//gm;
858 ...we will have peace, when you and all your works have
859 perished--and the works of your dark master to whom you
860 would deliver us. You are a liar, Saruman, and a corrupter
861 of men's hearts. --Theoden in /usr/src/perl/taint.c
862 FINIS
863 \$quote =~ s/\s*--/\n--/;
864
865A nice general-purpose fixer-upper function for indented here documents
866follows. It expects to be called with a here document as its argument.
867It looks to see whether each line begins with a common substring, and
a6dd486b
JB
868if so, strips that substring off. Otherwise, it takes the amount of leading
869whitespace found on the first line and removes that much off each
5a964f20
TC
870subsequent line.
871
872 sub fix {
873 local \$_ = shift;
a6dd486b 874 my (\$white, \$leader); # common whitespace and common leading string
5a964f20
TC
875 if (/^\s*(?:([^\w\s]+)(\s*).*\n)(?:\s*\1\2?.*\n)+\$/) {
876 (\$white, \$leader) = (\$2, quotemeta(\$1));
877 } else {
878 (\$white, \$leader) = (/^(\s+)/, '');
879 }
881 return \$_;
882 }
883
c8db1d39 884This works with leading special strings, dynamically determined:
5a964f20
TC
885
886 \$remember_the_main = fix<<' MAIN_INTERPRETER_LOOP';
887 @@@ int
888 @@@ runops() {
889 @@@ SAVEI32(runlevel);
890 @@@ runlevel++;
d92eb7b0 891 @@@ while ( op = (*op->op_ppaddr)() );
5a964f20
TC
892 @@@ TAINT_NOT;
893 @@@ return 0;
894 @@@ }
895 MAIN_INTERPRETER_LOOP
896
a6dd486b 897Or with a fixed amount of leading whitespace, with remaining
5a964f20
TC
898indentation correctly preserved:
899
900 \$poem = fix<<EVER_ON_AND_ON;
902 And I must follow, if I can,
903 Pursuing it with eager feet,
904 Until it joins some larger way
905 Where many paths and errands meet.
906 And whither then? I cannot say.
907 --Bilbo in /usr/src/perl/pp_ctl.c
908 EVER_ON_AND_ON
909
911
65acb1b1
TC
912=head2 What is the difference between a list and an array?
913
914An array has a changeable length. A list does not. An array is something
915you can push or pop, while a list is a set of values. Some people make
916the distinction that a list is a value while an array is a variable.
917Subroutines are passed and return lists, you put things into list
918context, you initialize arrays with lists, and you foreach() across
919a list. C<@> variables are arrays, anonymous arrays are arrays, arrays
920in scalar context behave like the number of elements in them, subroutines
a6dd486b 921access their arguments through the array C<@_>, and push/pop/shift only work
65acb1b1
TC
922on arrays.
923
924As a side note, there's no such thing as a list in scalar context.
925When you say
926
927 \$scalar = (2, 5, 7, 9);
928
d92eb7b0
GS
929you're using the comma operator in scalar context, so it uses the scalar
930comma operator. There never was a list there at all! This causes the
931last value to be returned: 9.
65acb1b1 932
68dc0745 933=head2 What is the difference between \$array[1] and @array[1]?
934
a6dd486b 935The former is a scalar value; the latter an array slice, making
68dc0745 936it a list with one (scalar) value. You should use \$ when you want a
937scalar value (most of the time) and @ when you want a list with one
938scalar value in it (very, very rarely; nearly never, in fact).
939
940Sometimes it doesn't make a difference, but sometimes it does.
941For example, compare:
942
943 \$good[0] = `some program that outputs several lines`;
944
945with
946
947 @bad[0] = `same program that outputs several lines`;
948
9f1b1f2d
GS
949The C<use warnings> pragma and the B<-w> flag will warn you about these
950matters.
68dc0745 951
d92eb7b0 952=head2 How can I remove duplicate elements from a list or array?
68dc0745 953
954There are several possible ways, depending on whether the array is
955ordered and whether you wish to preserve the ordering.
956
957=over 4
958
551e1d92
RB
959=item a)
960
961If @in is sorted, and you want @out to be sorted:
5a964f20 962(this assumes all true values in the array)
68dc0745 963
a4341a65 964 \$prev = "not equal to \$in[0]";
3bc5ef3e 965 @out = grep(\$_ ne \$prev && (\$prev = \$_, 1), @in);
68dc0745 966
c8db1d39 967This is nice in that it doesn't use much extra memory, simulating
3bc5ef3e
HG
968uniq(1)'s behavior of removing only adjacent duplicates. The ", 1"
969guarantees that the expression is true (so that grep picks it up)
970even if the \$_ is 0, "", or undef.
