Hash.pod6

=begin pod :kind("Type") :subkind("class") :category("composite")

=TITLE class Hash

=SUBTITLE Mapping from strings to itemized values

    class Hash is Map { }

A C<Hash> is a mutable L<Map|/type/Map>; it implements C<Associative> through
its inheritance of C<Map> and as such provides support for looking up values
using keys, providing support for L<associative
subscripting|/language/subscripts#Methods_to_implement_for_associative_subscripting>.

Although the order of the hashes is guaranteed to be random in every single
call, still successive calls to C<.keys> and C<.values> are guaranteed to return
them in the same order:

    my %orig = :1a, :2b; my %new = :5b, :6c;
    %orig{ %new.keys } = %new.values;
    say %orig.perl; # OUTPUT: «{:a(1), :b(5), :c(6)}␤»

In this case, C<b> will always be associated to 5 and C<c> to 6; even if two
successive calls to C<keys> will return them in different order. Successive
calls to any of them separately and repeatedly will always return the same order
in any program invocation.

Please see the section on L<hash literals|/language/syntax#Hash_literals> for
different ways to declare a hash. Additionally, they can be declared using curly
braces as long as these rules are followed:

=item Empty curly braces will always declare an empty hash.
=item A reference to $_ (even implicit) will instead declare a block.
=item A C<Pair> or variable with C<%> as the first element will declare a hash.

=for code
given 3 { say WHAT {3 => 4, :b}  };     # OUTPUT: «(Hash)␤»
given 3 { say WHAT {3 => 4, :b($_)} };  # OUTPUT: «(Block)␤»
given 3 { say WHAT {3 => 4, :b(.Num)} };# OUTPUT: «(Block)␤»
say { 'a',:b(3), 'c' }.^name;           # OUTPUT: «Block␤»

The next-to-last two cases are examples of the generation of C<Block>s in the
presence of the topic variable C<$_>. The last case does not meet the third
criterium for generating a hash, and thus generates a C<Block>.

A % in front of parentheses or square brackets will generate a Hash as long as
the elements can be paired.

    say %( 'a', 3, :b(3), 'c', 3 ).^name; # OUTPUT: «Hash␤»

Elements in this hash can be paired both sides of the Pair C<:b(3)>.

    say %(«a b c 1 2 3»).^name;           # OUTPUT: «Hash␤»

An empty hash can be initialized either with empty curly braces or, since 6.d, C<%()>.

    say %().^name; # OUTPUT: «Hash␤»
    say {}.^name;  # OUTPUT: «Hash␤»

=head1 Methods

=head2 method classify-list

Defined as:

    multi method classify-list(&mapper, *@list, :&as --> Hash:D)
    multi method classify-list(%mapper, *@list, :&as --> Hash:D)
    multi method classify-list(@mapper, *@list, :&as --> Hash:D)

Populates a L«C<Hash>|/type/Hash» by classifying the possibly-empty C<@list> of
values using the given C<mapper>, optionally altering the values using the
C<:&as> L«C<Callable>|/type/Callable». The C<@list> cannot be lazy.

The mapper can be a L«C<Callable>|/type/Callable» that takes a single argument,
an L«C<Associative>|/type/Associative», or an L«C<Iterable>|/type/Iterable»;
this C<Callable> is guaranteed to be called only once per item.
With L«C<Associative>|/type/Associative» and an L«C<Iterable>|/type/Iterable»
mappers, the values in the C<@list> represent the key and index of the mapper's
value respectively. A L«C<Callable>|/type/Callable» mapper will be executed
once per each item in the C<@list>, with that item as the argument and its
return value will be used as the mapper's value.

