=encoding utf8

=head1 NAME

DBIx::Custom::Guide - DBIx::Custom Guide

=head1 GUIDE

(This module contain many experimental featrues.

 in each pod of module, experimental marked method

 is maybe changed without warnings)

L<DBIx::Custom> is the class to make easy to execute SQL.

This is L<DBI> wrapper class like L<DBIx::Class> or L<DBIx::Simple>.

You can do thing more easy than L<DBIx::Class>, more flexible

than L<DBIx::Simple>.

L<DBIx::Custom> is not O/R mapper, O/R mapper is usefule, but

you must learn many things. Created SQL is sometimes inefficient,

and in many cases you create raw SQL because

O/R mapper can't make complex SQL

L<DBIx::Custom> is opposit of O/R mapper.

The main purpose is that we respect SQL

and make easy difficult works if you use only L<DBI>.

If you already learn SQL, it is easy to use L<DBIx::Custom>.

I explain L<DBIx::Custom> a little in this section.

In L<DBIx::Custom>, you embbed tag in SQL.

    select * from book where {= title} and {=author};

The part arround {} is tag.

This SQL is converted to the one which contains place holder.

    select * from book where title = ? and author = ?;

Maybe you ask me that this conversion is meaningful.

On the top of this, usuful features is implemented.

See the following descriptions.

=over 4

=item 1. Specify place holder binding value as hash refernce

If you use L<DBI>, you must specify place holder binding value

as array.

    $sth->execute(@bind);

If you use L<DBIx::Custom>, you specify it as hash reference.

    my $param = {title => 'Perl', author => 'Ken'};

    $dbi->execute($sql, $param);

=item 2. Filtering

L<DBIx::Custom> provides filtering system.

For example, You think that about date value you want to 

manipulate it as date object like L<Time::Piece> in Perl,

and want to convert it to database DATE format.

and want to do reverse.

You can use filtering system.

At first, register filter.

    $dbi->register_filter(

        tp_to_date => sub {

            ...

        },

        date_to_tp => sub {

            ...

        }

    );

next, apply this filter to each column.

    $dbi->apply_filter('book',

        'issue_date' => {out => 'tp_to_date', in => 'date_to_tp'}

    );

C<out> is perl-to-database way. C<in> is perl-from-database way.

This filter is automatically enabled in many method.

    $dbi->insert(table => 'book', param => {issue_date => $tp});

=item 3. Selective search condition

It is difficult to create selective where clause in L<DBI>.

For example, If C<title> and C<author> is specified, we create 

the following SQL.

    select * from book where title = ? and author = ?;

If only C<title> is specified, the following one

    select * from book where title = ?;

If only C<author> is specified, the following one,

    select * from book where author = ?;

This is hard work. Generally we use modules like L<SQL::Abstract>.

L<DBIx::Custom> prepare the way to make it easy.

    # Where object

    my $where = $dbi->where;

    # Search condition

    $where->clause(

        ['and', '{= title}', {'= author'}]

    );

    # Setting to automatically select needed column

    $where->param({title => 'Perl'});

    # Embbed where clause to SQL

    my $sql = "select * from book $where";

You can create where clause which has selected search condition.

You can write nesting of where clause and C<or> condition

=item 4. Methods for insert, update, delete, select

L<DBIx::Custom> provides methods for insert, update, delete, select

There are C<insert()>, C<update()>, C<delete()>,C<select()>.

    my $param = {title => 'Perl', author => 'Ken'};

    $dbi->insert(table => 'book', param => $param);

=item 5. Register method for table.

You can register method for table.

    $dbi->table('book')->method(

        list => sub {

            ...

        },

        something => sub {

            ...

        }

    );

use the mehtod.

    $dbi->table('book')->list;

Many O/R mapper must create class for table,

but L<DBIx::Custom> make it easy.

=back

L<DBIx::Custom> is very useful.

See the following if you are interested in it.

=head2 1. Connect to database

Load L<DBIx::Custom>.

    use DBIx::Custom;

use C<connect()> to connect to database.