68dc0745 971
551e1d92
RB
972=item b)
973
974If you don't know whether @in is sorted:
68dc0745 975
976 undef %saw;
977 @out = grep(!\$saw{\$_}++, @in);
978
551e1d92
RB
979=item c)
980
981Like (b), but @in contains only small integers:
68dc0745 982
983 @out = grep(!\$saw[\$_]++, @in);
984
551e1d92
RB
985=item d)
986
987A way to do (b) without any loops or greps:
68dc0745 988
989 undef %saw;
990 @saw{@in} = ();
991 @out = sort keys %saw; # remove sort if undesired
992
551e1d92
RB
993=item e)
994
995Like (d), but @in contains only small positive integers:
68dc0745 996
997 undef @ary;
998 @ary[@in] = @in;
87275199 999 @out = grep {defined} @ary;
68dc0745 1000
1001=back
1002
65acb1b1
TC
1003But perhaps you should have been using a hash all along, eh?
1004
ddbc1f16 1005=head2 How can I tell whether a certain element is contained in a list or array?
5a964f20
TC
1006
1007Hearing the word "in" is an I<in>dication that you probably should have
1008used a hash, not a list or array, to store your data. Hashes are
1009designed to answer this question quickly and efficiently. Arrays aren't.
68dc0745 1010
5a964f20
TC
1011That being said, there are several ways to approach this. If you
1012are going to make this query many times over arbitrary string values,
1013the fastest way is probably to invert the original array and keep an
68dc0745 1014associative array lying about whose keys are the first array's values.
1015
1016 @blues = qw/azure cerulean teal turquoise lapis-lazuli/;
1017 undef %is_blue;
1018 for (@blues) { \$is_blue{\$_} = 1 }
1019
1020Now you can check whether \$is_blue{\$some_color}. It might have been a
1021good idea to keep the blues all in a hash in the first place.
1022
1023If the values are all small integers, you could use a simple indexed
1024array. This kind of an array will take up less space:
1025
1026 @primes = (2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31);
1027 undef @is_tiny_prime;
d92eb7b0
GS
1028 for (@primes) { \$is_tiny_prime[\$_] = 1 }
1029 # or simply @istiny_prime[@primes] = (1) x @primes;
68dc0745 1030
1031Now you check whether \$is_tiny_prime[\$some_number].
1032
1033If the values in question are integers instead of strings, you can save
1034quite a lot of space by using bit strings instead:
1035
1036 @articles = ( 1..10, 150..2000, 2017 );
7b8d334a 1038 for (@articles) { vec(\$read,\$_,1) = 1 }
68dc0745 1039
1040Now check whether C<vec(\$read,\$n,1)> is true for some C<\$n>.
1041
1043
a6dd486b 1044 (\$is_there) = grep \$_ eq \$whatever, @array;
68dc0745 1045
1046or worse yet
1047
a6dd486b 1048 (\$is_there) = grep /\$whatever/, @array;
68dc0745 1049
1050These are slow (checks every element even if the first matches),
1051inefficient (same reason), and potentially buggy (what if there are
d92eb7b0 1052regex characters in \$whatever?). If you're only testing once, then
65acb1b1
TC
1053use:
1054
1055 \$is_there = 0;
1056 foreach \$elt (@array) {
1057 if (\$elt eq \$elt_to_find) {
1058 \$is_there = 1;
1059 last;
1060 }
1061 }
1062 if (\$is_there) { ... }
68dc0745 1063
1064=head2 How do I compute the difference of two arrays? How do I compute the intersection of two arrays?
1065
1066Use a hash. Here's code to do both and more. It assumes that
1067each element is unique in a given array:
1068
1069 @union = @intersection = @difference = ();
1070 %count = ();
1071 foreach \$element (@array1, @array2) { \$count{\$element}++ }
1072 foreach \$element (keys %count) {
1073 push @union, \$element;
1074 push @{ \$count{\$element} > 1 ? \@intersection : \@difference }, \$element;
1075 }
1076
d92eb7b0 1077Note that this is the I<symmetric difference>, that is, all elements in
a6dd486b 1078either A or in B but not in both. Think of it as an xor operation.
d92eb7b0 1079
65acb1b1
TC
1080=head2 How do I test whether two arrays or hashes are equal?
1081
1082The following code works for single-level arrays. It uses a stringwise
1083comparison, and does not distinguish defined versus undefined empty
1084strings. Modify if you have other needs.
