NAME
    Complete::Util - General completion routine

VERSION
    This document describes version 0.619 of Complete::Util (from Perl
    distribution Complete-Util), released on 2023-12-26.

DESCRIPTION
    This package provides some generic completion routines that follow the
    Complete convention. (If you are looking for bash/shell tab completion
    routines, take a look at the See Also section.) The main routine is
    "complete_array_elem" which tries to complete a word using choices from
    elements of supplied array. For example:

     complete_array_elem(word => "a", array => ["apple", "apricot", "banana"]);

    The routine will first try a simple substring prefix matching. If that
    fails, will try some other methods like word-mode, character-mode, or
    fuzzy matching. These methods can be disabled using settings.

    There are other utility routines e.g. for converting completion answer
    structure from hash to array/array to hash, combine or modify answer,
    etc. These routines are usually used by the other more specific or
    higher-level completion modules.

FUNCTIONS
  answer_has_entries
    Usage:

     answer_has_entries($answer) -> int

    Check if answer has entries.

    It is equivalent to:

     ref $answer eq 'ARRAY' ? (@$answer ? 1:0) : (@{$answer->{words}} ? 1:0);

    This function is not exported by default, but exportable.

    Arguments ('*' denotes required arguments):

    *   $answer* => *array|hash*

        Completion answer structure.

    Return value: (int)

  answer_num_entries
    Usage:

     answer_num_entries($answer) -> int

    Get the number of entries in an answer.

    It is equivalent to:

     ref $answer eq 'ARRAY' ? (@$answer // 0) : (@{$answer->{words}} // 0);

    This function is not exported by default, but exportable.

    Arguments ('*' denotes required arguments):

    *   $answer* => *array|hash*

        Completion answer structure.

    Return value: (int)

  arrayify_answer
    Usage:

     arrayify_answer($answer) -> array

    Make sure we return completion answer in array form.

    This is the reverse of "hashify_answer". It accepts a hash or an array.
    If it receives a hash, will return its "words" key.

    This function is not exported by default, but exportable.

    Arguments ('*' denotes required arguments):

    *   $answer* => *array|hash*

        Completion answer structure.

    Return value: (array)

  combine_answers
    Usage:

     combine_answers($answers, ...) -> hash

    Given two or more answers, combine them into one.

    This function is useful if you want to provide a completion answer that
    is gathered from multiple sources. For example, say you are providing
    completion for the Perl tool cpanm, which accepts a filename (a tarball
    like "*.tar.gz"), a directory, or a module name. You can do something
    like this:

     combine_answers(
         complete_file(word=>$word),
         complete_module(word=>$word),
     );

    But if a completion answer has a metadata "final" set to true, then that
    answer is used as the final answer without any combining with the other
    answers.

    This function is not exported by default, but exportable.

    Arguments ('*' denotes required arguments):

    *   $answers* => *array[hash|array]*

        (No description)

    Return value: (hash)

    Return a combined completion answer. Words from each input answer will
    be combined, order preserved and duplicates removed. The other keys from
    each answer will be merged.

  complete_array_elem
    Usage:

     complete_array_elem(%args) -> array

    Complete from array.

    Try to find completion from an array of strings. Will attempt several
    methods, from the cheapest and most discriminating to the most expensive
    and least discriminating.

    First method is normal/exact string prefix matching (either
    case-sensitive or insensitive depending on the $Complete::Common::OPT_CI
    variable or the "COMPLETE_OPT_CI" environment variable). If at least one
    match is found, return result. Else, proceed to the next method.

    Word-mode matching (can be disabled by setting
    $Complete::Common::OPT_WORD_MODE or "COMPLETE_OPT_WORD_MODE" environment
    varialbe to false). Word-mode matching is described in Complete::Common.
    If at least one match is found, return result. Else, proceed to the next
    method.

    Prefix char-mode matching (can be disabled by settings
    $Complete::Common::OPT_CHAR_MODE or "COMPLETE_OPT_CHAR_MODE" environment
    variable to false). Prefix char-mode matching is just like char-mode
    matching (see next paragraph) except the first character must match. If
    at least one match is found, return result. Else, proceed to the next
    method.

