.\" -*- mode: troff; coding: utf-8 -*- .\" Automatically generated by Pod::Man 5.0102 (Pod::Simple 3.45) .\" .\" Standard preamble: .\" ======================================================================== .de Sp \" Vertical space (when we can't use .PP) .if t .sp .5v .if n .sp .. .de Vb \" Begin verbatim text .ft CW .nf .ne \\$1 .. .de Ve \" End verbatim text .ft R .fi .. .\" \*(C` and \*(C' are quotes in nroff, nothing in troff, for use with C<>. .ie n \{\ . ds C` "" . ds C' "" 'br\} .el\{\ . ds C` . ds C' 'br\} .\" .\" Escape single quotes in literal strings from groff's Unicode transform. .ie \n(.g .ds Aq \(aq .el .ds Aq ' .\" .\" If the F register is >0, we'll generate index entries on stderr for .\" titles (.TH), headers (.SH), subsections (.SS), items (.Ip), and index .\" entries marked with X<> in POD. Of course, you'll have to process the .\" output yourself in some meaningful fashion. .\" .\" Avoid warning from groff about undefined register 'F'. .de IX .. .nr rF 0 .if \n(.g .if rF .nr rF 1 .if (\n(rF:(\n(.g==0)) \{\ . if \nF \{\ . de IX . tm Index:\\$1\t\\n%\t"\\$2" .. . if !\nF==2 \{\ . nr % 0 . nr F 2 . \} . \} .\} .rr rF .\" ======================================================================== .\" .IX Title "List::UtilsBy 3" .TH List::UtilsBy 3 2024-09-01 "perl v5.40.0" "User Contributed Perl Documentation" .\" For nroff, turn off justification. Always turn off hyphenation; it makes .\" way too many mistakes in technical documents. .if n .ad l .nh .SH NAME "List::UtilsBy" \- higher\-order list utility functions .SH SYNOPSIS .IX Header "SYNOPSIS" .Vb 1 \& use List::UtilsBy qw( nsort_by min_by ); \& \& use File::stat qw( stat ); \& my @files_by_age = nsort_by { stat($_)\->mtime } @files; \& \& my $shortest_name = min_by { length } @names; .Ve .SH DESCRIPTION .IX Header "DESCRIPTION" This module provides a number of list utility functions, all of which take an initial code block to control their behaviour. They are variations on similar core perl or \f(CW\*(C`List::Util\*(C'\fR functions of similar names, but which use the block to control their behaviour. For example, the core Perl function \f(CW\*(C`sort\*(C'\fR takes a list of values and returns them, sorted into order by their string value. The "sort_by" function sorts them according to the string value returned by the extra function, when given each value. .PP .Vb 1 \& my @names_sorted = sort @names; \& \& my @people_sorted = sort_by { $_\->name } @people; .Ve .SH FUNCTIONS .IX Header "FUNCTIONS" All functions added since version 0.04 unless otherwise stated, as the original names for earlier versions were renamed. .SS sort_by .IX Subsection "sort_by" .Vb 1 \& @vals = sort_by { KEYFUNC } @vals .Ve .PP Returns the list of values sorted according to the string values returned by the \f(CW\*(C`KEYFUNC\*(C'\fR block or function. A typical use of this may be to sort objects according to the string value of some accessor, such as .PP .Vb 1 \& sort_by { $_\->name } @people .Ve .PP The key function is called in scalar context, being passed each value in turn as both \f(CW$_\fR and the only argument in the parameters, \f(CW@_\fR. The values are then sorted according to string comparisons on the values returned. .PP This is equivalent to .PP .Vb 1 \& sort { $a\->name cmp $b\->name } @people .Ve .PP except that it guarantees the \f(CW\*(C`name\*(C'\fR accessor will be executed only once per value. .PP One interesting use-case is to sort strings which may have numbers embedded in them "naturally", rather than lexically. .PP .Vb 1 \& sort_by { s/(\ed+)/sprintf "%09d", $1/eg; $_ } @strings .Ve .PP This sorts strings by generating sort keys which zero-pad the embedded numbers to some level (9 digits in this case), helping to ensure the lexical sort puts them in the correct order. .SS nsort_by .IX Subsection "nsort_by" .Vb 1 \& @vals = nsort_by { KEYFUNC } @vals .Ve .PP Similar to "sort_by" but compares its key values numerically. .SS rev_sort_by .IX Subsection "rev_sort_by" .SS rev_nsort_by .IX Subsection "rev_nsort_by" .Vb 1 \& @vals = rev_sort_by { KEYFUNC } @vals \& \& @vals = rev_nsort_by { KEYFUNC } @vals .Ve .PP \&\fISince version 0.06.