1085
1087
1088 sub compare_arrays {
1089 my (\$first, \$second) = @_;
9f1b1f2d 1090 no warnings; # silence spurious -w undef complaints
65acb1b1
TC
1091 return 0 unless @\$first == @\$second;
1092 for (my \$i = 0; \$i < @\$first; \$i++) {
1093 return 0 if \$first->[\$i] ne \$second->[\$i];
1094 }
1095 return 1;
1096 }
1097
1098For multilevel structures, you may wish to use an approach more
1099like this one. It uses the CPAN module FreezeThaw:
1100
1101 use FreezeThaw qw(cmpStr);
1102 @a = @b = ( "this", "that", [ "more", "stuff" ] );
1103
1104 printf "a and b contain %s arrays\n",
1105 cmpStr(\@a, \@b) == 0
1106 ? "the same"
1107 : "different";
1108
1109This approach also works for comparing hashes. Here
1111
1112 use FreezeThaw qw(cmpStr cmpStrHard);
1113
1114 %a = %b = ( "this" => "that", "extra" => [ "more", "stuff" ] );
1115 \$a{EXTRA} = \%b;
1116 \$b{EXTRA} = \%a;
1117
1118 printf "a and b contain %s hashes\n",
1119 cmpStr(\%a, \%b) == 0 ? "the same" : "different";
1120
1121 printf "a and b contain %s hashes\n",
1122 cmpStrHard(\%a, \%b) == 0 ? "the same" : "different";
1123
1124
1125The first reports that both those the hashes contain the same data,
1126while the second reports that they do not. Which you prefer is left as
1128
68dc0745 1129=head2 How do I find the first array element for which a condition is true?
1130
1131You can use this if you care about the index:
1132
65acb1b1 1133 for (\$i= 0; \$i < @array; \$i++) {
68dc0745 1134 if (\$array[\$i] eq "Waldo") {
1135 \$found_index = \$i;
1136 last;
1137 }
1138 }
1139
1140Now C<\$found_index> has what you want.
1141
1143
1144In general, you usually don't need a linked list in Perl, since with
1145regular arrays, you can push and pop or shift and unshift at either end,
5a964f20 1146or you can use splice to add and/or remove arbitrary number of elements at
87275199 1147arbitrary points. Both pop and shift are both O(1) operations on Perl's
5a964f20
TC
1148dynamic arrays. In the absence of shifts and pops, push in general
1149needs to reallocate on the order every log(N) times, and unshift will
1150need to copy pointers each time.
68dc0745 1151
1152If you really, really wanted, you could use structures as described in
1153L<perldsc> or L<perltoot> and do just what the algorithm book tells you
65acb1b1
TC
1154to do. For example, imagine a list node like this:
1155
1156 \$node = {
1157 VALUE => 42,
1159 };
1160
1161You could walk the list this way:
1162
1163 print "List: ";
1165 print \$node->{VALUE}, " ";
1166 }
1167 print "\n";
1168
a6dd486b 1169You could add to the list this way:
65acb1b1
TC
1170
1173 for \$value ( 2 .. 10 ) {
1174 \$tail = append(\$tail, \$value);
1175 }
1176
1177 sub append {
1178 my(\$list, \$value) = @_;
1179 my \$node = { VALUE => \$value };
1180 if (\$list) {
1183 } else {
1184 \$_[0] = \$node; # replace caller's version
1185 }
1186 return \$node;
1187 }
1188
1189But again, Perl's built-in are virtually always good enough.
68dc0745 1190
1191=head2 How do I handle circular lists?
1192
1194lists, or you could just do something like this with an array:
1195
1196 unshift(@array, pop(@array)); # the last shall be first
1197 push(@array, shift(@array)); # and vice versa
1198
1199=head2 How do I shuffle an array randomly?
1200
5a964f20
TC
1201Use this:
1202
1203 # fisher_yates_shuffle( \@array ) :
1204 # generate a random permutation of @array in place
1205 sub fisher_yates_shuffle {
1206 my \$array = shift;
1207 my \$i;
1208 for (\$i = @\$array; --\$i; ) {
1209 my \$j = int rand (\$i+1);
5a964f20
TC
1210 @\$array[\$i,\$j] = @\$array[\$j,\$i];
1211 }
1212 }
1213
1214 fisher_yates_shuffle( \@array ); # permutes @array in place
1215
d92eb7b0 1216You've probably seen shuffling algorithms that work using splice,
a6dd486b 1217randomly picking another element to swap the current element with
68dc0745 1218
1219 srand;
1220 @new = ();
1221 @old = 1 .. 10; # just a demo
1222 while (@old) {
1223 push(@new, splice(@old, rand @old, 1));
1224 }
1225
5a964f20
TC
1226This is bad because splice is already O(N), and since you do it N times,
1227you just invented a quadratic algorithm; that is, O(N**2). This does
1228not scale, although Perl is so efficient that you probably won't notice
1229this until you have rather largish arrays.
68dc0745 1230
1231=head2 How do I process/modify each element of an array?