=head3 Simple classification

In simple classification mode, each mapper's value is any non-C<Iterable> and
represents a key to classify C<@list>'s item under:

=begin code
say % .classify-list: { $_ %% 2 ?? 'even' !! 'odd' }, ^10;
# OUTPUT: «{even => [0 2 4 6 8], odd => [1 3 5 7 9]}␤»

my @mapper = <zero one two three four five>;
my %hash = foo => 'bar';
say %hash.classify-list: @mapper, 1, 2, 3, 4, 4;
# OUTPUT: «{foo => bar, four => [4 4], one => [1], three => [3], two => [2]}␤»
=end code

The mapper's value is used as the key of the L«C<Hash>|/type/Hash» to
which the C<@list>'s item will be L«C<push>ed|/routine/push». See
L«C<.categorize-list>|/routine/categorize-list» if you wish to classify an item
into multiple categories at once.

=head3 Multi-level classification

In multi-level classification mode, each mapper's value is an
L«C<Iterable>|/type/Iterable» that represents a tree of hash keys to classify
C<@list>'s item under:

    say % .classify-list: {
        [
            (.is-prime ?? 'prime' !! 'non-prime'),
            ($_ %% 2   ?? 'even'  !! 'odd'      ),
        ]
    }, ^10;
    # OUTPUT:
    # {
    #     non-prime => {
    #         even => [0 4 6 8],
    #         odd  => [1 9]
    #     },
    #     prime => {
    #         even => [2],
    #         odd  => [3 5 7]
    #     }
    # }

In the case we are using C<Iterables> and not C<Callables>, each of those
L«C<Iterable>|/type/Iterable»s must
have the same number
of elements, or the method will throw an exception. This restriction exists to
avoid conflicts when the same key is a leaf of one value's classification but a
node of another value's classification.

=for code
my @mapper = [['1a','1b','1c'],['2a','2b','2c'],['3a','3b','3c']];
say % .classify-list: @mapper, 1,2,1,1,2,0;
# OUTPUT: «{1a => {1b => {1c => [0]}}, 2a => {2b => {2c => [1 1 1]}}, 3a => {3b => {3c => [2 2]}}}␤»

Every element of the array represents a different level in the tree, with the
elements of the list that is being mapped used as index, and the elements of
the mapper array used as keys to the different levels. So C<0> selects the
first sub-array and then the subsequent levels are built by running over the
rest of the elements of that sub-array.

=for code
my @mapper = [['1a','1b'],['2a','2b'],['3a','3b']];
say % .classify-list: @mapper, 1,0,1,1,1,0,2;
# OUTPUT: «{1a => {1b => [0 0]}, 2a => {2b => [1 1 1 1]}, 3a => {3b => [2]}}␤»

From version 6.d, trying to use C<Iterable>s of different size will throw an
error:

=for code :skip-test<Illustrates error>
my @mapper = [<1a 1b>, <2a 2b 2fail>];
say % .classify-list: @mapper, 1,0,1,1,1,0;
# OUTPUT: «mapper on classify-list computed to an item with different number
# of elements in it than previous items, which cannot be used because all
# values need to have the same number of elements. Mixed-level classification
# is not supported.␤  in block <unit>…»

=head3 C<:&as> value modifier

If C<:&as> L«C<Callable>|/type/Callable» argument is specified, it will be
called once per each item of C<@list>, with the value as the argument, and
its return value will be used instead of the original C<@list>'s item:

    say % .classify-list: :as{"Value is $_"}, { $_ %% 2 ?? 'even' !! 'odd' }, ^5;
    # OUTPUT (slightly altered manually, for clarity):
    # {
    #     even => ['Value is 0', 'Value is 2', 'Value is 4'],
    #     odd  => ['Value is 1', 'Value is 3']
    # }

=head2 method categorize-list

Defined as:

    multi method categorize-list(&mapper, *@list, :&as --> Hash:D)
    multi method categorize-list(%mapper, *@list, :&as --> Hash:D)
    multi method categorize-list(@mapper, *@list, :&as --> Hash:D)

Populates a L«C<Hash>|/type/Hash» by classifying the
possibly-empty C<@list> of values using the given C<mapper>, optionally
altering the values using the C<:&as> L«C<Callable>|/type/Callable». The
C<@list> cannot be lazy.