    Char-mode matching (can be disabled by settings
    $Complete::Common::OPT_CHAR_MODE or "COMPLETE_OPT_CHAR_MODE" environment
    variable to false). Char-mode matching is described in Complete::Common.
    If at least one match is found, return result. Else, proceed to the next
    method.

    Fuzzy matching (can be disabled by setting $Complete::Common::OPT_FUZZY
    or "COMPLETE_OPT_FUZZY" to false). Fuzzy matching is described in
    Complete::Common. If at least one match is found, return result. Else,
    return empty string.

    Will sort the resulting completion list, so you don't have to presort
    the array.

    This function is not exported by default, but exportable.

    Arguments ('*' denotes required arguments):

    *   array* => *array[str]*

        (No description)

    *   exclude => *array*

        (No description)

    *   replace_map => *hash*

        You can supply correction entries in this option. An example is when
        array if "['mount','unmount']" and "umount" is a popular "typo" for
        "unmount". When someone already types "um" it cannot be completed
        into anything (even the current fuzzy mode will return *both* so it
        cannot complete immediately).

        One solution is to add replace_map "{'unmount'=>['umount']}". This
        way, "umount" will be regarded the same as "unmount" and when user
        types "um" it can be completed unambiguously into "unmount".

    *   summaries => *array[str]*

        (No description)

    *   word* => *str* (default: "")

        Word to complete.

    Return value: (array)

  complete_comma_sep
    Usage:

     complete_comma_sep(%args) -> array

    Complete a comma-separated list string.

    This function is not exported by default, but exportable.

    Arguments ('*' denotes required arguments):

    *   elems* => *array[str]*

        (No description)

    *   exclude => *array*

        (No description)

    *   remaining => *code*

        What elements should remain for completion.

        This is a more general mechanism if the "uniq" option does not
        suffice. Suppose you are offering completion for sorting fields. The
        elements are field names as well as field names prefixed with dash
        ("-") to mean sorting with a reverse order. So for example "elems"
        is "["name","-name","age","-age"]". When current word is "name", it
        doesn't make sense to offer "name" nor "-name" again as the next
        sorting field. So we can set "remaining" to this code:

         sub {
             my ($seen_elems, $elems) = @_;
 
             my %seen;
             for (@$seen_elems) {
                 (my $nodash = $_) =~ s/^-//;
                 $seen{$nodash}++;
             }
 
             my @remaining;
             for (@$elems) {
                 (my $nodash = $_) =~ s/^-//;
                 push @remaining, $_ unless $seen{$nodash};
             }
 
             \@remaining;
         }

        As you can see above, the code is given $seen_elems and $elems as
        arguments and is expected to return remaining elements to offer.

    *   replace_map => *hash*

        You can supply correction entries in this option. An example is when
        array if "['mount','unmount']" and "umount" is a popular "typo" for
        "unmount". When someone already types "um" it cannot be completed
        into anything (even the current fuzzy mode will return *both* so it
        cannot complete immediately).

        One solution is to add replace_map "{'unmount'=>['umount']}". This
        way, "umount" will be regarded the same as "unmount" and when user
        types "um" it can be completed unambiguously into "unmount".

    *   sep => *str* (default: ",")

        (No description)

    *   summaries => *array[str]*

        (No description)

    *   uniq => *bool*

        Whether list should contain unique elements.

        When this option is set to true, if the formed list in the current
        word already contains an element, the element will not be offered
        again as completion answer. For example, if "elems" is "[1,2,3,4]"
        and "word" is "2,3," then without "uniq" set to true the completion
        answer is:

         2,3,1
         2,3,2
         2,3,3
         2,3,4

        but with "uniq" set to true, the completion answer becomes:

         2,3,1
         2,3,4

        See also the "remaining" option for a more general mechanism of
        offering fewer elements.

    *   word* => *str* (default: "")

        Word to complete.