1232
1233Use C<for>/C<foreach>:
1234
1235 for (@lines) {
5a964f20
TC
1236 s/foo/bar/; # change that word
1237 y/XZ/ZX/; # swap those letters
68dc0745 1238 }
1239
1240Here's another; let's compute spherical volumes:
1241
5a964f20 1242 for (@volumes = @radii) { # @volumes has changed parts
68dc0745 1243 \$_ **= 3;
1244 \$_ *= (4/3) * 3.14159; # this will be constant folded
1245 }
1246
5a964f20
TC
1247If you want to do the same thing to modify the values of the hash,
1248you may not use the C<values> function, oddly enough. You need a slice:
1249
1250 for \$orbit ( @orbits{keys %orbits} ) {
1251 (\$orbit **= 3) *= (4/3) * 3.14159;
1252 }
1253
68dc0745 1254=head2 How do I select a random element from an array?
1255
1256Use the rand() function (see L<perlfunc/rand>):
1257
5a964f20 1258 # at the top of the program:
68dc0745 1259 srand; # not needed for 5.004 and later
5a964f20
TC
1260
1261 # then later on
68dc0745 1262 \$index = rand @array;
1263 \$element = \$array[\$index];
1264
5a964f20
TC
1265Make sure you I<only call srand once per program, if then>.
1266If you are calling it more than once (such as before each
1267call to rand), you're almost certainly doing something wrong.
1268
68dc0745 1269=head2 How do I permute N elements of a list?
1270
1271Here's a little program that generates all permutations
1272of all the words on each line of input. The algorithm embodied
5a964f20 1273in the permute() function should work on any list:
68dc0745 1274
1275 #!/usr/bin/perl -n
5a964f20
TC
1276 # tsc-permute: permute each word of input
1277 permute([split], []);
1278 sub permute {
1279 my @items = @{ \$_[0] };
1280 my @perms = @{ \$_[1] };
1281 unless (@items) {
1282 print "@perms\n";
68dc0745 1283 } else {
5a964f20
TC
1284 my(@newitems,@newperms,\$i);
1285 foreach \$i (0 .. \$#items) {
1286 @newitems = @items;
1287 @newperms = @perms;
1288 unshift(@newperms, splice(@newitems, \$i, 1));
1289 permute([@newitems], [@newperms]);
68dc0745 1290 }
1291 }
1292 }
1293
b8d2732a
RH
1294Unfortunately, this algorithm is very inefficient. The Algorithm::Permute
1295module from CPAN runs at least an order of magnitude faster. If you don't
1296have a C compiler (or a binary distribution of Algorithm::Permute), then
1297you can use List::Permutor which is written in pure Perl, and is still
1298several times faster than the toy algorithm above.
1299
68dc0745 1300=head2 How do I sort an array by (anything)?
1301
1302Supply a comparison function to sort() (described in L<perlfunc/sort>):
1303
1304 @list = sort { \$a <=> \$b } @list;
1305
1306The default sort function is cmp, string comparison, which would
c47ff5f1 1307sort C<(1, 2, 10)> into C<(1, 10, 2)>. C<< <=> >>, used above, is
68dc0745 1308the numerical comparison operator.
1309
1310If you have a complicated function needed to pull out the part you
1311want to sort on, then don't do it inside the sort function. Pull it
1312out first, because the sort BLOCK can be called many times for the
1313same element. Here's an example of how to pull out the first word
1314after the first number on each item, and then sort those words
1315case-insensitively.
1316
1317 @idx = ();
1318 for (@data) {
1319 (\$item) = /\d+\s*(\S+)/;
1320 push @idx, uc(\$item);
1321 }
1322 @sorted = @data[ sort { \$idx[\$a] cmp \$idx[\$b] } 0 .. \$#idx ];
1323
a6dd486b 1324which could also be written this way, using a trick
68dc0745 1325that's come to be known as the Schwartzian Transform:
1326
1327 @sorted = map { \$_->[0] }
1328 sort { \$a->[1] cmp \$b->[1] }
d92eb7b0 1329 map { [ \$_, uc( (/\d+\s*(\S+)/)[0]) ] } @data;
68dc0745 1330
1331If you need to sort on several fields, the following paradigm is useful.
1332
1333 @sorted = sort { field1(\$a) <=> field1(\$b) ||
1334 field2(\$a) cmp field2(\$b) ||
1335 field3(\$a) cmp field3(\$b)
1336 } @data;
1337
1338This can be conveniently combined with precalculation of keys as given
1339above.
1340
1342this approach.
1343
1345
1346=head2 How do I manipulate arrays of bits?
1347
1348Use pack() and unpack(), or else vec() and the bitwise operations.
