#============================================================= -*-perl-*-
#
# Template::Manual::VMethods
#
# AUTHOR
#   Andy Wardley  <abw@wardley.org>
#
# COPYRIGHT
#   Copyright (C) 1996-2022 Andy Wardley.  All Rights Reserved.
#
#   This module is free software; you can redistribute it and/or
#   modify it under the same terms as Perl itself.
#
#========================================================================

=head1 NAME

Template::Manual::VMethods - Virtual Methods

=head1 Scalar Virtual Methods

=head2 chunk(size)

Splits the value into a list of chunks of a certain size.

    [% ccard_no = "1234567824683579";
       ccard_no.chunk(4).join
    %]

Output:

    1234 5678 2468 3579

If the size is specified as a negative number then the text will
be chunked from right-to-left.  This gives the correct grouping
for numbers, for example.

    [% number = 1234567;
       number.chunk(-3).join(',')
    %]

Output:

    1,234,567

=head2 collapse

Returns the text with any leading and trailing whitespace removed and any
internal sequences of whitespace converted to a single space

    [% text = "  The bird\n  is the word" %]
    [% text.collapse %]       # The bird is the word

=head2 defined

Returns true if the value is defined.

    [% user = get_user(uid) IF uid.defined %]

=head2 dquote

Returns the text with any double quote characters escaped with a backslash
prefix.  Any newline characters in the text will be replaced with "\n".

    [% quote = 'He said "Oh really?"' %]
    [% quote.dquote %]        # He said \"Oh really?\"

=head2 hash

Return the value as a hash reference containing a single entry with
the key C<value> indicating the original scalar value.  As with the
C<list> virtual method, this is generally used to help massage data
into different formats.

=head2 lcfirst

Returns the text with the first letter converted to lower case.

    [% word = 'BIRD' %]
    [% word.lcfirst %]        # bIRD

=head2 length

Returns the length of the string representation of the item:

    [% IF password.length < 8 %]
       Password too short, dumbass!
    [% END %]

=head2 empty

Returns true if the string is empty:

    [% IF details.empty %]
       No details specified
    [% END %]

=head2 list

Return the value as a single element list.  This can be useful if you
have a variable which may contain a single item or a list and you want
to treat them equally.  The C<list> method can be called against a list
reference and will simply return the original reference, effectively
a no-op.

    [% thing.list.size %]     # thing can be a scalar or a list

=head2 lower

Returns the text in lower case.

    [% word = 'BIRD' %]
    [% word.lower %]          # bird

=head2 match(pattern, global)

Performs a regular expression match on the string using the pattern
passed as an argument.  If the pattern matches the string then the
method returns a reference to a list of any strings captured within
parenthesis in the pattern.

    [% name = 'Larry Wall' %]
    [% matches = name.match('(\w+) (\w+)') %]
    [% matches.1 %], [% matches.0 %]    # Wall, Larry

If the pattern does not match then the method returns false, rather
than returning an empty list which Perl and the Template Toolkit both
consider to be a true value.  This allows you to write expression like
this.

    [% "We're not worthy!" IF name.match('Larry Wall') %]

    [% IF (matches = name.match('(\w+) (\w+)')) %]
       pattern matches: [% matches.join(', ') %]
    [% ELSE %]
       pattern does not match
    [% END %]

Any regex modifiers, like C</s>, should be added in the regex using
the C<(?s)> syntax.  For example, to modify the regex to disregard
whitespace (the C</x> switch), use:

    [% re = '(?x)
               (\w+)
               [ ]
               (\w+)
             ';
      matches = name.match(re);
    %]

To perform a global search to match the pattern as many times as it
appears in the source string, provide a true value for the C<global>
argument following the pattern.

    [% text = 'bandanna';
       text.match('an+', 1).join(', ')      # an, ann
    %]

=head2 repeat(n)

Repeat the string a specified number of times.

    [% name = 'foo' %]
    [% name.repeat(3) %]		# foofoofoo

=head2 replace(search, replace)

Outputs the string with all instances of the first argument (specified
as a Perl regular expression) with the second.