Return value is L<DBIx::Custom> object.

    my $dbi = DBIx::Custom->connect(

        data_source => "dbi:mysql:database=bookstore",

        user => 'ken',

        password => '!LFKD%$&',

        dbi_options => {mysql_enable_utf8 => 1}

    );

C<data_source> must be one corresponding to the database system.

The following ones are data source example.

B<MySQL>

    "dbi:mysql:database=$database"

    "dbi:mysql:database=$database;host=$hostname;port=$port"

B<SQLite>

    "dbi:SQLite:dbname=$database"

    "dbi:SQLite:dbname=:memory:"

B<PostgreSQL>

    "dbi:Pg:dbname=$dbname"

B<Oracle>

    "dbi:Oracle:$dbname"

    "dbi:Oracle:host=$host;sid=$sid"

B<ODBC(Microsoft Access)>

    "dbi:ODBC:driver=Microsoft Access Driver (*.mdb);dbq=hoge.mdb"

B<ODBC(SQL Server)>

   "dbi:ODBC:driver={SQL Server};Server=(local);database=test;Trusted_Connection=yes;AutoTranslate=No;"

If authentication is needed, you can specify C<user> and C<password>

L<DBIx::Custom> is wrapper class of L<DBI>.

You can use all methods of L<DBI> from L<DBIx::Custom> object.

    $dbi->do(...);

    $dbi->begin_work;

use C<dhb()> to get database handle of L<DBI>

    my $dbh = $dbi->dbh;

By default, the following ones is set to database handle attributes.

    RaiseError  ->  1

    PrintError  ->  0

    AutoCommit  ->  1

If fatal error occuer, program terminate.

If SQL is executed, commit is executed automatically.

=head2 2. Methods for insert, update, delete, or insert

There are following methods.

=head3 C<insert()>

use C<insert()> to insert row into database

    $dbi->insert(table  => 'book',

                 param  => {title => 'Perl', author => 'Ken'});

C<table> is table name, C<param> is insert data.

Following SQL is executed.

    insert into (title, author) values (?, ?);

=head3 C<update()>

use C<update()> to update row in database.

    $dbi->update(table  => 'book', 

                 param  => {title => 'Perl', author => 'Ken'}, 

                 where  => {id => 5});

C<table> is table name, C<param> is update data, C<where> is condition.

Following SQL is executed.

    update book set title = ?, author = ?;

You can't execute C<update()> without C<where> for safety.

use C<update_all()> if you want to update all rows.

    $dbi->update_all(table  => 'book', 

                     param  => {title => 'Perl', author => 'Ken'});

=head3 C<delete()>

use C<delete()> to delete rows from database.

    $dbi->delete(table  => 'book',

                 where  => {author => 'Ken'});

C<table> is table name, C<where> is condition.

Following SQL is executed.

    delete from book where id = ?;

You can't execute C<delete()> without C<where> for safety.

use C<delete_all()> if you want to delete all rows.

    $dbi->delete_all(table  => 'book');

=head3 C<select()>

use C<select()> to select rows from database

    my $result = $dbi->select(table => 'book');

Following SQL is executed.

    select * from book;

Return value is L<DBIx::Custom::Result> object.

use C<fetch()> to fetch row.

    while (my $row = $result->fetch) {

        my $title  = $row->[0];

        my $author = $row->[1];

    }

See L<3. Fetch row/"3. Fetch row"> about L<DBIx::Custom::Result>.

Continue more examples.

    my $result = $dbi->select(

        table  => 'book',

        column => ['author',  'title'],

        where  => {author => 'Ken'}

    );

C<column> is column names, C<where> is condition.

Following SQL is executed.

    select author, title from book where author = ?;

Next example.

    my $result = $dbi->select(

        table    => 'book',

        where    => {'book.name' => 'Perl'},

        relation => {'book.id' => 'rental.book_id'}

    );

C<relation> is relation of tables. This is inner join.

Following SQL is executed.

    select * from book, rental where book.name = ? and book.id = rental.book_id;

Next example.

    my $result = $dbi->select(

        table  => 'book',

        where  => {author => 'Ken'},

        append => 'for update',

    );

C<append> is string appending to end of SQL.

Following SQL is executed.

    select * book where author = ? for update;

C<appned> is also used at C<insert()>, C<update()>, C<update_all()>

C<delete()>, C<delete_all()>, and C<select()>.