1349
1350For example, this sets \$vec to have bit N set if \$ints[N] was set:
1351
1352 \$vec = '';
1353 foreach(@ints) { vec(\$vec,\$_,1) = 1 }
1354
1355And here's how, given a vector in \$vec, you can
1356get those bits into your @ints array:
1357
1358 sub bitvec_to_list {
1359 my \$vec = shift;
1360 my @ints;
1361 # Find null-byte density then select best algorithm
1362 if (\$vec =~ tr/\0// / length \$vec > 0.95) {
1363 use integer;
1364 my \$i;
1365 # This method is faster with mostly null-bytes
1366 while(\$vec =~ /[^\0]/g ) {
1367 \$i = -9 + 8 * pos \$vec;
1368 push @ints, \$i if vec(\$vec, ++\$i, 1);
1369 push @ints, \$i if vec(\$vec, ++\$i, 1);
1370 push @ints, \$i if vec(\$vec, ++\$i, 1);
1371 push @ints, \$i if vec(\$vec, ++\$i, 1);
1372 push @ints, \$i if vec(\$vec, ++\$i, 1);
1373 push @ints, \$i if vec(\$vec, ++\$i, 1);
1374 push @ints, \$i if vec(\$vec, ++\$i, 1);
1375 push @ints, \$i if vec(\$vec, ++\$i, 1);
1376 }
1377 } else {
1378 # This method is a fast general algorithm
1379 use integer;
1380 my \$bits = unpack "b*", \$vec;
1381 push @ints, 0 if \$bits =~ s/^(\d)// && \$1;
1382 push @ints, pos \$bits while(\$bits =~ /1/g);
1383 }
1384 return \@ints;
1385 }
1386
1387This method gets faster the more sparse the bit vector is.
1388(Courtesy of Tim Bunce and Winfried Koenig.)
1389
65acb1b1
TC
1390Here's a demo on how to use vec():
1391
1392 # vec demo
1393 \$vector = "\xff\x0f\xef\xfe";
1394 print "Ilya's string \\xff\\x0f\\xef\\xfe represents the number ",
1395 unpack("N", \$vector), "\n";
1396 \$is_set = vec(\$vector, 23, 1);
1397 print "Its 23rd bit is ", \$is_set ? "set" : "clear", ".\n";
1398 pvec(\$vector);
1399
1400 set_vec(1,1,1);
1401 set_vec(3,1,1);
1402 set_vec(23,1,1);
1403
1404 set_vec(3,1,3);
1405 set_vec(3,2,3);
1406 set_vec(3,4,3);
1407 set_vec(3,4,7);
1408 set_vec(3,8,3);
1409 set_vec(3,8,7);
1410
1411 set_vec(0,32,17);
1412 set_vec(1,32,17);
1413
1414 sub set_vec {
1415 my (\$offset, \$width, \$value) = @_;
1416 my \$vector = '';
1417 vec(\$vector, \$offset, \$width) = \$value;
1418 print "offset=\$offset width=\$width value=\$value\n";
1419 pvec(\$vector);
1420 }
1421
1422 sub pvec {
1423 my \$vector = shift;
1424 my \$bits = unpack("b*", \$vector);
1425 my \$i = 0;
1426 my \$BASE = 8;
1427
1428 print "vector length in bytes: ", length(\$vector), "\n";
1429 @bytes = unpack("A8" x length(\$vector), \$bits);
1430 print "bits are: @bytes\n\n";
1431 }
1432
68dc0745 1433=head2 Why does defined() return true on empty arrays and hashes?
1434
65acb1b1
TC
1435The short story is that you should probably only use defined on scalars or
1436functions, not on aggregates (arrays and hashes). See L<perlfunc/defined>
1437in the 5.004 release or later of Perl for more detail.
68dc0745 1438
1440
1441=head2 How do I process an entire hash?
1442
1443Use the each() function (see L<perlfunc/each>) if you don't care
1444whether it's sorted:
1445
5a964f20 1446 while ( (\$key, \$value) = each %hash) {
68dc0745 1447 print "\$key = \$value\n";
1448 }
1449
1450If you want it sorted, you'll have to use foreach() on the result of
1451sorting the keys as shown in an earlier question.
1452
1453=head2 What happens if I add or remove keys from a hash while iterating over it?
1454
d92eb7b0
GS
1455Don't do that. :-)
1456
1457[lwall] In Perl 4, you were not allowed to modify a hash at all while
87275199 1458iterating over it. In Perl 5 you can delete from it, but you still
d92eb7b0
GS
1459can't add to it, because that might cause a doubling of the hash table,
1460in which half the entries get copied up to the new top half of the
87275199 1461table, at which point you've totally bamboozled the iterator code.
d92eb7b0
GS
1462Even if the table doesn't double, there's no telling whether your new
1463entry will be inserted before or after the current iterator position.
1464
a6dd486b 1465Either treasure up your changes and make them after the iterator finishes
d92eb7b0
GS
1466or use keys to fetch all the old keys at once, and iterate over the list
1467of keys.