\fR .PP Similar to "sort_by" and "nsort_by" but returns the list in the reverse order. Equivalent to .PP .Vb 1 \& @vals = reverse sort_by { KEYFUNC } @vals .Ve .PP except that these functions are slightly more efficient because they avoid the final \f(CW\*(C`reverse\*(C'\fR operation. .SS max_by .IX Subsection "max_by" .Vb 1 \& $optimal = max_by { KEYFUNC } @vals \& \& @optimal = max_by { KEYFUNC } @vals .Ve .PP Returns the (first) value from \f(CW@vals\fR that gives the numerically largest result from the key function. .PP .Vb 1 \& my $tallest = max_by { $_\->height } @people \& \& use File::stat qw( stat ); \& my $newest = max_by { stat($_)\->mtime } @files; .Ve .PP In scalar context, the first maximal value is returned. In list context, a list of all the maximal values is returned. This may be used to obtain positions other than the first, if order is significant. .PP If called on an empty list, an empty list is returned. .PP For symmetry with the "nsort_by" function, this is also provided under the name \f(CW\*(C`nmax_by\*(C'\fR since it behaves numerically. .SS min_by .IX Subsection "min_by" .Vb 1 \& $optimal = min_by { KEYFUNC } @vals \& \& @optimal = min_by { KEYFUNC } @vals .Ve .PP Similar to "max_by" but returns values which give the numerically smallest result from the key function. Also provided as \f(CW\*(C`nmin_by\*(C'\fR .SS minmax_by .IX Subsection "minmax_by" .Vb 1 \& ( $minimal, $maximal ) = minmax_by { KEYFUNC } @vals .Ve .PP \&\fISince version 0.11.\fR .PP Similar to calling both "min_by" and "max_by" with the same key function on the same list. This version is more efficient than calling the two other functions individually, as it has less work to perform overall. In the case of ties, only the first optimal element found in each case is returned. Also provided as \f(CW\*(C`nminmax_by\*(C'\fR. .SS uniq_by .IX Subsection "uniq_by" .Vb 1 \& @vals = uniq_by { KEYFUNC } @vals .Ve .PP Returns a list of the subset of values for which the key function block returns unique values. The first value yielding a particular key is chosen, subsequent values are rejected. .PP .Vb 1 \& my @some_fruit = uniq_by { $_\->colour } @fruit; .Ve .PP To select instead the last value per key, reverse the input list. If the order of the results is significant, don't forget to reverse the result as well: .PP .Vb 1 \& my @some_fruit = reverse uniq_by { $_\->colour } reverse @fruit; .Ve .PP Because the values returned by the key function are used as hash keys, they ought to either be strings, or at least well-behaved as strings (such as numbers, or object references which overload stringification in a suitable manner). .SS partition_by .IX Subsection "partition_by" .Vb 1 \& %parts = partition_by { KEYFUNC } @vals .Ve .PP Returns a key/value list of ARRAY refs containing all the original values distributed according to the result of the key function block. Each value will be an ARRAY ref containing all the values which returned the string from the key function, in their original order. .PP .Vb 1 \& my %balls_by_colour = partition_by { $_\->colour } @balls; .Ve .PP Because the values returned by the key function are used as hash keys, they ought to either be strings, or at least well-behaved as strings (such as numbers, or object references which overload stringification in a suitable manner). .SS count_by .IX Subsection "count_by" .Vb 1 \& %counts = count_by { KEYFUNC } @vals .Ve .PP \&\fISince version 0.07.