Instead of C<column> and C<table>,

you can use C<selection>.

This is used to specify column names and table names at once

    my $selection = <<"EOS";

    title, author, company_name

    from book inner join company on book.company_id = company.id

    EOS

    $dbi->select(selection => $selection);

Note that you can't use where clause in C<selection>.

use clause like "inner join".

=head3 C<execute()>

use C<execute()> to execute SQL

    $dbi->execute("select * from book;");

Process tag and execute SQL.

    $dbi->execute(

        "select * from book {= title} and {= author};"

        param => {title => 'Perl', author => 'Ken'}

    );

Following SQL is executed.

    select * from book title = ? and author = ?;

Values of title and author is embbdeded into placeholder.

See L<5. Tag/"5. Tag"> about tag.

You don't have to wirte last semicolon in C<execute()>. 

    $dbi->execute('select * from book');

=head3 insert by using primary key : C<insert_at()>

To insert row by using primary key, use C<insert_at()>

    $dbi->insert_at(

        table => 'book', primary_key => ['id'],

        where => ['123'], param => {name => 'Ken'}

    );

In this example, row which id column is 123 is inserted.

NOTE that you must pass array reference as C<where>.

If C<param> contains primary key, the key and value is delete from C<param>.

=head3 Update by using primary key : C<update_at()>

To update row by using primary key, use C<update_at()>

    $dbi->update_at(

        table => 'book', primary_key => ['id'],

        where => ['123'], param => {name => 'Ken'}

    );

In this example, row which id column is 123 is updated.

NOTE that you must pass array reference as C<where>.

If C<param> contains primary key, the key and value is delete from C<param>.

=head3 Delete by using primary key : C<delete_at()>

To delete row by using primary key, use C<delete_at()>

    $dbi->delete_at(table => 'book', primary_key => ['id'], where => ['123']);

In this example, row which id column is 123 is deleted.

NOTE that you must pass array reference as C<where>.

You can also write arguments like this.

    $dbi->delete_at(table => 'book', primary_key => ['id'], param => {id => '123'});

=head3 Select by using primary key : C<select_at()>

To select row by using primary key, use C<select_at()>.

    $dbi->select_at(table => 'book', primary_key => ['id'], where => ['123']);

In this example, row which id colunm is 123 is selected.

NOTE that you must pass array reference as C<where>.

You can also write arguments like this.

    $dbi->select_at(table => 'book', primary_key => ['id'], param => {id => '123'});

=head2 3. Fetch row

Return value of C<select()> is L<DBIx::Custom::Result> object.

There are many methods to fetch row.

=head3 Fetch a row (array) : C<fetch()>

use C<fetch()> to fetch a row and assign it into array reference.

    my $row = $result->fetch;

You can get all rows.

    while (my $row = $result->fetch) {

        my $title  = $row->[0];

        my $author = $row->[1];

    }

=head3 Fetch only first row (array) : C<fetch_first()>

use C<fetch_first()> to fetch only first row.

    my $row = $result->fetch_first;

You can't fetch rest rows

because statement handle C<finish()> is executed.

=head3 Fetch rows (array) : C<fetch_multi()>

use C<fetch_multi()> to fetch rows and assign it into

array reference which has array references as element.

    while (my $rows = $result->fetch_multi(2)) {

        my $title0   = $rows->[0][0];

        my $author0  = $rows->[0][1];

        my $title1   = $rows->[1][0];

        my $author1  = $rows->[1][1];

    }

Specify row count as argument.

You can get the following data.

    [

        ['Perl', 'Ken'],

        ['Ruby', 'Mark']

    ]

=head3 Fetch all rows (array) : C<fetch_all>

use C<fetch_all()> to fetch all rows and assign it into

array reference which has array reference as element.

    my $rows = $result->fetch_all;

You can get the following data.

    [

        ['Perl', 'Ken'],

        ['Ruby', 'Mark']

    ]

=head3 Fetch a row (hash) : C<fetch_hash()>

use C<fetch_hash()> to fetch a row and assign it into hash reference.

    while (my $row = $result->fetch_hash) {

        my $title  = $row->{title};

        my $author = $row->{author};

    }

=head3 Fetch only first row (hash) : C<fetch_hash_first()>

use C<fetch_hash_first()> to fetch only first row

and assign it into hash reference.

    my $row = $result->fetch_hash_first;

You can't fetch rest rows

because statement handle C<finish()> is executed.