    [% name = 'foo, bar & baz' %]
    [% name.replace('\W+', '_') %]        # foo_bar_baz

You can use C<$1>, C<$2>, etc., to reference captured parts (in parentheses)
in the regular expression.  Just be careful to I<single> quote the replacement
string.  If you use I<double> quotes then TT will try and interpolate the
variables before passing the string to the C<replace> vmethod.

    [% name = 'FooBarBaz' %]
    [% name.replace('([A-Z])', ' $1') %]  # Foo Bar Baz

=head2 remove(pattern)

Outputs the string with all instances of the pattern (specified
as a Perl regular expression) removed.

    [% name = 'foo, bar & baz' %]
    [% name.remove('\W+') %]    # foobarbaz

=head2 search(pattern)

Performs a similar function to L<match> but simply returns true if the
string matches the regular expression pattern passed as an argument.

    [% name = 'foo bar baz' %]
    [% name.search('bar') ? 'bar' : 'no bar' %]	    # bar

This virtual method is now deprecated in favour of L<match>.  Move along
now, there's nothing more to see here.

=head2 size

Always returns 1 for scalar values.  This method is provided for
consistency with the hash and list size methods.

=head2 split(pattern)

Calls Perl's C<split()> function to split a string into a list of
strings.

    [% FOREACH dir IN mypath.split(':') %]
       [% dir %]
    [% END %]

=head2 substr(offset, length, replacement)

Returns a substring starting at C<offset>, for C<length> characters.

    [% str 'foo bar baz wiz waz woz') %]
    [% str.substr(4, 3) %]    # bar

If C<length> is not specified then it returns everything from the
C<offset> to the end of the string.

    [% str.substr(12) %]      # wiz waz woz

If both C<length> and C<replacement> are specified, then the method
replaces everything from C<offset> for C<length> characters with
C<$replacement>.  The substring removed from the string is then returned.

    [% str.substr(0, 11, 'FOO') %]   # foo bar baz
    [% str %]                        # FOO wiz waz woz

=head2 squote

Returns the text with any single quote characters escaped with a backslash
prefix.

    [% tim = "Tim O'Reilly" %]
    [% tim.squote %]          # Tim O\'Reilly

=head2 trim

Returns the text with any leading and trailing whitespace removed.

    [% text = '  hello  world  ' %]
    [% text.trim %]           # hello  world

=head2 ucfirst

Returns the text with the first letter converted to upper case.

    [% word = 'bird' %]
    [% word.ucfirst %]        # Bird

=head2 upper

Returns the text in upper case.

    [% word = 'bird' %]
    [% word.upper %]          # BIRD

=head1 Hash Virtual Methods

=head2 keys

Returns a list of keys in the hash.  They are not returned in any
particular order, but the order is the same as for the corresponding
values method.

    [% FOREACH key IN hash.keys %]
       * [% key %]
    [% END %]

If you want the keys in sorted order, use the list C<sort> method.

    [% FOREACH key IN hash.keys.sort %]
       * [% key %]
    [% END %]

Having got the keys in sorted order, you can then use variable
interpolation to fetch the value.  This is shown in the following
example by the use of C<$key> to fetch the item from C<hash> whose
key is stored in the C<key> variable.

    [% FOREACH key IN hash.keys.sort %]
       * [% key %] = [% hash.$key %]
    [% END %]

Alternately, you can use the C<pairs> method to get a list of
key/value pairs in sorted order.

=head2 values

Returns a list of the values in the hash.  As with the C<keys> method,
they are not returned in any particular order, although it is the same
order that the keys are returned in.

    [% hash.values.join(', ') %]

=head2 items

Returns a list of both the keys and the values expanded into a single list.

    [% hash = {
          a = 10
          b = 20
       };

       hash.items.join(', ')    # a, 10, b, 20
    %]

=head2 each

This method currently returns the same thing as the C<items> method.