68dc0745 1468
1469=head2 How do I look up a hash element by value?
1470
1471Create a reverse hash:
1472
1473 %by_value = reverse %by_key;
1474 \$key = \$by_value{\$value};
1475
1476That's not particularly efficient. It would be more space-efficient
1477to use:
1478
1479 while ((\$key, \$value) = each %by_key) {
1480 \$by_value{\$value} = \$key;
1481 }
1482
d92eb7b0
GS
1483If your hash could have repeated values, the methods above will only find
1484one of the associated keys. This may or may not worry you. If it does
1485worry you, you can always reverse the hash into a hash of arrays instead:
1486
1487 while ((\$key, \$value) = each %by_key) {
1488 push @{\$key_list_by_value{\$value}}, \$key;
1489 }
68dc0745 1490
1491=head2 How can I know how many entries are in a hash?
1492
1493If you mean how many keys, then all you have to do is
1494take the scalar sense of the keys() function:
1495
3fe9a6f1 1496 \$num_keys = scalar keys %hash;
68dc0745 1497
a6dd486b 1498The keys() function also resets the iterator, which in void context is
d92eb7b0
GS
1499faster for tied hashes than would be iterating through the whole
1500hash, one key-value pair at a time.
68dc0745 1501
1502=head2 How do I sort a hash (optionally by value instead of key)?
1503
1504Internally, hashes are stored in a way that prevents you from imposing
1505an order on key-value pairs. Instead, you have to sort a list of the
1506keys or values:
1507
1508 @keys = sort keys %hash; # sorted by key
1509 @keys = sort {
1510 \$hash{\$a} cmp \$hash{\$b}
1511 } keys %hash; # and by value
1512
1513Here we'll do a reverse numeric sort by value, and if two keys are
a6dd486b
JB
1514identical, sort by length of key, or if that fails, by straight ASCII
1515comparison of the keys (well, possibly modified by your locale--see
68dc0745 1516L<perllocale>).
1517
1518 @keys = sort {
1519 \$hash{\$b} <=> \$hash{\$a}
1520 ||
1521 length(\$b) <=> length(\$a)
1522 ||
1523 \$a cmp \$b
1524 } keys %hash;
1525
1526=head2 How can I always keep my hash sorted?
1527
1528You can look into using the DB_File module and tie() using the
1529\$DB_BTREE hash bindings as documented in L<DB_File/"In Memory Databases">.
5a964f20 1530The Tie::IxHash module from CPAN might also be instructive.
68dc0745 1531
1532=head2 What's the difference between "delete" and "undef" with hashes?
1533
1534Hashes are pairs of scalars: the first is the key, the second is the
1535value. The key will be coerced to a string, although the value can be
1536any kind of scalar: string, number, or reference. If a key C<\$key> is
1537present in the array, C<exists(\$key)> will return true. The value for
1538a given key can be C<undef>, in which case C<\$array{\$key}> will be
1539C<undef> while C<\$exists{\$key}> will return true. This corresponds to
1540(C<\$key>, C<undef>) being in the hash.
1541
1542Pictures help... here's the C<%ary> table:
1543
1544 keys values
1545 +------+------+
1546 | a | 3 |
1547 | x | 7 |
1548 | d | 0 |
1549 | e | 2 |
1550 +------+------+
1551
1552And these conditions hold
1553
1554 \$ary{'a'} is true
1555 \$ary{'d'} is false
1556 defined \$ary{'d'} is true
1557 defined \$ary{'a'} is true
87275199 1558 exists \$ary{'a'} is true (Perl5 only)
68dc0745 1559 grep (\$_ eq 'a', keys %ary) is true
1560
1561If you now say
1562
1563 undef \$ary{'a'}
1564
1566
1567
1568 keys values
1569 +------+------+
1570 | a | undef|
1571 | x | 7 |
1572 | d | 0 |
1573 | e | 2 |
1574 +------+------+
1575
1576and these conditions now hold; changes in caps:
1577
1578 \$ary{'a'} is FALSE
1579 \$ary{'d'} is false
1580 defined \$ary{'d'} is true
1581 defined \$ary{'a'} is FALSE
87275199 1582 exists \$ary{'a'} is true (Perl5 only)
68dc0745 1583 grep (\$_ eq 'a', keys %ary) is true
1584
1585Notice the last two: you have an undef value, but a defined key!
1586
1587Now, consider this:
1588
1589 delete \$ary{'a'}
1590
1592
1593 keys values
1594 +------+------+
1595 | x | 7 |
1596 | d | 0 |
1597 | e | 2 |
1598 +------+------+
1599
1600and these conditions now hold; changes in caps:
1601
1602 \$ary{'a'} is false
1603 \$ary{'d'} is false
1604 defined \$ary{'d'} is true
1605 defined \$ary{'a'} is false
87275199 1606 exists \$ary{'a'} is FALSE (Perl5 only)
68dc0745 1607 grep (\$_ eq 'a', keys %ary) is FALSE
1608
1609See, the whole entry is gone!