=head3 Fetch rows (hash) : C<fetch_hash_multi()>

use C<fetch_hash_multi()> to fetch rows and

assign it into array reference which has hash references as element.

    while (my $rows = $result->fetch_hash_multi(5)) {

        my $title0   = $rows->[0]{title};

        my $author0  = $rows->[0]{author};

        my $title1  = $rows->[1]{title};

        my $author1 = $rows->[1]{author};

    }

Specify row count as argument.

You can get the following data.

    [

        {title => 'Perl', author => 'Ken'},

        {title => 'Ruby', author => 'Mark'}

    ]

=head3 Fetch all rows (hash) : C<fetch_hash_all()>

use C<fetch_hash_all()> to fetch all rows and

assign it into array reference which has hash 

references as element.

    my $rows = $result->fetch_hash_all;

You can get the following data.

    [

        {title => 'Perl', author => 'Ken'},

        {title => 'Ruby', author => 'Mark'}

    ]

=head3 Statement handle : C<sth()>

use <sth()> to get statement handle.

    my $sth = $result->sth;

=head2 4. Filtering

L<DBIx::Custom> provide value filtering. 

For example, You maybe want to convert L<Time::Piece> object to

database date format when register data into database.

and convert database date fromat to L<Time::Piece> object

when get data from database.

=head3 Register filter : C<register_filter()>

use C<register_filter()> to register filter.

    $dbi->register_filter(

        # Time::Piece object to DATE format

        tp_to_date => sub {

            my $date = shift;

            return '0000-00-00' unless $tp;

            return $tp->strftime('%Y-%m-%d');

        },

        # DATE to Time::Piece object

        date_to_tp => sub {

            my $date = shift;

            return if $date eq '0000-00-00';

            return Time::Piece->strptime($date, '%Y-%m-%d');

        },

    );

Registered filter is used by C<apply_filter()> or etc.

=head3 Apply filter : C<apply_filter()>

use C<apply_filter()> to apply registered filter.

    $dbi->apply_filter('book',

        issue_date => {out => 'tp_to_date', in => 'date_to_tp'},

        first_issue_date => {out => 'tp_to_date', in => 'date_to_tp'}

    );

First argument is table name. Arguments after first argument are pairs of column

name and fitering rule. C<out> of filtering rule is filter which is used when data

is send to database. C<in> of filtering rule is filter which is used when data

is got from database. 

You can specify code reference as filter.

    issue_date => {out => sub { ... }, in => sub { ... }}

Applied filter become effective at insert()>, C<update()>, C<update_all()>,

C<delete()>, C<delete_all()>, C<select()>.

    my $tp = Time::Piece->strptime('2010/10/14', '%Y/%m/%d');

    my $result = $dbi->select(table => 'book', where => {issue_date => $tp});

When data is send to database, L<Time::Piece> object is converted

to database date format "2010-10-14"

When data is fetched, database date format is

converted to L<Time::Piece> object.

    my $row = $resutl->fetch_hash_first;

    my $tp = $row->{issue_date};

You can also use column name which contains table name.

    $dbi->select(

        table => 'book',

        where => {'book.issue_date' => $tp}

    );

In fetching, Filter is effective if you use "TABLE__COLUMN" as column name.

    my $result = $dbi->execute(

       "select issue_date as book__issue_date from book");

You can apply C<end> filter execute after C<in> filter.

    $dbi->apply_filter('book',

        issue_date => {out => 'tp_to_date', in => 'date_to_tp',

                       end => 'tp_to_displaydate'},

    );

=head3 Individual filter C<filter>

You can apply individual filter .

This filter overwrite the filter by C<apply_filter()>

use C<filter> option to apply individual filter

when data is send to database.

This option is used at C<insert()>, C<update()>,

C<update_all()>, C<delete()>, C<delete_all()>, C<select()>,

C<execute()>.