    Return value: (array)

  complete_comma_sep_pair
    Usage:

     complete_comma_sep_pair(%args) -> array

    Complete a comma-separated list of key-value pairs.

    This function is not exported by default, but exportable.

    Arguments ('*' denotes required arguments):

    *   complete_value => *code*

        Code to supply possible values for a key.

        Code should accept hash arguments and will be given the arguments
        "word" (word that is part of the value), and "key" (the key being
        evaluated) and is expected to return a completion answer.

    *   keys* => *array[str]*

        (No description)

    *   keys_summaries => *array[str]*

        Summary for each key.

    *   remaining_keys => *code*

        What keys should remain for completion.

        This is a more general mechanism if the "uniq" option does not
        suffice. Suppose you are offering completion for arguments. Possible
        arguments are "foo", "bar", "baz" but the "bar" and "baz" arguments
        are mutually exclusive. We can set "remaining_keys" to this code:

         my %possible_args = {foo=>1, bar=>1, baz=>1};
         sub {
             my ($seen_elems, $elems) = @_;
 
             my %remaining = %possible_args;
             for (@$seen_elems) {
                 delete $remaining{$_};
                 delete $remaining{baz} if $_ eq 'bar';
                 delete $remaining{bar} if $_ eq 'baz';
             }
 
             [keys %remaining];
         }

        As you can see above, the code is given $seen_elems and $elems as
        arguments and is expected to return remaining elements to offer.

    *   uniq => *bool* (default: 1)

        (No description)

    *   word* => *str* (default: "")

        Word to complete.

    Return value: (array)

  complete_hash_key
    Usage:

     complete_hash_key(%args) -> array

    Complete from hash keys.

    This function is not exported by default, but exportable.

    Arguments ('*' denotes required arguments):

    *   hash* => *hash*

        (No description)

    *   summaries => *hash*

        (No description)

    *   summaries_from_hash_values => *true*

        (No description)

    *   word* => *str* (default: "")

        Word to complete.

    Return value: (array)

  complete_hash_value
    Usage:

     complete_hash_value(%args) -> array

    Complete from hash values.

    This function is not exported by default, but exportable.

    Arguments ('*' denotes required arguments):

    *   hash* => *hash*

        (No description)

    *   summaries => *hash*

        (No description)

    *   summaries_from_hash_keys => *true*

        (No description)

    *   word* => *str* (default: "")

        Word to complete.

    Return value: (array)

  get_answer_summaries
    Usage:

     get_answer_summaries($answer) -> array

    Extract just the entry summaries from answer structure.

    This routine accepts a hash or an array answer structure. It then
    returns an arrayref containing just the summaries from each answer
    entry. If the answer is undef, it returns an empty arrayref.

    See also: get_answer_words().

    This function is not exported by default, but exportable.

    Arguments ('*' denotes required arguments):

    *   $answer* => *array|hash*

        Completion answer structure.

    Return value: (array)

  get_answer_words
    Usage:

     get_answer_words($answer) -> array

    Extract just the words from answer structure.

    This routine accepts a hash or an array answer structure. It then
    returns an arrayref containing just the words from each answer entry. If
    the answer is undef, it returns an empty arrayref.

    See also: get_answer_summaries().

    This function is not exported by default, but exportable.

    Arguments ('*' denotes required arguments):

    *   $answer* => *array|hash*

        Completion answer structure.

    Return value: (array)

  hashify_answer
    Usage:

     hashify_answer($answer, $meta) -> hash

    Make sure we return completion answer in hash form.

    This function accepts a hash or an array. If it receives an array, will
    convert the array into `{words=>$ary}' first to make sure the completion
    answer is in hash form.

    Then will add keys from "meta" to the hash.

    This function is not exported by default, but exportable.

    Arguments ('*' denotes required arguments):

    *   $answer* => *array|hash*

        Completion answer structure.

    *   $meta => *hash*

        Metadata (extra keys) for the hash.

    Return value: (hash)

  modify_answer
    Usage:

     modify_answer(%args) -> undef

    Modify answer (add prefix/suffix, etc).