\fR .PP Returns a key/value list of integers, giving the number of times the key function block returned the key, for each value in the list. .PP .Vb 1 \& my %count_of_balls = count_by { $_\->colour } @balls; .Ve .PP Because the values returned by the key function are used as hash keys, they ought to either be strings, or at least well-behaved as strings (such as numbers, or object references which overload stringification in a suitable manner). .SS zip_by .IX Subsection "zip_by" .Vb 1 \& @vals = zip_by { ITEMFUNC } \e@arr0, \e@arr1, \e@arr2,... .Ve .PP Returns a list of each of the values returned by the function block, when invoked with values from across each each of the given ARRAY references. Each value in the returned list will be the result of the function having been invoked with arguments at that position, from across each of the arrays given. .PP .Vb 1 \& my @transposition = zip_by { [ @_ ] } @matrix; \& \& my @names = zip_by { "$_[1], $_[0]" } \e@firstnames, \e@surnames; \& \& print zip_by { "$_[0] => $_[1]\en" } [ keys %hash ], [ values %hash ]; .Ve .PP If some of the arrays are shorter than others, the function will behave as if they had \f(CW\*(C`undef\*(C'\fR in the trailing positions. The following two lines are equivalent: .PP .Vb 2 \& zip_by { f(@_) } [ 1, 2, 3 ], [ "a", "b" ] \& f( 1, "a" ), f( 2, "b" ), f( 3, undef ) .Ve .PP The item function is called by \f(CW\*(C`map\*(C'\fR, so if it returns a list, the entire list is included in the result. This can be useful for example, for generating a hash from two separate lists of keys and values .PP .Vb 2 \& my %nums = zip_by { @_ } [qw( one two three )], [ 1, 2, 3 ]; \& # %nums = ( one => 1, two => 2, three => 3 ) .Ve .PP (A function having this behaviour is sometimes called \f(CW\*(C`zipWith\*(C'\fR, e.g. in Haskell, but that name would not fit the naming scheme used by this module). .SS unzip_by .IX Subsection "unzip_by" .Vb 1 \& $arr0, $arr1, $arr2, ... = unzip_by { ITEMFUNC } @vals .Ve .PP \&\fISince version 0.09.\fR .PP Returns a list of ARRAY references containing the values returned by the function block, when invoked for each of the values given in the input list. Each of the returned ARRAY references will contain the values returned at that corresponding position by the function block. That is, the first returned ARRAY reference will contain all the values returned in the first position by the function block, the second will contain all the values from the second position, and so on. .PP .Vb 1 \& my ( $firstnames, $lastnames ) = unzip_by { m/^(.*?) (.*)$/ } @names; .Ve .PP If the function returns lists of differing lengths, the result will be padded with \f(CW\*(C`undef\*(C'\fR in the missing elements. .PP This function is an inverse of "zip_by", if given a corresponding inverse function. .SS extract_by .IX Subsection "extract_by" .Vb 1 \& @vals = extract_by { SELECTFUNC } @arr .Ve .PP \&\fISince version 0.05.\fR .PP Removes elements from the referenced array on which the selection function returns true, and returns a list containing those elements. This function is similar to \f(CW\*(C`grep\*(C'\fR, except that it modifies the referenced array to remove the selected values from it, leaving only the unselected ones. .PP .Vb 1 \& my @red_balls = extract_by { $_\->color eq "red" } @balls; \& \& # Now there are no red balls in the @balls array .Ve .PP This function modifies a real array, unlike most of the other functions in this module. Because of this, it requires a real array, not just a list. .PP This function is implemented by invoking \f(CW\*(C`splice\*(C'\fR on the array, not by constructing a new list and assigning it. One result of this is that weak references will not be disturbed. .PP .Vb 1 \& extract_by { !defined $_ } @refs; .Ve .PP will leave weak references weakened in the \f(CW@refs\fR array, whereas .PP .Vb 1 \& @refs = grep { defined $_ } @refs; .Ve .PP will strengthen them all again. .SS extract_first_by .IX Subsection "extract_first_by" .Vb 1 \& $val = extract_first_by { SELECTFUNC } @arr .Ve .PP \&\fISince version 0.10.