C<insert()> example:

    $dbi->insert(

        table => 'book',

        param => {issue_date => $tp, first_issue_date => $tp},

        filter => {issue_date => 'tp_to_date', first_issue_date => 'tp_to_date'}

    );

C<execute()> example:

my $sql = <<"EOS";

select YEAR(issue_date) as issue_year

from book

where YEAR(issue_date) = {? issue_year}

EOS

    my $result = $dbi->execute(

        $sql,

        param => {issue_year => '2010'},

        filter => {issue_year => 'tp_to_year'}

    );

You can also apply indivisual filter when you fetch row.

use C<DBIx::Custom::Result>'s C<filter()>.

    $result->filter(issue_year => 'year_to_tp');

You can remove filter by C<remove_filter()>

    $result->remove_filter;

=head3 End filtering : C<end_filter()>

You can add filter at end.

It is useful to create last output.

use C<end_filter()> to add end filter.

    $result->end_filter(issue_date => sub {

        my $tp = shift;

        return '' unless $tp;

        return $tp->strftime('%Y/%m/%d %h:%m:%s (%a)');

    });

In this example, L<Time::Piece> object is converted to readable format.

You can remove end_filter by C<end_filter>

    $result->remove_end_filter;

=head3 Automate applying filter : C<each_column()>

It is useful to apply filter automatically at date type columns.

You can use C<each_column()> to process all column infos.

    $dbi->each_column(

        sub {

            my ($self, $table, $column, $info) = @_;

            my $type = $info->{TYPE_NAME};

            my $filter = $type eq 'DATE'     ? {out => 'tp_to_date', in => 'date_to_tp'}

                       : $type eq 'DATETIME' ? {out => 'tp_to_datetime', in => 'datetime_to_tp'}

                                             : undef;

            $self->apply_filter($table, $column, $filter)

              if $filter;

        }

    );

C<each_column()> receive callback.

callback arguments are L<DBIx::Custom> object, table name, column name, column information.

Filter is applied automatically by column type.

=head2 5. Tag

=head3 Basic of Tag

You can embedd tag into SQL.

    select * from book where {= title} and {like author};

{= title} and {like author} are tag. Tag has the folloring format.

    {TAG_NAME ARG1 ARG2 ...}

Tag start C<{> and end C<}>. 

Don't insert space between C<{}> and tag name.

C<{> and C<}> are reserved word.

If you want to use these, escape it by '\';

    select from book \\{ ... \\}

\ is perl's escape character, you need two \.

Tag is expanded before executing SQL.

    select * from book where title = ? and author like ?;

use C<execute()> to execute SQL which contains tag

    my $sql = "select * from book where {= author} and {like title};"

    $dbi->execute($sql, param => {title => 'Perl', author => '%Ken%'});

You can specify values embedded into place holder as hash reference using

C<param> option.

You can specify C<filter()> at C<execute()>.

    $dbi->execute($sql, param => {title => 'Perl', author => '%Ken%'}

                  filter => {title => 'to_something');

Note that at C<execute()> the filter applied by C<apply_filter()>

don't has effective to columns.

You have to use C<table> tag in SQL

    my $sql = "select * from {table book} where {= author} and {like title};"

=head3 Tag list

The following tag is available.

=head4 C<table>

    {table NAME} -> NAME

This is used to specify table name in SQL.

If you specify table name, Filtering by 

C<apply_filter()> is effective.

=head4 C<?>

    {? NAME}    ->   ?

=head4 C<=>

    {= NAME}    ->   NAME = ?

=head4 C<E<lt>E<gt>>

    {<> NAME}   ->   NAME <> ?

=head4 C<E<lt>>

    {< NAME}    ->   NAME < ?

=head4 C<E<gt>>

    {> NAME}    ->   NAME > ?

=head4 C<E<gt>=>

    {>= NAME}   ->   NAME >= ?

=head4 C<E<lt>=>

    {<= NAME}   ->   NAME <= ?

=head4 C<like>

    {like NAME}   ->   NAME like ?

=head4 C<in>

    {in NAME COUNT}   ->   NAME in [?, ?, ..]

=head4 C<insert_param>

    {insert_param NAME1 NAME2}   ->   (NAME1, NAME2) values (?, ?)