However, please note that this method will change in the next major
version of the Template Toolkit (v3) to return the same thing as the
C<pairs> method.  This will be done in an effort to make these virtual
method more consistent with each other and how Perl works.

In anticipation of this, we recommend that you stop using C<hash.each>
and instead use C<hash.items>.

=head2 pairs

This method returns a list of key/value pairs.  They are returned in
sorted order according to the keys.

    [% FOREACH pair IN product.pairs %]
       * [% pair.key %] is [% pair.value %]
    [% END %]

=head2 list

Returns the contents of the hash in list form.  An argument can be
passed to indicate the desired items required in the list: C<keys> to
return a list of the keys (same as C<hash.keys>), C<values> to return a
list of the values (same as C<hash.values>), C<each> to return as list
of key and values (same as C<hash.each>), or C<pairs> to return a list
of key/value pairs (same as C<hash.pairs>).

    [% keys   = hash.list('keys') %]
    [% values = hash.list('values') %]
    [% items  = hash.list('each') %]
    [% pairs  = hash.list('pairs') %]

When called without an argument it currently returns the same thing as
the C<pairs> method.  However, please note that this method will change
in the next major version of the Template Toolkit (v3) to return a
reference to a list containing the single hash reference (as per the
scalar list method).

In anticipation of this, we recommend that you stop using C<hash.list>
and instead use C<hash.pairs>.

=head2 sort, nsort

Return a list of the keys, sorted alphabetically (C<sort>) or numerically
(C<nsort>) according to the corresponding values in the hash.

    [% FOREACH n IN phones.sort %]
       [% phones.$n %] is [% n %],
    [% END %]

=head2 import

The C<import> method can be called on a hash array to import the contents
of another hash array.

    [% hash1 = {
         foo = 'Foo'
         bar = 'Bar'
       }
       hash2 = {
           wiz = 'Wiz'
           woz = 'Woz'
       }
    %]

    [% hash1.import(hash2) %]
    [% hash1.wiz %]             # Wiz

You can also call the C<import()> method by itself to import a hash array
into the current namespace hash.

    [% user = { id => 'lwall', name => 'Larry Wall' } %]
    [% import(user) %]
    [% id %]: [% name %]        # lwall: Larry Wall

=head2 defined, exists

Returns a true or false value if an item in the hash denoted by the key
passed as an argument is defined or exists, respectively.

    [% hash.defined('somekey') ? 'yes' : 'no' %]
    [% hash.exists('somekey') ? 'yes' : 'no' %]

When called without any argument, C<hash.defined> returns true if the hash
itself is defined (e.g. the same effect as C<scalar.defined>).

=head2 delete

Delete one or more items from the hash.

    [% hash.delete('foo', 'bar') %]

=head2 size

Returns the number of key/value pairs in the hash.

=head2 empty

Returns true if the hash is empty:

    [% IF config.empty %]
       No configuration available
    [% END %]

=head2 item

Returns an item from the hash using a key passed as an argument.

    [% hash.item('foo') %]  # same as hash.foo

=head1 List Virtual Methods

=head2 first, last

Returns the first/last item in the list.  The item is not removed from the
list.

    [% results.first %] to [% results.last %]

If either is given a numeric argument C<n>, they return the first or
last C<n> elements:

    The first 5 results are [% results.first(5).join(", ") %].

=head2 size, max

Returns the size of a list (number of elements) and the maximum
index number (size - 1), respectively.

    [% results.size %] search results matched your query

=head2 empty

Returns true if the list is empty:

    [% IF results.empty %]
       No results found
    [% END %]

=head2 defined

Returns a true or false value if the item in the list denoted by the
argument is defined.

    [% list.defined(3) ? 'yes' : 'no' %]

When called without any argument, C<list.defined> returns true if the list
itself is defined (e.g. the same effect as C<scalar.defined>).

=head2 reverse

Returns the items of the list in reverse order.

    [% FOREACH s IN scores.reverse %]
       ...
    [% END %]

=head2 join

Joins the items in the list into a single string, using Perl's C<join()>
function.