    This function is not exported by default, but exportable.

    Arguments ('*' denotes required arguments):

    *   answer* => *hash|array*

        (No description)

    *   prefix => *str*

        (No description)

    *   suffix => *str*

        (No description)

    Return value: (undef)

FAQ
  Why is fuzzy matching slow?
    Example:

     use Benchmark qw(timethis);
     use Complete::Util qw(complete_array_elem);

     # turn off the other non-exact matching methods
     $Complete::Common::OPT_CI = 0;
     $Complete::Common::OPT_WORD_MODE = 0;
     $Complete::Common::OPT_CHAR_MODE = 0;

     my @ary = ("aaa".."zzy"); # 17575 elems
     timethis(20, sub { complete_array_elem(array=>\@ary, word=>"zzz") });

    results in:

     timethis 20:  7 wallclock secs ( 6.82 usr +  0.00 sys =  6.82 CPU) @  2.93/s (n=20)

    Answer: fuzzy matching is slower than exact matching due to having to
    calculate Levenshtein distance. But if you find fuzzy matching too slow
    using the default pure-perl implementation, you might want to install
    Text::Levenshtein::Flexible (an optional prereq) to speed up fuzzy
    matching. After Text::Levenshtein::Flexible is installed:

     timethis 20:  1 wallclock secs ( 1.04 usr +  0.00 sys =  1.04 CPU) @ 19.23/s (n=20)

ENVIRONMENT
  COMPLETE_UTIL_TRACE
    Bool. If set to true, will generate more log statements for debugging
    (at the trace level).

  COMPLETE_UTIL_LEVENSHTEIN => str ('pp'|'xs'|'flexible')
    Can be used to force which Levenshtein distance implementation to use.
    "pp" means the included PP implementation, which is the slowest (1-2
    orders of magnitude slower than XS implementations), "xs" which means
    Text::Levenshtein::XS, or "flexible" which means
    Text::Levenshtein::Flexible (performs best).

    If this is not set, the default is to use Text::Levenshtein::Flexible
    when it's available, then fallback to the included PP implementation.

HOMEPAGE
    Please visit the project's homepage at
    <https://metacpan.org/release/Complete-Util>.

SOURCE
    Source repository is at
    <https://github.com/perlancar/perl-Complete-Util>.

SEE ALSO
    Complete

    If you want to do bash tab completion with Perl, take a look at
    Complete::Bash or Getopt::Long::Complete or Perinci::CmdLine.

    Other "Complete::*" modules.

    Bencher::Scenarios::CompleteUtil

AUTHOR
    perlancar <perlancar@cpan.org>

CONTRIBUTORS
    *   A. Sinan Unur <nanis@cpan.org>

    *   Steven Haryanto <stevenharyanto@gmail.com>

CONTRIBUTING
    To contribute, you can send patches by email/via RT, or send pull
    requests on GitHub.

    Most of the time, you don't need to build the distribution yourself. You
    can simply modify the code, then test via:

     % prove -l

    If you want to build the distribution (e.g. to try to install it locally
    on your system), you can install Dist::Zilla,
    Dist::Zilla::PluginBundle::Author::PERLANCAR,
    Pod::Weaver::PluginBundle::Author::PERLANCAR, and sometimes one or two
    other Dist::Zilla- and/or Pod::Weaver plugins. Any additional steps
    required beyond that are considered a bug and can be reported to me.

COPYRIGHT AND LICENSE
    This software is copyright (c) 2023, 2022, 2020, 2019, 2017, 2016, 2015,
    2014, 2013 by perlancar <perlancar@cpan.org>.

    This is free software; you can redistribute it and/or modify it under
    the same terms as the Perl 5 programming language system itself.

BUGS
    Please report any bugs or feature requests on the bugtracker website
    <https://rt.cpan.org/Public/Dist/Display.html?Name=Complete-Util>

    When submitting a bug or request, please include a test-file or a patch
    to an existing test-file that illustrates the bug or desired feature.