=head4 C<update_param>

    {update_param NAME1 NAME2}   ->   set NAME1 = ?, NAME2 = ?

=head3 Manipulate same name's columns

It is ok if there are same name's columns.

Let's think two date comparison.

    my $sql = "select * from table where {> date} and {< date};";

In this case, You specify paramter values as array reference.

    my $dbi->execute($sql, param => {date => ['2010-10-01', '2012-02-10']});

=head3 Register Tag : C<register_tag()>

You can register custom tag.

use C<register_tag()> to register tag.

    $dbi->register_tag(

        '=' => sub {

            my $column = shift;

            return ["$column = ?", [$column]];

        }

    );

This is implementation of C<=> tag.

Tag format is the following one.

    {TAG_NAME ARG1 ARG2 ...}

In case C<=> tag. Format is

    {= title}

So subroutine receive one argument "title".

You have to return array reference in the following format.

    [

        String after expanding,

        [COLUMN1(This is used for place holder), COLUMN2 , ...]

    ]

First element is expanded stirng. In this example,

    'title = ?'

Secount element is array reference which is used to embedd value to

place holder. In this example,

    ['title']

If there are more than one placeholders,

This elements is multipul.

You return the following array reference.

    ['title = ?', ['title']]

See source of L<DBIx::Custom::Tag> to see many implementation.

=head2 6. Dinamically create where clause

=head3 Dinamically create where clause : where()

You want to search multiple conditions in many times.

Let's think the following three cases.

Case1: Search only C<title>

    where {= title}

Case2: Search only C<author>

    where {= author}

Case3: Search C<title> and C<author>

    where {= title} and {=author}

L<DBIx::Custom> support dinamic where clause creating.

At first, create L<DBIx::Custom::Where> object by C<where()>.

    my $where = $dbi->where;

Set clause by C<clause()>

    $where->clause(

        ['and', '{= title'}, '{= author}']

    );

C<clause> is the following format.

    ['or' or 'and', TAG1, TAG2, TAG3]

First argument is 'or' or 'and'.

Later than first argument are tag names.

You can write more complex format.

    ['and', 

      '{= title}', 

      ['or', '{= author}', '{like date}']

    ]

This mean "{=title} and ( {=author} or {like date} )".

After setting C<clause>, set C<param>.

    $where->param({title => 'Perl'});

In this example, parameter contains only title.

If you execute C<string_to()>, you can get where clause

which contain only parameter name.

    my $where_clause = $where->to_string;

Parameter name is only title, the following where clause is created.

    where {= title}

You can also create where clause by stringification.

    my $where_clause = "$where";

This is useful to embbed it into SQL. 

=head3 In case where clause contains same name columns

Even if same name tags exists, you can create where clause.

Let's think that there are starting date and ending date.

    my $param = {start_date => '2010-11-15', end_date => '2011-11-21'};

In this case, you set parameter value as array reference.

    my $p = {date => ['2010-11-15', '2011-11-21']};

You can embbed these values into same name tags.

    $where->clause(

        ['and', '{> date}', '{< date}']

    );

    $where->param($p);

If starting date isn't exists, create the following parameter.

    my $p = {date => [$dbi->not_exists, '2011-11-21']};

You can get DBIx::Custom::NotExists object by C<not_exists()>

This mean correnspondinf value isn't exists.

If ending date isn't exists, create the following parameter.

    my $p = {date => ['2010-11-15']};

If both date isn't exists, create the following parameter.

    my $p = {date => []};

This logic is a little difficut. See the following ones.

    my @date;

    push @date, exists $param->{start_date} ? $param->{start_date}

                                            : $dbi->not_exists;

    push @date, $param->{end_date} if exists $param->{end_date};

    my $p = {date => \@date};

=head3 With C<select()>

You can pass L<DBIx::Custom::Where> object to C<where> of C<select()>.

    my $where = $dbi->where;

    $where->clause(['and', '{= title}', '{= author}']);

    $where->param({title => 'Perl'});

    my $result = $dbi->select(table => 'book', where => $where);

You can also pass it to C<where> of C<update()>�AC<delete()>

=head3 With C<execute()>

L<DBIx::Custom::Where> object is embedded into SQL.