1610
1611=head2 Why don't my tied hashes make the defined/exists distinction?
1612
1613They may or may not implement the EXISTS() and DEFINED() methods
1614differently. For example, there isn't the concept of undef with hashes
1615that are tied to DBM* files. This means the true/false tables above
1616will give different results when used on such a hash. It also means
1617that exists and defined do the same thing with a DBM* file, and what
1618they end up doing is not what they do with ordinary hashes.
1619
1620=head2 How do I reset an each() operation part-way through?
1621
5a964f20 1622Using C<keys %hash> in scalar context returns the number of keys in
68dc0745 1623the hash I<and> resets the iterator associated with the hash. You may
1624need to do this if you use C<last> to exit a loop early so that when you
46fc3d4c 1625re-enter it, the hash iterator has been reset.
68dc0745 1626
1627=head2 How can I get the unique keys from two hashes?
1628
d92eb7b0
GS
1629First you extract the keys from the hashes into lists, then solve
1630the "removing duplicates" problem described above. For example:
68dc0745 1631
1632 %seen = ();
1633 for \$element (keys(%foo), keys(%bar)) {
1634 \$seen{\$element}++;
1635 }
1636 @uniq = keys %seen;
1637
1638Or more succinctly:
1639
1640 @uniq = keys %{{%foo,%bar}};
1641
1642Or if you really want to save space:
1643
1644 %seen = ();
1645 while (defined (\$key = each %foo)) {
1646 \$seen{\$key}++;
1647 }
1648 while (defined (\$key = each %bar)) {
1649 \$seen{\$key}++;
1650 }
1651 @uniq = keys %seen;
1652
1653=head2 How can I store a multidimensional array in a DBM file?
1654
1655Either stringify the structure yourself (no fun), or else
1656get the MLDBM (which uses Data::Dumper) module from CPAN and layer
1657it on top of either DB_File or GDBM_File.
1658
1659=head2 How can I make my hash remember the order I put elements into it?
1660
1661Use the Tie::IxHash from CPAN.
1662
46fc3d4c 1663 use Tie::IxHash;
1664 tie(%myhash, Tie::IxHash);
1665 for (\$i=0; \$i<20; \$i++) {
1666 \$myhash{\$i} = 2*\$i;
1667 }
1668 @keys = keys %myhash;
1669 # @keys = (0,1,2,3,...)
1670
68dc0745 1671=head2 Why does passing a subroutine an undefined element in a hash create it?
1672
1673If you say something like:
1674
1675 somefunc(\$hash{"nonesuch key here"});
1676
1677Then that element "autovivifies"; that is, it springs into existence
1678whether you store something there or not. That's because functions
1679get scalars passed in by reference. If somefunc() modifies C<\$_[0]>,
1680it has to be ready to write it back into the caller's version.
1681
87275199 1682This has been fixed as of Perl5.004.
68dc0745 1683
1684Normally, merely accessing a key's value for a nonexistent key does
1685I<not> cause that key to be forever there. This is different than
1686awk's behavior.
1687
fc36a67e 1688=head2 How can I make the Perl equivalent of a C structure/C++ class/hash or array of hashes or arrays?
68dc0745 1689
65acb1b1
TC
1690Usually a hash ref, perhaps like this:
1691
1692 \$record = {
1693 NAME => "Jason",
1694 EMPNO => 132,
1695 TITLE => "deputy peon",
1696 AGE => 23,
1697 SALARY => 37_000,
1698 PALS => [ "Norbert", "Rhys", "Phineas"],
1699 };
1700
1701References are documented in L<perlref> and the upcoming L<perlreftut>.
1702Examples of complex data structures are given in L<perldsc> and
1703L<perllol>. Examples of structures and object-oriented classes are
1704in L<perltoot>.
68dc0745 1705
1706=head2 How can I use a reference as a hash key?
1707
1708You can't do this directly, but you could use the standard Tie::Refhash
87275199 1709module distributed with Perl.
68dc0745 1710
1712
1713=head2 How do I handle binary data correctly?
1714
1715Perl is binary clean, so this shouldn't be a problem. For example,
1716this works fine (assuming the files are found):
1717
1718 if (`cat /vmunix` =~ /gzip/) {
1719 print "Your kernel is GNU-zip enabled!\n";
1720 }
1721
d92eb7b0
GS
1722On less elegant (read: Byzantine) systems, however, you have
1723to play tedious games with "text" versus "binary" files. See
1724L<perlfunc/"binmode"> or L<perlopentut>. Most of these ancient-thinking
1725systems are curses out of Microsoft, who seem to be committed to putting
1726the backward into backward compatibility.