\fR .PP A hybrid between "extract_by" and \f(CW\*(C`List::Util::first\*(C'\fR. Removes the first element from the referenced array on which the selection function returns true, returning it. .PP As with "extract_by", this function requires a real array and not just a list, and is also implemented using \f(CW\*(C`splice\*(C'\fR so that weak references are not disturbed. .PP If this function fails to find a matching element, it will return an empty list in list context. This allows a caller to distinguish the case between no matching element, and the first matching element being \f(CW\*(C`undef\*(C'\fR. .SS weighted_shuffle_by .IX Subsection "weighted_shuffle_by" .Vb 1 \& @vals = weighted_shuffle_by { WEIGHTFUNC } @vals .Ve .PP \&\fISince version 0.07.\fR .PP Returns the list of values shuffled into a random order. The randomisation is not uniform, but weighted by the value returned by the \f(CW\*(C`WEIGHTFUNC\*(C'\fR. The probability of each item being returned first will be distributed with the distribution of the weights, and so on recursively for the remaining items. .SS bundle_by .IX Subsection "bundle_by" .Vb 1 \& @vals = bundle_by { BLOCKFUNC } $number, @vals .Ve .PP \&\fISince version 0.07.\fR .PP Similar to a regular \f(CW\*(C`map\*(C'\fR functional, returns a list of the values returned by \f(CW\*(C`BLOCKFUNC\*(C'\fR. Values from the input list are given to the block function in bundles of \f(CW$number\fR. .PP If given a list of values whose length does not evenly divide by \f(CW$number\fR, the final call will be passed fewer elements than the others. .SH EXAMPLES .IX Header "EXAMPLES" As many of these functions both take and return lists of values, more useful behaviours can be created by combining them together. .SS "Modal Value" .IX Subsection "Modal Value" To pick the modal value in a list (i.e. the value that appears most frequently according to some key function) combine \f(CW\*(C`count_by\*(C'\fR and \f(CW\*(C`max_by\*(C'\fR: .PP .Vb 2 \& my %counts = count_by { KEYFUNC } ITEMS; \& my $modal = max_by { $counts{$_} } keys %counts; .Ve .PP This can be performed in a single step by use of the \f(CW\*(C`pairs\*(C'\fR function from List::Util: .PP .Vb 1 \& use List::Util qw( pairs ); \& \& my $modal = ( max_by { $_\->value } pairs count_by { KEYFUNC } ITEMS )\->key; .Ve .SH TODO .IX Header "TODO" .IP \(bu 4 XS implementations .Sp These functions are currently all written in pure perl. Some at least, may benefit from having XS implementations to speed up their logic. .IP \(bu 4 Merge into List::Util or List::MoreUtils .Sp This module shouldn't really exist. The functions should instead be part of one of the existing modules that already contain many list utility functions. Having Yet Another List Utilty Module just worsens the problem. .Sp I have attempted to contact the authors of both of the above modules, to no avail; therefore I decided it best to write and release this code here anyway so that it is at least on CPAN. Once there, we can then see how best to merge it into an existing module. .Sp \&\fIUpdated 2015/07/16\fR: As I am now the maintainer of List::Util, some amount of merging/copying should be possible. However, given the latter's key position in the core \fIperl\fR distribution and head of the "CPAN River" I am keen not to do this wholesale, but a selected pick of what seems best, by a popular consensus. .IP \(bu 4 \&\f(CW\*(C`head\*(C'\fR and \f(CW\*(C`tail\*(C'\fR\-like functions .Sp Consider perhaps .Sp .Vb 2 \& head_before { COND } LIST # excludes terminating element \& head_upto { COND } LIST # includes terminating element \& \& tail_since { COND } LIST # includes initiating element \& tail_after { COND } LIST # excludes initiating element .Ve .Sp (See also ). .SH AUTHOR .IX Header "AUTHOR" Paul Evans