The mapper can be a L«C<Callable>|/type/Callable» that takes a single argument,
an L«C<Associative>|/type/Associative», or an L«C<Iterable>|/type/Iterable».
With L«C<Associative>|/type/Associative» and an L«C<Iterable>|/type/Iterable»
mappers, the values in the C<@list> represent the key and index of the mapper's
value respectively. A L«C<Callable>|/type/Callable» mapper will be executed
once per each item in the C<@list>, with that item as the argument and its
return value will be used as the mapper's value.

=head3 Simple categorization

The mapper's value is expected to be a possibly empty list of
non-L«C<Iterables>|/type/Iterable» that represent categories to place the value
into:

    say % .categorize-list: {
        gather {
            take 'prime'   if .is-prime;
            take 'largish' if $_ > 5;
            take $_ %% 2 ?? 'even' !! 'odd';
        }
    }, ^10;

    # OUTPUT:
    # {
    #     prime   => [2 3 5 7]
    #     even    => [0 2 4 6 8],
    #     odd     => [1 3 5 7 9],
    #     largish => [6 7 8 9],
    # }

Notice how some items, e.g. C<6> and C<7>, are present in several categories.

=head3 Multi-level categorization

In multi-level categorization, the categories produced by the mapper are
L<Iterables|/type/Iterable> and categorization combines features
of L<classify|/routine/classify>, by producing nested hashes of classifications
for each category.

    say % .categorize-list: {
        [
            $_ > 5    ?? 'largish' !! 'smallish',
            .is-prime ?? 'prime'   !! 'non-prime',
        ],
    }, ^10;

    # OUTPUT:
    # {
    #     largish => {
    #         non-prime => [6 8 9],
    #         prime     => [7]
    #     },
    #     smallish => {
    #         non-prime => [0 1 4],
    #         prime     => [2 3 5]
    #     }
    # }

The mapper in the snippet above produces a single-item list (note the
significant
trailing comma) with a two-item C<Array> in it. The first item in that array
indicates the first level of classification: the C<largish>/C<smallish>
categories the routine produces. The second item in that array indicates
further levels of classification, in our case the classification into
C<prime>/C<non-prime> inside of each category.

B<NOTE:>: every L«C<Iterables>|/type/Iterable» category
must have the same number of elements, or the method will throw an exception.
This restriction exists to avoid conflicts when the same key is a
leaf of one value's classification but a node of another value's classification.

=head3 C<:&as> value modifier

If C<:&as> L«C<Callable>|/type/Callable» argument is specified, it will be
called once per each item of C<@list>, with the value as the argument, and
its return value will be used instead of the original C<@list>'s item:

    say % .categorize-list: :as{"Value is $_"}, { $_ %% 2 ?? 'even' !! 'odd' }, ^5;
    # OUTPUT (slightly altered manually, for clarity):
    # {
    #     even => ['Value is 0', 'Value is 2', 'Value is 4'],
    #     odd  => ['Value is 1', 'Value is 3']
    # }

=head2 method push

Defined as:

    multi method push(Hash:D: *@new)

Adds the C<@new> elements to the hash with the same semantics as hash
assignment, but with three exceptions:

=item The hash isn't emptied first, i.e. old pairs are not deleted.

=item If a key already exists in the hash, and the corresponding value is an
L<Array|/type/Array>, the new value is pushed onto the array (instead of replacing it).

=item If a key already exists in the hash, and the corresponding value is not
an L<Array|/type/Array>, old and new value are both placed into an array in the place
of the old value.