68dc0745 1727
1728If you're concerned about 8-bit ASCII data, then see L<perllocale>.
1729
54310121 1730If you want to deal with multibyte characters, however, there are
68dc0745 1731some gotchas. See the section on Regular Expressions.
1732
1733=head2 How do I determine whether a scalar is a number/whole/integer/float?
1734
1735Assuming that you don't care about IEEE notations like "NaN" or
1736"Infinity", you probably just want to use a regular expression.
1737
65acb1b1
TC
1738 if (/\D/) { print "has nondigits\n" }
1739 if (/^\d+\$/) { print "is a whole number\n" }
1740 if (/^-?\d+\$/) { print "is an integer\n" }
1741 if (/^[+-]?\d+\$/) { print "is a +/- integer\n" }
1742 if (/^-?\d+\.?\d*\$/) { print "is a real number\n" }
1743 if (/^-?(?:\d+(?:\.\d*)?|\.\d+)\$/) { print "is a decimal number" }
1744 if (/^([+-]?)(?=\d|\.\d)\d*(\.\d*)?([Ee]([+-]?\d+))?\$/)
1745 { print "a C float" }
68dc0745 1746
5a964f20
TC
1747If you're on a POSIX system, Perl's supports the C<POSIX::strtod>
1748function. Its semantics are somewhat cumbersome, so here's a C<getnum>
1749wrapper function for more convenient access. This function takes
1750a string and returns the number it found, or C<undef> for input that
1751isn't a C float. The C<is_numeric> function is a front end to C<getnum>
1752if you just want to say, ``Is this a float?''
1753
1754 sub getnum {
1755 use POSIX qw(strtod);
1756 my \$str = shift;
1757 \$str =~ s/^\s+//;
1758 \$str =~ s/\s+\$//;
1759 \$! = 0;
1760 my(\$num, \$unparsed) = strtod(\$str);
1761 if ((\$str eq '') || (\$unparsed != 0) || \$!) {
1762 return undef;
1763 } else {
1764 return \$num;
1765 }
1766 }
1767
072dc14b 1768 sub is_numeric { defined getnum(\$_[0]) }
5a964f20 1769
6cecdcac
GS
1770Or you could check out the String::Scanf module on CPAN instead. The
1771POSIX module (part of the standard Perl distribution) provides the
bf4acbe4 1772C<strtod> and C<strtol> for converting strings to double and longs,
6cecdcac 1773respectively.
68dc0745 1774
1775=head2 How do I keep persistent data across program calls?
1776
1777For some specific applications, you can use one of the DBM modules.
65acb1b1 1778See L<AnyDBM_File>. More generically, you should consult the FreezeThaw,
83df6a1d
JH
1779Storable, or Class::Eroot modules from CPAN. Starting from Perl 5.8
1780Storable is part of the standard distribution. Here's one example using
65acb1b1
TC
1781Storable's C<store> and C<retrieve> functions:
1782
1783 use Storable;
1784 store(\%hash, "filename");
1785
1786 # later on...
1787 \$href = retrieve("filename"); # by ref
1788 %hash = %{ retrieve("filename") }; # direct to hash
68dc0745 1789
1790=head2 How do I print out or copy a recursive data structure?
1791
65acb1b1
TC
1792The Data::Dumper module on CPAN (or the 5.005 release of Perl) is great
1793for printing out data structures. The Storable module, found on CPAN,
1794provides a function called C<dclone> that recursively copies its argument.
1795
1796 use Storable qw(dclone);
1797 \$r2 = dclone(\$r1);
68dc0745 1798
65acb1b1
TC
1799Where \$r1 can be a reference to any kind of data structure you'd like.
1800It will be deeply copied. Because C<dclone> takes and returns references,
1801you'd have to add extra punctuation if you had a hash of arrays that
1802you wanted to copy.
68dc0745 1803
65acb1b1 1804 %newhash = %{ dclone(\%oldhash) };
68dc0745 1805
1806=head2 How do I define methods for every class/object?
1807
1808Use the UNIVERSAL class (see L<UNIVERSAL>).
1809
1810=head2 How do I verify a credit card checksum?
1811
1812Get the Business::CreditCard module from CPAN.
1813
65acb1b1
TC
1814=head2 How do I pack arrays of doubles or floats for XS code?
1815
1816The kgbpack.c code in the PGPLOT module on CPAN does just this.
1817If you're doing a lot of float or double processing, consider using
1818the PDL module from CPAN instead--it makes number-crunching easy.
1819
1821
65acb1b1 1822Copyright (c) 1997-1999 Tom Christiansen and Nathan Torkington.
5a964f20
TC