    [% items.join(', ') %]

=head2 grep

Returns a list of the items in the list that match a regular expression
pattern.

    [% FOREACH directory.files.grep('\.txt$') %]
       ...
    [% END %]

=head2 sort, nsort

Returns the items in alpha (C<sort>) or numerical (C<nsort>) order.

    [% library = books.sort %]

An argument can be provided to specify a search key.  Where an item in
the list is a hash reference, the search key will be used to retrieve a
value from the hash which will then be used as the comparison value.
Where an item is an object which implements a method of that name, the
method will be called to return a comparison value.

    [% library = books.sort('author') %]

In the example, the C<books> list can contains hash references with
an C<author> key or objects with an C<author> method.

You can also specify multiple sort keys.

    [% library = books.sort('author', 'title') %]

In this case the books will be sorted primarily by author.  If two or more
books have authors with the same name then they will be sorted by title.

=head2 unshift(item), push(item)

The C<push()> method adds an item or items to the end of list.

    [% mylist.push(foo) %]
    [% mylist.push(foo, bar) %]

The C<unshift()> method adds an item or items to the start of a list.

    [% mylist.unshift(foo) %]
    [% mylist.push(foo, bar)    %]

=head2 shift, pop

Removes the first/last item from the list and returns it.

    [% first = mylist.shift %]
    [% last  = mylist.pop   %]

=head2 unique

Returns a list of the unique elements in a list, in the same order
as in the list itself.

    [% mylist = [ 1, 2, 3, 2, 3, 4, 1, 4, 3, 4, 5 ] %]
    [% numbers = mylist.unique %]

While this can be explicitly sorted, it is not required that the list
be sorted before the unique elements are pulled out (unlike the Unix
command line utility).

    [% numbers = mylist.unique.sort %]

=head2 import

Appends the contents of one or more other lists to the end of the
current list.

    [% one   = [ 1 2 3 ];
       two   = [ 4 5 6 ];
       three = [ 7 8 9 ];
       one.import(two, three);
       one.join(', ');     # 1, 2, 3, 4, 5, 6, 7, 8, 9
    %]

Import also allows chaining. The below syntax is equivalent.

    [% one   = [ 1 2 3 ];
       two   = [ 4 5 6 ];
       three = [ 7 8 9 ];
       one.import(two, three).join(', ');        # 1, 2, 3, 4, 5, 6, 7, 8, 9
# or:  one.import(two).import(three).join(', '); # 1, 2, 3, 4, 5, 6, 7, 8, 9
    %]

=head2 merge

Returns a list composed of zero or more other lists:

    [% list_one = [ 1 2 3 ];
       list_two = [ 4 5 6 ];
       list_three = [ 7 8 9 ];
       list_four = list_one.merge(list_two, list_three);
    %]

The original lists are not modified.

=head2 slice(from, to)

Returns a slice of items in the list between the bounds passed as
arguments.  If the second argument, C<to>, isn't specified, then it
defaults to the last item in the list.  The original list is not
modified.

    [% first_three = list.slice(0,2) %]
    [% last_three  = list.slice(-3, -1) %]

=head2 splice(offset, length, list)

Behaves just like Perl's C<splice()> function allowing you to selectively
remove and/or replace elements in a list.  It removes C<length> items
from the list, starting at C<offset> and replaces them with the items
in C<list>.

    [% play_game = [ 'play', 'scrabble' ];
       ping_pong = [ 'ping', 'pong' ];
       redundant = play_game.splice(1, 1, ping_pong);
       redundant.join;     # scrabble
       play_game.join;     # play ping pong
    %]

The method returns a list of the items removed by the splice.
You can use the C<CALL> directive to ignore the output if you're
not planning to do anything with it.

    [% CALL play_game.splice(1, 1, ping_pong) %]

As well as providing a reference to a list of replacement values,
you can pass in a list of items.