Example:

    my %h  = a => 1;
    %h.push: (a => 1);              # a => [1,1]
    %h.push: (a => 1) xx 3 ;        # a => [1,1,1,1,1]
    %h.push: (b => 3);              # a => [1,1,1,1,1], b => 3
    %h.push('c' => 4);              # a => [1,1,1,1,1], b => 3, c => 4
    push %h, 'd' => 5;              # a => [1,1,1,1,1], b => 3, c => 4, d => 5

Please note that C<Pair>s or
L<colon pairs|/language/glossary#index-entry-Colon_Pair> as arguments to push
will be treated as extra named arguments and as such wont end up the C<Hash>.
The same applies to the sub C<push>.

    my %h .= push(e => 6);
    push %h, f => 7;
    say %h.perl;
    # OUTPUT: «{}␤»

Also note that push can be used as a replacement for assignment during hash
initialization very useful ways. Take for instance the case of an inverted
index:

    my %wc = 'hash' => 323, 'pair' => 322, 'pipe' => 323;
    (my %inv).push: %wc.invert;
    say %inv;                     # OUTPUT: «{322 => pair, 323 => [pipe hash]}␤»

Note that such an initialization could also be written as

    my %wc = 'hash' => 323, 'pair' => 322, 'pipe' => 323;
    my %inv .= push: %wc.invert;

B<Note:> Compared to L«C<append>|/routine/append», C<push> will add the given
value as is, whereas C<append> will L«C<slip>|/routine/slip» it in:

    my %ha = :a[42, ]; %ha.push: "a" => <a b c a>;
    say %ha; # OUTPUT: «{a => [42 (a b c a)]}␤»

    my %hb = :a[42, ]; %hb.append: "a" => <a b c a>;
    say %hb; # OUTPUT: «{a => [42 a b c a]}␤»

=head2 method append

Defined as:

    method append(+@values)

Append the provided Pairs or even sized list to the Hash. If a key already
exists, turn the existing value into an L<Array|/type/Array> and push new value
onto that C<Array>. Please note that you can't mix even sized lists and lists
of Pairs. Also, bare C<Pair>s or colon pairs will be treated as L<named
arguments|/type/Signature#Positional_vs._Named> to C<.append>.

    my %h = a => 1;
    %h.append('b', 2, 'c', 3);
    %h.append( %(d => 4) );
    say %h;
    # OUTPUT: «{a => 1, b => 2, c => 3, d => 4}␤»
    %h.append('a', 2);
    # OUTPUT: «{{a => [1 2], b => 2, c => 3, d => 4}␤»

B<Note:> Compared to L«C<push>|/routine/push», C<append> will
L«C<slip>|/routine/slip» in the given value, whereas C<push> will add it as
is:

    my %hb = :a[42, ]; %hb.append: "a" => <a b c a>;
    say %hb; # OUTPUT: «{a => [42 a b c a]}␤»

    my %ha = :a[42, ]; %ha.push: "a" => <a b c a>;
    say %ha; # OUTPUT: «{a => [42 (a b c a)]}␤»

=head2 method default

Defined as:

    method default()

Returns the default value of the invocant, i.e. the value which is returned when
a non existing key is used to access an element in the C<Hash>. Unless the
C<Hash> is declared as having a default value by using the
L<is default|/syntax/trait is default> trait the method returns the type object
C<(Any)>.

    my %h1 = 'apples' => 3, 'oranges' => 7;
    say %h1.default;                                       # OUTPUT: «(Any)␤»
    say %h1{'bananas'};                                    # OUTPUT: «(Any)␤»

    my %h2 is default(1) = 'apples' => 3, 'oranges' => 7;
    say %h2.default;                                       # OUTPUT: «1␤»
    say %h2{'apples'} + %h2{'bananas'};                    # OUTPUT: «4␤»

=head2 method keyof

Defined as:

    method keyof()

Returns the type constraint for the keys of the invocant. For
normal hashes the method returns the coercion type C<(Str(Any))>
while for L<non-string keys|/type/Hash#Non-string_keys>
hashes the type used in the declaration of the C<Hash> is returned.

    my %h1 = 'apples' => 3, 'oranges' => 7;  # (no key type specified)
    say %h1.keyof;                           # OUTPUT: «(Str(Any))␤»

    my %h2{Str} = 'oranges' => 7;            # (keys must be of type Str)
    say %h2.keyof;                           # (Str)
    %h2{3} = 'apples';                       # throws exception
    CATCH { default { put .^name, ': ', .Str } };
    # OUTPUT: «X::TypeCheck::Binding: Type check failed in binding to key; expected Str but got Int (3)␤»

    my %h3{Int};                             # (this time, keys must be of type Int)
    %h3{42} = 4096;
    say %h3.keyof;                           # (Int)

=head2 method of

Defined as:

    method of()

Returns the type constraint for the values of the invocant. By default,
i.e., if no type constraint is given during declaration, the method
returns C<(Mu)>.

    my %h1 = 'apples' => 3, 'oranges' => 7;  # (no type constraint specified)
    say %h1.of;                              # OUTPUT: «(Mu)␤»

    my Int %h2 = 'oranges' => 7;             # (values must be of type Int)
    say %h2.of;                              # OUTPUT: «(Int)␤»

=head2 routine dynamic

Defined as:

    method dynamic(--> Bool:D)

Returns C<True> if the invocant has been declared with the
L<is dynamic|/routine/is dynamic> trait.

    my %a;
    say %a.dynamic;                          # OUTPUT: «False␤»

    my %b is dynamic;
    say %b.dynamic;                          # OUTPUT: «True␤»

If you declare a variable with the C<*> twigil C<is dynamic> is implied.

    my %*b;
    say %*b.dynamic;                         # OUTPUT: «True␤»

Note that in the L<Scalar|/type/Scalar> case you have to use the C<VAR> method
in order to get correct information.

    my $s is dynamic = %('apples' => 5);
    say $s.dynamic;                   # OUTPUT: «False␤»  (wrong, don't do this)
    say $s.VAR.dynamic;               # OUTPUT: «True␤»   (correct approach)

=head1 Subscript Adverbs

Some methods are implemented as adverbs on subscripts
(consult the L<operators|/language/operators#postcircumfix_{_}> documentation
for more information).

=head2 C<:exists>

The adverb C<:exists> returns C<Bool::True> if a key exists in the Hash. If more
than one key is supplied it returns a C<List> of C<Bool>.

    my %h = a => 1, b => 2;
    say %h<a>:exists;   # OUTPUT: «True␤»
    say %h<a b>:exists; # OUTPUT: «(True True)␤»

=head2 C<:delete>

Use C<:delete> to remove a C<Pair> from the C<Hash>.

    my %h = a => 1;
    say %h;         # OUTPUT: «{a => 1}␤»
    say %h.elems;   # OUTPUT: «1␤»

    %h<a>:delete;
    say %h;         # OUTPUT: «{}␤»
    say %h.elems;   # OUTPUT: «0␤»

=head2 C<:p>

The adverb C<:p> returns a C<Pair> or a List of C<Pair> instead of just the
value.

    my %h = a => 1, b => 2;
    say %h<a>:p;    # OUTPUT: «a => 1␤»
    say %h<a b>:p;  # OUTPUT: «(a => 1 b=> 2)␤»

=head2 C<:v> and C<:k>

The adverbs C<:v> and C<:k> return the key or value or a list thereof.

    my %h = a => 1, b => 2;
    say %h<a>:k;    # OUTPUT: «a␤»
    say %h<a b>:k;  # OUTPUT: «(a b)␤»

The adverb C<:kv> returns a list of keys and values.

    my %h = a => 1, b => 2, c => 3;
    say %h<a c>:kv;  # OUTPUT: «(a 1 c 3)␤»

You can also use the adverbs without knowing anything about the hash by using
empty angle brackets in which case all the keys and values will be listed:

    my %h1 = a => 1;
    my %h2 = a => 1, b => 2;
    say %h1<>:k; # OUTPUT: «(a)␤»
    say %h1<>:v; # OUTPUT: «(1)␤»
    say %h2<>:k; # OUTPUT: «(a b)␤»
    say %h2<>:v; # OUTPUT: «(1 2)␤»

=end pod

# vim: expandtab softtabstop=4 shiftwidth=4 ft=perl6