    [% CALL list.splice(-1, 0, 'foo', 'bar') %]

Be careful about passing just one item in as a replacement value.
If it is a reference to a list then the contents of the list will
be used.  If it's not a list, then it will be treated as a single
value.  You can use square brackets around a single item if you
need to be explicit:

    [% # push a single item, an_item
       CALL list.splice(-1, 0, an_item);

       # push the items from another_list
       CALL list.splice(-1, 0, another_list);

       # push a reference to another_list
       CALL list.splice(-1, 0, [ another_list ]);
    %]

=head2 hash

Returns a reference to a hash array comprised of the elements in the
list.  The even-numbered elements (0, 2, 4, etc) become the keys and
the odd-numbered elements (1, 3, 5, etc) the values.

    [% list = ['pi', 3.14, 'e', 2.718] %]
    [% hash = list.hash %]
    [% hash.pi %]               # 3.14
    [% hash.e  %]               # 2.718

If a numerical argument is provided then the hash returned will have
keys generated for each item starting at the number specified.

    [% list = ['beer', 'peanuts'] %]
    [% hash = list.hash(1) %]
    [% hash.1  %]               # beer
    [% hash.2  %]               # peanuts

=head2 item

Returns an item from the list using an index passed as an argument.

    [% list.item(0) %]  # same as list.0

=head1 Automagic Promotion of Scalar to List for Virtual Methods

In addition to the scalar virtual methods listed in the previous
section, you can also call any list virtual method against a scalar.
The item will be automagically promoted to a single element list and
the appropriate list virtual method will be called.

One particular benefit of this comes when calling subroutines or
object methods that return a list of items, rather than the
preferred reference to a list of items.  In this case, the
Template Toolkit automatically folds the items returned into
a list.

The upshot is that you can continue to use existing Perl modules or
code that returns lists of items, without having to refactor it
just to keep the Template Toolkit happy (by returning references
to list).  C<Class::DBI> module is just one example of a particularly
useful module which returns values this way.

If only a single item is returned from a subroutine then the
Template Toolkit assumes it meant to return a single item (rather
than a list of 1 item) and leaves it well alone, returning the
single value as it is.  If you're executing a database query,
for example, you might get 1 item returned, or perhaps many
items which are then folded into a list.

The C<FOREACH> directive will happily accept either a list or a single item
which it will treat as a list. So it's safe to write directives like this,
where we assume that the C<something> variable is bound to a subroutine which
may return one or more items:

    [% FOREACH item IN something %]
       ...
    [% END %]

The automagic promotion of scalars to single item lists means
that you can also use list virtual methods safely, even if you
only get one item returned.  For example:

    [% something.first   %]
    [% something.join    %]
    [% something.reverse.join(', ') %]

Note that this is very much a last-ditch behaviour.  If the single
item return is an object with a C<first> method, for example, then that
will be called, as expected, in preference to the list virtual method.

=head1 Defining Custom Virtual Methods

You can define your own virtual methods for scalars, lists and hash arrays.
The L<Template::Stash> package variables C<$SCALAR_OPS>, C<$LIST_OPS> and
C<$HASH_OPS> are references to hash arrays that define these virtual methods.
C<HASH_OPS> and C<LIST_OPS> methods are subroutines that accept a hash/list
reference as the first item. C<SCALAR_OPS> are subroutines that accept a scalar
value as the first item. Any other arguments specified when the method is
called will be passed to the subroutine.

    # load Template::Stash to make method tables visible
    use Template::Stash;

    # define list method to return new list of odd numbers only
    $Template::Stash::LIST_OPS->{ odd } = sub {
        my $list = shift;
        return [ grep { $_ % 2 } @$list ];
    };

Example template:

    [% primes = [ 2, 3, 5, 7, 9 ] %]
    [% primes.odd.join(', ') %]		# 3, 5, 7, 9

TODO: document the define_vmethod() method which makes this even easier

=cut

# Local Variables:
# mode: perl
# perl-indent-level: 4
# indent-tabs-mode: nil
# End:
#
# vim: expandtab shiftwidth=4: