use 5.008007;
package DBIx::Custom;
use Object::Simple -base;

our $VERSION = '0.22';

use Carp 'croak';
use DBI;
use DBIx::Custom::Result;
use DBIx::Custom::Query;
use DBIx::Custom::QueryBuilder;
use DBIx::Custom::Where;
use DBIx::Custom::Model;
use DBIx::Custom::Tag;
use DBIx::Custom::Order;
use DBIx::Custom::Util qw/_array_to_hash _subname/;
use DBIx::Custom::Mapper;
use DBIx::Custom::NotExists;
use Encode qw/encode encode_utf8 decode_utf8/;
use Scalar::Util qw/weaken/;


has [qw/connector dsn password quote user exclude_table user_table_info
     user_column_info safety_character/],
  async_conf => sub { {} },
  cache => 0,
  cache_method => sub {
    sub {
      my $self = shift;
      $self->{_cached} ||= {};
      if (@_ > 1) { $self->{_cached}{$_[0]} = $_[1] }
      else { return $self->{_cached}{$_[0]} }
    }
  },
  option => sub { {} },
  default_option => sub {
    {
      RaiseError => 1,
      PrintError => 0,
      AutoCommit => 1
    }
  },
  filters => sub {
    {
      encode_utf8 => sub { encode_utf8($_[0]) },
      decode_utf8 => sub { decode_utf8($_[0]) }
    }
  },
  last_sql => '',
  models => sub { {} },
  now => sub {
    sub {
      my ($sec, $min, $hour, $mday, $mon, $year) = localtime;
      $mon++;
      $year += 1900;
      my $now = sprintf("%04d-%02d-%02d %02d:%02d:%02d",
        $year, $mon, $mday, $hour, $min, $sec);
      return $now;
    }
  },
  query_builder => sub {
    my $self = shift;
    my $builder = DBIx::Custom::QueryBuilder->new(dbi => $self);
    weaken $builder->{dbi};
    return $builder;
  },
  result_class  => 'DBIx::Custom::Result',
  separator => '.',
  stash => sub { {} };

sub available_datatype {
  my $self = shift;
  
  my $data_types = '';
  for my $i (-1000 .. 1000) {
     my $type_info = $self->dbh->type_info($i);
     my $data_type = $type_info->{DATA_TYPE};
     my $type_name = $type_info->{TYPE_NAME};
     $data_types .= "$data_type ($type_name)\n"
       if defined $data_type;
  }
  return "Data Type maybe equal to Type Name" unless $data_types;
  $data_types = "Data Type (Type name)\n" . $data_types;
  return $data_types;
}

sub available_typename {
  my $self = shift;
  
  # Type Names
  my $type_names = {};
  $self->each_column(sub {
    my ($self, $table, $column, $column_info) = @_;
    $type_names->{$column_info->{TYPE_NAME}} = 1
      if $column_info->{TYPE_NAME};
  });
  my @output = sort keys %$type_names;
  unshift @output, "Type Name";
  return join "\n", @output;
}

our $AUTOLOAD;
sub AUTOLOAD {
  my $self = shift;

  # Method name
  my ($package, $mname) = $AUTOLOAD =~ /^([\w\:]+)\:\:(\w+)$/;

  # Call method
  $self->{_methods} ||= {};
  if (my $method = $self->{_methods}->{$mname}) {
    return $self->$method(@_)
  }
  elsif ($self->{dbh} && (my $dbh_method = $self->dbh->can($mname))) {
    $self->dbh->$dbh_method(@_);
  }
  else {
    croak qq{Can't locate object method "$mname" via "$package" }
      . _subname;
  }
}

sub assign_clause {
  my ($self, $param, $opts) = @_;
  
  my $wrap = $opts->{wrap} || {};
  my ($q, $p) = split //, $self->q('');
  
  # Assign clause (performance is important)
  join(
    ', ',
    map {
      ref $param->{$_} eq 'SCALAR' ? "$q$_$p = " . ${$param->{$_}}
      : $wrap->{$_} ? "$q$_$p = " . $wrap->{$_}->(":$_")
      : "$q$_$p = :$_";
    } sort keys %$param
  );
}

sub column {
  my $self = shift;
  my $option = pop if ref $_[-1] eq 'HASH';
  my $real_table = shift;
  my $columns = shift;
  my $table = $option->{alias} || $real_table;
  
  # Columns
  unless (defined $columns) {
    $columns ||= $self->model($real_table)->columns;
  }
  
  # Separator
  my $separator = $self->separator;
  
  # Column clause
  my @column;
  $columns ||= [];
  push @column, $self->q($table) . "." . $self->q($_) .
    " as " . $self->q("${table}${separator}$_")
    for @$columns;
  
  return join (', ', @column);
}

sub connect {
  my $self = ref $_[0] ? shift : shift->new(@_);
  
  my $connector = $self->connector;
  
  if (!ref $connector && $connector) {
    require DBIx::Connector;
    
    my $dsn = $self->dsn;
    my $user = $self->user;
    my $password = $self->password;
    my $option = $self->_option;
    my $connector = DBIx::Connector->new($dsn, $user, $password,
      {%{$self->default_option} , %$option});
    $self->connector($connector);
  }
  
  # Connect
  $self->dbh;
  
  return $self;
}

sub count { shift->select(column => 'count(*)', @_)->fetch_one->[0] }

sub dbh {
  my $self = shift;
  
  # Set
  if (@_) {
    $self->{dbh} = $_[0];
    
    return $self;
  }
  
  # Get
  else {
    # From Connction manager
    if (my $connector = $self->connector) {
      croak "connector must have dbh() method " . _subname
        unless ref $connector && $connector->can('dbh');
        
      $self->{dbh} = $connector->dbh;
    }
    
    # Connect
    $self->{dbh} ||= $self->_connect;
    
    # Quote
    if (!defined $self->reserved_word_quote && !defined $self->quote) {
      my $driver = $self->_driver;
      my $quote =  $driver eq 'odbc' ? '[]'
        : $driver eq 'ado' ? '[]'
        : $driver eq 'mysql' ? '`'
        : '"';
      $self->quote($quote);
    }
    
    return $self->{dbh};
  }
}

sub delete {
  my ($self, %opt) = @_;
  warn "delete method where_param option is DEPRECATED!"
    if $opt{where_param};
  
  # Don't allow delete all rows
  croak qq{delete method where or id option must be specified } . _subname
    if !$opt{where} && !defined $opt{id} && !$opt{allow_delete_all};
  
  # Where
  my $w = $self->_where_clause_and_param($opt{where}, $opt{where_param},
    delete $opt{id}, $opt{primary_key}, $opt{table});

  # Delete statement
  my $sql = "delete ";
  $sql .= "$opt{prefix} " if defined $opt{prefix};
  $sql .= "from " . $self->q($opt{table}) . " $w->{clause} ";
  
  # Execute query
  $opt{statement} = 'delete';
  $self->execute($sql, $w->{param}, %opt);
}

sub delete_all { shift->delete(@_, allow_delete_all => 1) }

sub DESTROY {}

sub create_model {
  my $self = shift;
  
  # Options
  my $opt = ref $_[0] eq 'HASH' ? $_[0] : {@_};
  $opt->{dbi} = $self;
  my $model_class = delete $opt->{model_class} || 'DBIx::Custom::Model';
  my $model_name  = delete $opt->{name};
  my $model_table = delete $opt->{table};
  $model_name ||= $model_table;
  
  # Create model
  my $model = $model_class->new($opt);
  weaken $model->{dbi};
  $model->name($model_name) unless $model->name;
  $model->table($model_table) unless $model->table;
  
  # Apply filter(DEPRECATED logic)
  if ($model->{filter}) {
    my $filter = ref $model->filter eq 'HASH'
      ? [%{$model->filter}]
      : $model->filter;
    $filter ||= [];
    warn "DBIx::Custom::Model filter method is DEPRECATED!"
      if @$filter;
    $self->_apply_filter($model->table, @$filter);
  }
  
  # Set model
  $self->model($model->name, $model);
  
  return $self->model($model->name);
}

sub each_column {
  my ($self, $cb, %options) = @_;

  my $user_column_info = $self->user_column_info;
  
  if ($user_column_info) {
    $self->$cb($_->{table}, $_->{column}, $_->{info}) for @$user_column_info;
  }
  else {
    my $re = $self->exclude_table || $options{exclude_table};
    # Tables
    my %tables;
    $self->each_table(sub { $tables{$_[1]}++ });

    # Iterate all tables
    my @tables = sort keys %tables;
    for (my $i = 0; $i < @tables; $i++) {
      my $table = $tables[$i];
      
      # Iterate all columns
      my $sth_columns;
      eval {$sth_columns = $self->dbh->column_info(undef, undef, $table, '%')};
      next if $@;
      while (my $column_info = $sth_columns->fetchrow_hashref) {
        my $column = $column_info->{COLUMN_NAME};
        $self->$cb($table, $column, $column_info);
      }
    }
  }
}

sub each_table {
  my ($self, $cb, %option) = @_;
  
  my $user_table_infos = $self->user_table_info;
  
  # Iterate tables
  if ($user_table_infos) {
      $self->$cb($_->{table}, $_->{info}) for @$user_table_infos;
  }
  else {
    my $re = $self->exclude_table || $option{exclude};
    my $sth_tables = $self->dbh->table_info;
    while (my $table_info = $sth_tables->fetchrow_hashref) {
      # Table
      my $table = $table_info->{TABLE_NAME};
      next if defined $re && $table =~ /$re/;
      $self->$cb($table, $table_info);
    }
  }
}

sub execute {
  my $self = shift;
  my $sql = shift;

  # Options
  my $params;
  $params = shift if @_ % 2;
  my %opt = @_;
  
  # Async query
  $opt{prepare_attr} = $self->async_conf->{prepare_attr} if $opt{async};
  if ($opt{async} && !$self->{_new_connection}) {
    my $dsn = $self->dsn;
    croak qq/Data source must be specified when "async" option is used/
      unless defined $dsn;
    
    my $user = $self->user;
    my $password = $self->password;
    my $option = $self->_option;
    
    my $new_dbi = bless {%$self}, ref $self;
    $new_dbi->connector(undef);
    $new_dbi->{dbh} = DBI->connect($dsn, $user, $password,
      {%{$new_dbi->default_option}, %$option});
    
    $new_dbi->{_new_connection} = 1;
    return $new_dbi->execute($sql, defined $params ? ($params) : (), %opt);
  }
  
  # Options
  warn "sqlfilter option is DEPRECATED" if $opt{sqlfilter};
  $params ||= $opt{param} || {};
  my $tables = $opt{table} || [];
  $tables = [$tables] unless ref $tables eq 'ARRAY';
  my $filter = ref $opt{filter} eq 'ARRAY' ?
    _array_to_hash($opt{filter}) : $opt{filter};
  
  # Merge second parameter
  my @cleanup;
  my $saved_param;
  $opt{statement} ||= '';
  $opt{statement} = 'select' if $opt{select};
  if (($opt{statement} || '') ne 'insert' && ref $params eq 'ARRAY') {
    my $params2 = $params->[1];
    $params = $params->[0];
    for my $column (keys %$params2) {
      if (!exists $params->{$column}) {
        $params->{$column} = $params2->{$column};
        push @cleanup, $column;
      }
      else {
        delete $params->{$_} for @cleanup;
        @cleanup = ();
        $saved_param  = $params;
        $params = $self->merge_param($params, $params2);
        delete $saved_param->{$_} for (@{$opt{cleanup} || []});
        last;
      }
    }
  }
  $params = [$params] unless ref $params eq 'ARRAY';
  
  # Append
  $sql .= $opt{append} if defined $opt{append} && !ref $sql;
  
  # Query
  my $query;
  if (ref $sql) {
    $query = $sql;
    warn "execute method receiving query object as first parameter is DEPRECATED!" .
      "because this is very buggy.";
  }
  else {
    $query = $opt{reuse}->{$sql} if $opt{reuse};
    unless ($query) {
      my $c = $self->{safety_character};
      # Check unsafety keys
      unless ((join('', keys %{$params->[0]}) || '') =~ /^[$c\.]+$/) {
        for my $column (keys %{$params->[0]}) {
          croak qq{"$column" is not safety column name } . _subname
            unless $column =~ /^[$c\.]+$/;
        }
      }
      $query = $self->_create_query($sql,
        $opt{after_build_sql} || $opt{sqlfilter}, $opt{prepare_attr});
    }
    $query->{statement} = $opt{statement} || '';
    $opt{reuse}->{$sql} = $query if $opt{reuse};
  }
      
  # Save query
  $self->{last_sql} = $query->{sql};

  # Return query
  if ($opt{query}) {
    for my $column (@cleanup, @{$opt{cleanup} || []}) {
      delete $_->{$column} for @$params;
    }
    return $query;
  };
  
  # Merge query filter(DEPRECATED!)
  $filter ||= $query->{filter} || {};
  
  # Tables
  unshift @$tables, @{$query->{tables} || []};
  my $main_table = @{$tables}[-1];

  # Merge id to parameter
  if (defined $opt{id}) {
    my $statement = $query->{statement};
    warn "execute method id option is DEPRECATED!" unless $statement;
    croak "execute id option must be specified with primary_key option"
      unless $opt{primary_key};
    $opt{primary_key} = [$opt{primary_key}] unless ref $opt{primary_key};
    $opt{id} = [$opt{id}] unless ref $opt{id};
    for (my $i = 0; $i < @{$opt{id}}; $i++) {
      my $key = $opt{primary_key}->[$i];
      $key = "$main_table.$key" if $statement eq 'update' ||
        $statement eq 'delete' || $statement eq 'select';
      next if exists $params->[0]->{$key};
      $params->[0]->{$key} = $opt{id}->[$i];
      push @cleanup, $key;1
    }
  }
  
  # Cleanup tables(DEPRECATED!)
  $tables = $self->_remove_duplicate_table($tables, $main_table)
    if @$tables > 1;
  
  # Type rule
  my $type_filters = {};
  my $type_rule_off = !$self->{_type_rule_is_called} || $opt{type_rule_off};
  unless ($type_rule_off) {
    my $type_rule_off_parts = {
      1 => $opt{type_rule1_off},
      2 => $opt{type_rule2_off}
    };
    for my $i (1, 2) {
      unless ($type_rule_off_parts->{$i}) {
        $type_filters->{$i} = {};
        my $table_alias = $opt{table_alias} || {};
        for my $alias (keys %$table_alias) {
          my $table = $table_alias->{$alias};
          
          for my $column (keys %{$self->{"_into$i"}{key}{$table} || {}}) {
            $type_filters->{$i}->{"$alias.$column"} = $self->{"_into$i"}{key}{$table}{$column};
          }
        }
        $type_filters->{$i} = {%{$type_filters->{$i}}, %{$self->{"_into$i"}{key}{$main_table} || {}}}
          if $main_table;
      }
    }
  }
  
  # Applied filter(DEPRECATED!)
  if ($self->{filter}{on}) {
    my $applied_filter = {};
    for my $table (@$tables) {
      $applied_filter = {
        %$applied_filter,
        %{$self->{filter}{out}->{$table} || {}}
      }
    }
    $filter = {%$applied_filter, %$filter};
  }
  
  # Replace filter name to code
  for my $column (keys %$filter) {
    my $name = $filter->{$column};
    if (!defined $name) {
      $filter->{$column} = undef;
    }
    elsif (ref $name ne 'CODE') {
      croak qq{Filter "$name" is not registered" } . _subname
        unless exists $self->filters->{$name};
      $filter->{$column} = $self->filters->{$name};
    }
  }

  # Execute
  my $sth = $query->{sth};
  my $affected;
  if ((!$query->{duplicate} || $opt{bulk_insert}) && $type_rule_off
    && !keys %$filter && !$self->{default_out_filter}
    && !$opt{bind_type} && !$opt{type} && !$ENV{DBIX_CUSTOM_DEBUG})
  {
    eval {
      if ($opt{bulk_insert}) {
        my %count;
        my $param = $params->[0];
        $affected = $sth->execute(map { $param->{$_}->[++$count{$_} - 1] }
          @{$query->{columns}});
      }
      else {
        for my $param (@$params) {
          $affected = $sth->execute(map { $param->{$_} }
            @{$query->{columns}});
        }
      }
    };
  }
  else {
    for my $param (@$params) {
      # Create bind values
      my ($bind, $bind_types) = $self->_create_bind_values($param, $query->{columns},
        $filter, $type_filters, $opt{bind_type} || $opt{type} || {});

      # Execute
      eval {
        if ($opt{bind_type} || $opt{type}) {
          $sth->bind_param($_ + 1, $bind->[$_],
              $bind_types->[$_] ? $bind_types->[$_] : ())
            for (0 .. @$bind - 1);
          $affected = $sth->execute;
        }
        else { $affected = $sth->execute(@$bind) }

        # DEBUG message
        if ($ENV{DBIX_CUSTOM_DEBUG}) {
          warn "SQL:\n" . $query->{sql} . "\n";
          my @output;
          for my $value (@$bind) {
            $value = 'undef' unless defined $value;
            $value = encode($ENV{DBIX_CUSTOM_DEBUG_ENCODING} || 'UTF-8', $value)
              if utf8::is_utf8($value);
            push @output, $value;
          }
          warn "Bind values: " . join(', ', @output) . "\n\n";
        }
      };
    }
  }
  
  $self->_croak($@, qq{. Following SQL is executed.\n}
    . qq{$query->{sql}\n} . _subname) if $@;

  # Remove id from parameter
  for my $column (@cleanup, @{$opt{cleanup} || []}) {
    delete $_->{$column} for @$params;
  }
  
  # Not select statement
  return $affected if !$sth->{NUM_OF_FIELDS} && $opt{statement} ne 'select';

  # Filter(DEPRECATED!)
  my $infilter = {};
  if ($self->{filter}{on}) {
    $infilter->{in}  = {};
    $infilter->{end} = {};
    push @$tables, $main_table if $main_table;
    for my $table (@$tables) {
      for my $way (qw/in end/) {
        $infilter->{$way} = {%{$infilter->{$way}},
          %{$self->{filter}{$way}{$table} || {}}};
      }
    }
  }
  
  # Result
  my $result = $self->result_class->new(
    sth => $sth,
    dbi => $self,
    default_filter => $self->{default_in_filter},
    filter => $infilter->{in} || {},
    end_filter => $infilter->{end} || {},
    type_rule => {
      from1 => $self->type_rule->{from1},
      from2 => $self->type_rule->{from2}
    },
  );
  
  if (my $cb = $opt{async}) {
    require AnyEvent;
    my $watcher;
    weaken $self;
    $watcher = AnyEvent->io(
      fh => $self->async_conf->{fh}->($self),
      poll => 'r',
      cb   => sub {
        $cb->($self, $result);
        undef $watcher;
        undef $result;
        undef $cb;
      },
    );
  }
  else { $result }
}

sub get_table_info {
  my ($self, %opt) = @_;
  
  my $exclude = delete $opt{exclude};
  croak qq/"$_" is wrong option/ for keys %opt;
  
  my $table_info = [];
  $self->each_table(
    sub { push @$table_info, {table => $_[1], info => $_[2] } },
    exclude => $exclude
  );
  
  return [sort {$a->{table} cmp $b->{table} } @$table_info];
}

sub get_column_info {
  my ($self, %opt) = @_;
  
  my $exclude_table = delete $opt{exclude_table};
  croak qq/"$_" is wrong option/ for keys %opt;
  
  my $column_info = [];
  $self->each_column(
    sub { push @$column_info, {table => $_[1], column => $_[2], info => $_[3] } },
    exclude_table => $exclude_table
  );
  
  return [
    sort {$a->{table} cmp $b->{table} || $a->{column} cmp $b->{column} }
      @$column_info];
}

sub helper {
  my $self = shift;
  
  # Register method
  my $methods = ref $_[0] eq 'HASH' ? $_[0] : {@_};
  $self->{_methods} = {%{$self->{_methods} || {}}, %$methods};
  
  return $self;
}

sub insert {
  my $self = shift;
  
  # Options
  my $params = @_ % 2 ? shift : undef;
  my %opt = @_;
  warn "insert method param option is DEPRECATED!" if $opt{param};
  $params ||= delete $opt{param} || {};
  
  my $multi;
  if (ref $params eq 'ARRAY') { $multi = 1 }
  else { $params = [$params] }
  
  # Timestamp(DEPRECATED!)
  if (!$multi && $opt{timestamp} && (my $insert_timestamp = $self->insert_timestamp)) {
    warn "insert timestamp option is DEPRECATED! use ctime option";
    my $columns = $insert_timestamp->[0];
    $columns = [$columns] unless ref $columns eq 'ARRAY';
    my $value = $insert_timestamp->[1];
    $value = $value->() if ref $value eq 'CODE';
    $params->[0]->{$_} = $value for @$columns;
  }

  # Created time and updated time
  my @timestamp_cleanup;
  warn "insert method created_at option is DEPRECATED! "
      . "use ctime option instead. " . _subname
    if $opt{created_at};
  warn "insert method updated_at option is DEPRECATED! "
      . "use mtime option instead. " . _subname
    if $opt{updated_at};
  $opt{ctime} ||= $opt{created_at};
  $opt{mtime} ||= $opt{updated_at};
  if (defined $opt{ctime} || defined $opt{mtime}) {
    my $now = $self->now;
    $now = $now->() if ref $now eq 'CODE';
    if (defined $opt{ctime}) {
      $_->{$opt{ctime}} = $now for @$params;
      push @timestamp_cleanup, $opt{ctime};
    }
    if (defined $opt{mtime}) {
      $_->{$opt{mtime}} = $now for @$params;
      push @timestamp_cleanup, $opt{mtime};
    }
  }
  
  # Merge id to parameter
  my @cleanup;
  my $id_param = {};
  if (defined $opt{id} && !$multi) {
    croak "insert id option must be specified with primary_key option"
      unless $opt{primary_key};
    $opt{primary_key} = [$opt{primary_key}] unless ref $opt{primary_key};
    $opt{id} = [$opt{id}] unless ref $opt{id};
    for (my $i = 0; $i < @{$opt{primary_key}}; $i++) {
      my $key = $opt{primary_key}->[$i];
      next if exists $params->[0]->{$key};
      $params->[0]->{$key} = $opt{id}->[$i];
      push @cleanup, $key;
    }
  }
  
  # Insert statement
  my $sql = "insert ";
  $sql .= "$opt{prefix} " if defined $opt{prefix};
  $sql .= "into " . $self->q($opt{table}) . " ";
  if ($opt{bulk_insert}) {
    $sql .= $self->_multi_values_clause($params, {wrap => $opt{wrap}}) . " ";
    my $new_param = {};
    $new_param->{$_} = [] for keys %{$params->[0]};
    for my $param (@$params) {
      push @{$new_param->{$_}}, $param->{$_} for keys %$param;
    }
    $params = [$new_param];
  }
  else {
    $sql .= $self->values_clause($params->[0], {wrap => $opt{wrap}}) . " ";
  }

  # Remove id from parameter
  delete $params->[0]->{$_} for @cleanup;
  
  # Execute query
  $opt{statement} = 'insert';
  $opt{cleanup} = \@timestamp_cleanup;
  $self->execute($sql, $params, %opt);
}

sub insert_timestamp {
  my $self = shift;
  
  warn "insert_timestamp method is DEPRECATED! use now attribute";
  
  if (@_) {
    $self->{insert_timestamp} = [@_];
    
    return $self;
  }
  return $self->{insert_timestamp};
}

sub include_model {
  my ($self, $name_space, $model_infos) = @_;
  
  # Name space
  $name_space ||= '';
  
  # Get Model infomations
  unless ($model_infos) {

    # Load name space module
    croak qq{"$name_space" is invalid class name } . _subname
      if $name_space =~ /[^\w:]/;
    eval "use $name_space";
    croak qq{Name space module "$name_space.pm" is needed. $@ } . _subname
      if $@;
    
    # Search model modules
    my $path = $INC{"$name_space.pm"};
    $path =~ s/\.pm$//;
    opendir my $dh, $path
      or croak qq{Can't open directory "$path": $! } . _subname
    $model_infos = [];
    while (my $module = readdir $dh) {
      push @$model_infos, $module
        if $module =~ s/\.pm$//;
    }
    close $dh;
  }
  
  # Include models
  for my $model_info (@$model_infos) {
    
    # Load model
    my $model_class;
    my $model_name;
    my $model_table;
    if (ref $model_info eq 'HASH') {
      $model_class = $model_info->{class};
      $model_name  = $model_info->{name};
      $model_table = $model_info->{table};
      
      $model_name  ||= $model_class;
      $model_table ||= $model_name;
    }
    else { $model_class = $model_name = $model_table = $model_info }
    my $mclass = "${name_space}::$model_class";
    croak qq{"$mclass" is invalid class name } . _subname
      if $mclass =~ /[^\w:]/;
    unless ($mclass->can('isa')) {
      eval "use $mclass";
      croak "$@ " . _subname if $@;
    }
    
    # Create model
    my $opt = {};
    $opt->{model_class} = $mclass if $mclass;
    $opt->{name}        = $model_name if $model_name;
    $opt->{table}       = $model_table if $model_table;
    $self->create_model($opt);
  }
  
  return $self;
}

sub like_value { sub { "%$_[0]%" } }

sub mapper {
  my $self = shift;
  return DBIx::Custom::Mapper->new(@_);
}

sub merge_param {
  my ($self, @params) = @_;
  
  # Merge parameters
  my $merge = {};
  for my $param (@params) {
    for my $column (keys %$param) {
      my $param_is_array = ref $param->{$column} eq 'ARRAY' ? 1 : 0;
      
      if (exists $merge->{$column}) {
        $merge->{$column} = [$merge->{$column}]
          unless ref $merge->{$column} eq 'ARRAY';
        push @{$merge->{$column}},
          ref $param->{$column} ? @{$param->{$column}} : $param->{$column};
      }
      else { $merge->{$column} = $param->{$column} }
    }
  }
  
  return $merge;
}

sub model {
  my ($self, $name, $model) = @_;
  
  # Set model
  if ($model) {
    $self->models->{$name} = $model;
    return $self;
  }
  
  # Check model existance
  croak qq{Model "$name" is not included } . _subname
    unless $self->models->{$name};
  
  # Get model
  return $self->models->{$name};
}

sub mycolumn {
  my ($self, $table, $columns) = @_;
  
  # Create column clause
  my @column;
  $columns ||= [];
  push @column, $self->q($table) . "." . $self->q($_) . " as " . $self->q($_)
    for @$columns;
  
  return join (', ', @column);
}

sub new {
  my $self = shift->SUPER::new(@_);
  
  # Check attributes
  my @attrs = keys %$self;
  for my $attr (@attrs) {
    croak qq{Invalid attribute: "$attr" } . _subname
      unless $self->can($attr);
  }
  
  $self->{safety_character} = 'a-zA-Z0-9_'
    unless exists $self->{safety_character};

  # DEPRECATED
  $self->{_tags} = {
    '?'     => \&DBIx::Custom::Tag::placeholder,
    '='     => \&DBIx::Custom::Tag::equal,
    '<>'    => \&DBIx::Custom::Tag::not_equal,
    '>'     => \&DBIx::Custom::Tag::greater_than,
    '<'     => \&DBIx::Custom::Tag::lower_than,
    '>='    => \&DBIx::Custom::Tag::greater_than_equal,
    '<='    => \&DBIx::Custom::Tag::lower_than_equal,
    'like'  => \&DBIx::Custom::Tag::like,
    'in'    => \&DBIx::Custom::Tag::in,
    'insert_param' => \&DBIx::Custom::Tag::insert_param,
    'update_param' => \&DBIx::Custom::Tag::update_param
  };
  $self->{tag_parse} = 1 unless exists $self->{tag_parse};
  $self->{cache} = 0 unless exists $self->{cache};
  
  return $self;
}

sub not_exists { DBIx::Custom::NotExists->singleton }

sub order {
  my $self = shift;
  return DBIx::Custom::Order->new(dbi => $self, @_);
}

sub q {
  my ($self, $value, $quotemeta) = @_;
  
  my $quote = $self->{reserved_word_quote}
    || $self->{quote} || $self->quote || '';
  return "$quote$value$quote"
    if !$quotemeta && ($quote eq '`' || $quote eq '"');
  
  my $q = substr($quote, 0, 1) || '';
  my $p;
  if (defined $quote && length $quote > 1) {
    $p = substr($quote, 1, 1);
  }
  else { $p = $q }
  
  if ($quotemeta) {
    $q = quotemeta($q);
    $p = quotemeta($p);
  }
  
  return "$q$value$p";
}

sub register_filter {
  my $self = shift;
  
  # Register filter
  my $filters = ref $_[0] eq 'HASH' ? $_[0] : {@_};
  $self->filters({%{$self->filters}, %$filters});
  
  return $self;
}

sub select {
  my $self = shift;
  my $column = shift if @_ % 2;
  my %opt = @_;
  $opt{statement} = 'select';
  $opt{column} = $column if defined $column;

  # Options
  my $table_is_empty;
  my $tables = ref $opt{table} eq 'ARRAY' ? $opt{table}
    : defined $opt{table} ? [$opt{table}]
    : [];
  $opt{table} = $tables;
  $table_is_empty = 1 unless @$tables;
  my $where_param = $opt{where_param} || delete $opt{param} || {};
  warn "select method where_param option is DEPRECATED!"
    if $opt{where_param};
  
  # Add relation tables(DEPRECATED!);
  if ($opt{relation}) {
    warn "select() relation option is DEPRECATED!";
    $self->_add_relation_table($tables, $opt{relation});
  }
  
  # Select statement
  my $sql = 'select ';
  
  # Prefix
  $sql .= "$opt{prefix} " if defined $opt{prefix};
  
  # Column
  if (defined $opt{column}) {
    my $columns
      = ref $opt{column} eq 'ARRAY' ? $opt{column} : [$opt{column}];
    for my $column (@$columns) {
      if (ref $column eq 'HASH') {
        $column = $self->column(%$column) if ref $column eq 'HASH';
      }
      elsif (ref $column eq 'ARRAY') {
        warn "select column option [COLUMN => ALIAS] syntax is DEPRECATED!" .
          "use q method to quote the value";
        if (@$column == 3 && $column->[1] eq 'as') {
          warn "[COLUMN, as => ALIAS] is DEPRECATED! use [COLUMN => ALIAS]";
          splice @$column, 1, 1;
        }
        
        $column = join(' ', $column->[0], 'as', $self->q($column->[1]));
      }
      unshift @$tables, @{$self->_search_tables($column)}
        unless $table_is_empty;
      $sql .= "$column, ";
    }
    $sql =~ s/, $/ /;
  }
  else { $sql .= '* ' }

  # Execute query without table
  return $self->execute($sql, {}, %opt) if $table_is_empty;

  # Table
  $sql .= 'from ';
  if ($opt{relation}) {
    my $found = {};
    for my $table (@$tables) {
      $sql .= $self->q($table) . ', ' unless $found->{$table};
      $found->{$table} = 1;
    }
  }
  else { $sql .= $self->q($tables->[-1] || '') . ' ' }
  $sql =~ s/, $/ /;

  # Add tables in parameter
  unshift @$tables,
    @{$self->_search_tables(join(' ', keys %$where_param) || '')};
  
  # Where
  my $w = $self->_where_clause_and_param($opt{where}, $where_param,
    delete $opt{id}, $opt{primary_key}, @$tables ? $tables->[-1] : undef);
  
  # Add table names in where clause
  unshift @$tables, @{$self->_search_tables($w->{clause})};
  
  # Join statement
  $self->_push_join(\$sql, $opt{join}, $tables) if defined $opt{join};
  
  # Add where clause
  $sql .= "$w->{clause} ";
  
  # Relation(DEPRECATED!);
  $self->_push_relation(\$sql, $tables, $opt{relation}, $w->{clause} eq '' ? 1 : 0)
    if $opt{relation};
  
  # Execute query
  return $self->execute($sql, $w->{param}, %opt);
}

sub setup_model {
  my $self = shift;
  
  # Setup model
  $self->each_column(
    sub {
      my ($self, $table, $column, $column_info) = @_;
      if (my $model = $self->models->{$table}) {
        push @{$model->columns}, $column;
      }
    }
  );
  return $self;
}

sub show_datatype {
  my ($self, $table) = @_;
  croak "Table name must be specified" unless defined $table;
  print "$table\n";
  
  my $result = $self->select(table => $table, where => "'0' <> '0'");
  my $sth = $result->sth;

  my $columns = $sth->{NAME};
  my $data_types = $sth->{TYPE};
  
  for (my $i = 0; $i < @$columns; $i++) {
    my $column = $columns->[$i];
    my $data_type = lc $data_types->[$i];
    print "$column: $data_type\n";
  }
}

sub show_typename {
  my ($self, $t) = @_;
  croak "Table name must be specified" unless defined $t;
  print "$t\n";
  
  $self->each_column(sub {
    my ($self, $table, $column, $infos) = @_;
    return unless $table eq $t;
    my $typename = lc $infos->{TYPE_NAME};
    print "$column: $typename\n";
  });
  
  return $self;
}

sub show_tables {
  my $self = shift;
  
  my %tables;
  $self->each_table(sub { $tables{$_[1]}++ });
  print join("\n", sort keys %tables) . "\n";
  return $self;
}

sub type_rule {
  my $self = shift;

  $self->{_type_rule_is_called} = 1;
  
  if (@_) {
    my $type_rule = ref $_[0] eq 'HASH' ? $_[0] : {@_};
    
    # Into
    for my $i (1 .. 2) {
      my $into = "into$i";
      my $exists_into = exists $type_rule->{$into};
      $type_rule->{$into} = _array_to_hash($type_rule->{$into});
      $self->{type_rule} = $type_rule;
      $self->{"_$into"} = {};
      for my $type_name (keys %{$type_rule->{$into} || {}}) {
        croak qq{type name of $into section must be lower case}
          if $type_name =~ /[A-Z]/;
      }
      
      $self->each_column(sub {
        my ($dbi, $table, $column, $column_info) = @_;
        
        my $type_name = lc $column_info->{TYPE_NAME};
        if ($type_rule->{$into} &&
            (my $filter = $type_rule->{$into}->{$type_name}))
        {
          return unless exists $type_rule->{$into}->{$type_name};
          if (defined $filter && ref $filter ne 'CODE') 
          {
            my $fname = $filter;
            croak qq{Filter "$fname" is not registered" } . _subname
              unless exists $self->filters->{$fname};
            
            $filter = $self->filters->{$fname};
          }

          $self->{"_$into"}{key}{$table}{$column} = $filter;
          $self->{"_$into"}{dot}{"$table.$column"} = $filter;
        }
      });
    }

    # From
    for my $i (1 .. 2) {
      $type_rule->{"from$i"} = _array_to_hash($type_rule->{"from$i"});
      for my $data_type (keys %{$type_rule->{"from$i"} || {}}) {
        croak qq{data type of from$i section must be lower case or number}
          if $data_type =~ /[A-Z]/;
        my $fname = $type_rule->{"from$i"}{$data_type};
        if (defined $fname && ref $fname ne 'CODE') {
          croak qq{Filter "$fname" is not registered" } . _subname
            unless exists $self->filters->{$fname};
          
          $type_rule->{"from$i"}{$data_type} = $self->filters->{$fname};
        }
      }
    }
    
    return $self;
  }
  
  return $self->{type_rule} || {};
}

sub update {
  my $self = shift;

  # Options
  my $param = @_ % 2 ? shift : undef;
  my %opt = @_;
  warn "update param option is DEPRECATED!" if $opt{param};
  warn "update method where_param option is DEPRECATED!"
    if $opt{where_param};
  $param ||= $opt{param} || {};
  
  # Don't allow update all rows
  croak qq{update method where option must be specified } . _subname
    if !$opt{where} && !defined $opt{id} && !$opt{allow_update_all};
  
  # Timestamp(DEPRECATED!)
  if ($opt{timestamp} && (my $update_timestamp = $self->update_timestamp)) {
    warn "update timestamp option is DEPRECATED! use mtime";
    my $columns = $update_timestamp->[0];
    $columns = [$columns] unless ref $columns eq 'ARRAY';
    my $value = $update_timestamp->[1];
    $value = $value->() if ref $value eq 'CODE';
    $param->{$_} = $value for @$columns;
  }

  # Created time and updated time
  my @timestamp_cleanup;
  warn "update method update_at option is DEPRECATED! "
      . "use mtime option instead " . _subname
    if $opt{updated_at};
  $opt{mtime} ||= $opt{updated_at};
  if (defined $opt{mtime}) {
    my $now = $self->now;
    $now = $now->() if ref $now eq 'CODE';
    $param->{$opt{mtime}} = $self->now->();
    push @timestamp_cleanup, $opt{mtime};
  }

  # Assign clause
  my $assign_clause = $self->assign_clause($param, {wrap => $opt{wrap}});
  
  # Where
  my $w = $self->_where_clause_and_param($opt{where}, $opt{where_param},
    delete $opt{id}, $opt{primary_key}, $opt{table});
  
  # Update statement
  my $sql = "update ";
  $sql .= "$opt{prefix} " if defined $opt{prefix};
  $sql .= $self->q($opt{table}) . " set $assign_clause $w->{clause} ";
  
  # Execute query
  $opt{statement} = 'update';
  $opt{cleanup} = \@timestamp_cleanup;
  $self->execute($sql, [$param, $w->{param}], %opt);
}

sub update_all { shift->update(@_, allow_update_all => 1) };

sub update_or_insert {
  my ($self, $param, %opt) = @_;
  croak "update_or_insert method need primary_key and id option "
    unless defined $opt{id} && defined $opt{primary_key};
  my $statement_opt = $opt{option} || {};

  my $rows = $self->select(%opt, %{$statement_opt->{select} || {}})->all;
  if (@$rows == 0) {
    return $self->insert($param, %opt, %{$statement_opt->{insert} || {}});
  }
  elsif (@$rows == 1) {
    return 0 unless keys %$param;
    return $self->update($param, %opt, %{$statement_opt->{update} || {}});
  }
  else { croak "selected row must be one " . _subname }
}

sub update_timestamp {
  my $self = shift;
  
  warn "update_timestamp method is DEPRECATED! use now method";
  
  if (@_) {
    $self->{update_timestamp} = [@_];
    
    return $self;
  }
  return $self->{update_timestamp};
}

sub values_clause {
  my ($self, $param, $opts) = @_;
  
  my $wrap = $opts->{wrap} || {};
  
  # Create insert parameter tag
  my ($q, $p) = split //, $self->q('');
  
  # values clause(performance is important)
  '(' .
  join(
    ', ',
    map { "$q$_$p" } sort keys %$param
  ) .
  ') values (' .
  join(
    ', ',
    map {
      ref $param->{$_} eq 'SCALAR' ? ${$param->{$_}} :
      $wrap->{$_} ? $wrap->{$_}->(":$_") :
      ":$_";
    } sort keys %$param
  ) .
  ')'
}

sub _multi_values_clause {
  my ($self, $params, $opts) = @_;
  
  my $wrap = $opts->{wrap} || {};
  
  # Create insert parameter tag
  my ($q, $p) = split //, $self->q('');
  
  # Multi values clause
  my $clause = '(' . join(', ', map { "$q$_$p" } sort keys %{$params->[0]}) . ') values ';
  
  for (1 .. @$params) {
    $clause .= '(' . join(', ', 
      map {
        ref $params->[0]->{$_} eq 'SCALAR' ? ${$params->[0]->{$_}} :
        $wrap->{$_} ? $wrap->{$_}->(":$_") :
        ":$_";
      } sort keys %{$params->[0]}
    ) . '), '
  }
  $clause =~ s/, $//;
  return $clause;
}

sub where { DBIx::Custom::Where->new(dbi => shift, @_) }

sub _create_query {
  
  my ($self, $source, $after_build_sql, $prepare_attr) = @_;
  
  $prepare_attr ||= {};
  
  # Cache
  my $cache = $self->{cache};
  
  # Query
  my $query;
  
  # Get cached query
  if ($cache) {
    
    # Get query
    my $q = $self->cache_method->($self, $source);
    
    # Create query
    if ($q) {
      $query = DBIx::Custom::Query->new($q);
      $query->{filters} = $self->filters;
    }
  }
  
  # Create query
  unless ($query) {

    # Create query
    my $tag_parse = exists $ENV{DBIX_CUSTOM_TAG_PARSE}
      ? $ENV{DBIX_CUSTOM_TAG_PARSE} : $self->{tag_parse};

    my $sql = " " . $source || '';
    if ($tag_parse && ($sql =~ /\s\{/)) {
      $query = $self->query_builder->build_query($sql);
    }
    else {
      my @columns;
      my $c = $self->{safety_character};
      my $re = $c eq 'a-zA-Z0-9_'
        ? qr/(.*?[^\\]):([$c\.]+)(?:\{(.*?)\})?(.*)/so
        : qr/(.*?[^\\]):([$c\.]+)(?:\{(.*?)\})?(.*)/s;
      my %duplicate;
      my $duplicate;
      # Parameter regex
      $sql =~ s/([0-9]):/$1\\:/g;
      my $new_sql = '';
      while ($sql =~ /$re/) {
        push @columns, $2;
        $duplicate = 1 if ++$duplicate{$columns[-1]} > 1;
        ($new_sql, $sql) = defined $3 ?
          ($new_sql . "$1$2 $3 ?", " $4") : ($new_sql . "$1?", " $4");
      }
      $new_sql .= $sql;
      $new_sql =~ s/\\:/:/g if index($new_sql, "\\:") != -1;

      # Create query
      $query = {sql => $new_sql, columns => \@columns, duplicate => $duplicate};
    }
    
    # Save query to cache
    $self->cache_method->(
      $self, $source,
      {
        sql     => $query->{sql}, 
        columns => $query->{columns},
        tables  => $query->{tables} || []
      }
    ) if $cache;
  }

  # Filter SQL
  $query->{sql} = $after_build_sql->($query->{sql}) if $after_build_sql;
  
  # Save sql
  $self->{last_sql} = $query->{sql};
  
  # Prepare statement handle
  my $sth;
  eval { $sth = $self->dbh->prepare($query->{sql}, $prepare_attr) };
  
  if ($@) {
    $self->_croak($@, qq{. Following SQL is executed.\n}
                    . qq{$query->{sql}\n} . _subname);
  }
  
  # Set statement handle
  $query->{sth} = $sth;
  
  # Set filters
  $query->{filters} = $self->{filters} || $self->filters;
  
  return $query;
}

sub _create_bind_values {
  my ($self, $params, $columns, $filter, $type_filters, $bind_type) = @_;
  
  $bind_type = _array_to_hash($bind_type) if ref $bind_type eq 'ARRAY';
  
  # Create bind values
  my @bind;
  my @types;
  my %count;
  my %not_exists;
  for my $column (@$columns) {
    
    # Bind value
    if(ref $params->{$column} eq 'ARRAY') {
      my $i = $count{$column} || 0;
      $i += $not_exists{$column} || 0;
      my $found;
      for (my $k = $i; $i < @{$params->{$column}}; $k++) {
        if (ref $params->{$column}->[$k] eq 'DBIx::Custom::NotExists') {
            $not_exists{$column}++;
        }
        else  {
          push @bind, $params->{$column}->[$k];
          $found = 1;
          last
        }
      }
      next unless $found;
    }
    else { push @bind, $params->{$column} }
    
    # Filter
    if (my $f = $filter->{$column} || $self->{default_out_filter} || '') {
      $bind[-1] = $f->($bind[-1]);
    }
    
    # Type rule
    if ($self->{_type_rule_is_called}) {
      my $tf1 = $self->{"_into1"}->{dot}->{$column}
        || $type_filters->{1}->{$column};
      $bind[-1] = $tf1->($bind[-1]) if $tf1;
      my $tf2 = $self->{"_into2"}->{dot}->{$column}
        || $type_filters->{2}->{$column};
      $bind[-1] = $tf2->($bind[-1]) if $tf2;
    }
   
    # Bind types
    push @types, $bind_type->{$column};
    
    # Count up 
    $count{$column}++;
  }
  
  return (\@bind, \@types);
}

sub _id_to_param {
  my ($self, $id, $primary_keys, $table) = @_;
  
  # Check primary key
  croak "primary_key option " .
        "must be specified when id option is used" . _subname
    unless defined $primary_keys;
  $primary_keys = [$primary_keys] unless ref $primary_keys eq 'ARRAY';
  
  # Create parameter
  my $param = {};
  if (defined $id) {
    $id = [$id] unless ref $id;
    for(my $i = 0; $i < @$id; $i++) {
      my $key = $primary_keys->[$i];
      $key = "$table." . $key if $table;
      $param->{$key} = $id->[$i];
    }
  }
  
  return $param;
}

sub _connect {
  my $self = shift;
  
  # Attributes
  my $dsn = $self->data_source;
  warn "data_source is DEPRECATED!\n"
    if $dsn;
  $dsn ||= $self->dsn;
  croak qq{"dsn" must be specified } . _subname
    unless $dsn;
  my $user        = $self->user;
  my $password    = $self->password;
  my $option = $self->_option;
  $option = {%{$self->default_option}, %$option};
  
  # Connect
  my $dbh;
  eval { $dbh = DBI->connect($dsn, $user, $password, $option) };
  
  # Connect error
  croak "$@ " . _subname if $@;
  
  return $dbh;
}

sub _croak {
  my ($self, $error, $append) = @_;
  
  # Append
  $append ||= "";
  
  # Verbose
  if ($Carp::Verbose) { croak $error }
  
  # Not verbose
  else {
    # Remove line and module infromation
    my $at_pos = rindex($error, ' at ');
    $error = substr($error, 0, $at_pos);
    $error =~ s/\s+$//;
    croak "$error$append";
  }
}

sub _driver { lc shift->{dbh}->{Driver}->{Name} }

sub _need_tables {
  my ($self, $tree, $need_tables, $tables) = @_;
  
  # Get needed tables
  for my $table (@$tables) {
    if ($tree->{$table}) {
      $need_tables->{$table} = 1;
      $self->_need_tables($tree, $need_tables, [$tree->{$table}{parent}])
    }
  }
}

sub _option {
  my $self = shift;
  my $option = {%{$self->dbi_options}, %{$self->dbi_option}, %{$self->option}};
  warn "dbi_options is DEPRECATED! use option instead\n"
    if keys %{$self->dbi_options};
  warn "dbi_option is DEPRECATED! use option instead\n"
    if keys %{$self->dbi_option};
  return $option;
}

sub _push_join {
  my ($self, $sql, $join, $join_tables) = @_;
  
  $join = [$join] unless ref $join eq 'ARRAY';
  
  # No join
  return unless @$join;
  
  # Push join clause
  my $tree = {};
  for (my $i = 0; $i < @$join; $i++) {
    
    # Arrange
    my $join_clause;;
    my $option;
    if (ref $join->[$i] eq 'HASH') {
      $join_clause = $join->[$i]->{clause};
      $option = {table => $join->[$i]->{table}};
    }
    else {
      $join_clause = $join->[$i];
      $option = {};
    };

    # Find tables in join clause
    my $table1;
    my $table2;
    if (my $table = $option->{table}) {
      $table1 = $table->[0];
      $table2 = $table->[1];
    }
    else {
      my $q = $self->_quote;
      my $j_clause = (split /\s+on\s+/, $join_clause)[-1];
      $j_clause =~ s/'.+?'//g;
      my $q_re = quotemeta($q);
      $j_clause =~ s/[$q_re]//g;
      
      my @j_clauses = reverse split /\s(and|on)\s/, $j_clause;
      my $c = $self->{safety_character};
      my $join_re = qr/([$c]+)\.[$c]+[^$c].*?([$c]+)\.[$c]+/sm;
      for my $clause (@j_clauses) {
        if ($clause =~ $join_re) {
          $table1 = $1;
          $table2 = $2;
          last;
        }                
      }
    }
    croak qq{join clause must have two table name after "on" keyword. } .
        qq{"$join_clause" is passed }  . _subname
      unless defined $table1 && defined $table2;
    croak qq{right side table of "$join_clause" must be unique } . _subname
      if exists $tree->{$table2};
    croak qq{Same table "$table1" is specified} . _subname
      if $table1 eq $table2;
    $tree->{$table2}
      = {position => $i, parent => $table1, join => $join_clause};
  }
  
  # Search need tables
  my $need_tables = {};
  $self->_need_tables($tree, $need_tables, $join_tables);
  my @need_tables = sort { $tree->{$a}{position} <=> $tree->{$b}{position} }
    keys %$need_tables;
  
  # Add join clause
  $$sql .= $tree->{$_}{join} . ' ' for @need_tables;
}

sub _quote {
  my $self = shift;
  return $self->{reserved_word_quote} || $self->quote || '';
}

sub _remove_duplicate_table {
  my ($self, $tables, $main_table) = @_;
  
  # Remove duplicate table
  my %tables = map {defined $_ ? ($_ => 1) : ()} @$tables;
  delete $tables{$main_table} if $main_table;
  
  my $new_tables = [keys %tables, $main_table ? $main_table : ()];
  if (my $q = $self->_quote) {
    $q = quotemeta($q);
    $_ =~ s/[$q]//g for @$new_tables;
  }

  return $new_tables;
}

sub _search_tables {
  my ($self, $source) = @_;
  
  # Search tables
  my $tables = [];
  my $safety_character = $self->{safety_character};
  my $q = $self->_quote;
  my $quoted_safety_character_re = $self->q("?([$safety_character]+)", 1);
  my $table_re = $q ? qr/(?:^|[^$safety_character])${quoted_safety_character_re}?\./
    : qr/(?:^|[^$safety_character])([$safety_character]+)\./;
  while ($source =~ /$table_re/g) { push @$tables, $1 }
  return $tables;
}

sub _where_clause_and_param {
  my ($self, $where, $where_param, $id, $primary_key, $table) = @_;

  $where ||= {};
  $where = $self->_id_to_param($id, $primary_key, $table) if defined $id;
  $where_param ||= {};
  my $w = {};

  if (ref $where eq 'HASH') {
    my $clause = [];
    my $column_join = '';
    for my $column (keys %$where) {
      $column_join .= $column;
      my $table;
      my $c;
      if ($column =~ /(?:(.*?)\.)?(.*)/) {
        $table = $1;
        $c = $2;
      }
      
      my $table_quote;
      $table_quote = $self->q($table) if defined $table;
      my $column_quote = $self->q($c);
      $column_quote = $table_quote . '.' . $column_quote
        if defined $table_quote;
      if (ref $where->{$column} eq 'ARRAY') {
        my $c = join(', ', (":$column") x @{$where->{$column}});
        push @$clause, "$column_quote in ( $c )";
      }
      else { push @$clause, "$column_quote = :$column" }
    }
    
    $w->{clause} = @$clause ? "where ( " . join(' and ', @$clause) . " ) " : '' ;
    $w->{param} = $where;
    $w->{param} = keys %$where_param
      ? $self->merge_param($where_param, $where)
      : $where;
  }  
  elsif (ref $where) {
    my $obj;

    if (ref $where eq 'DBIx::Custom::Where') { $obj = $where }
    elsif (ref $where eq 'ARRAY') {
      $obj = $self->where(clause => $where->[0], param => $where->[1]);
    }
    
    # Check where argument
    croak qq{"where" must be hash reference or DBIx::Custom::Where object}
        . qq{or array reference, which contains where clause and parameter}
        . _subname
      unless ref $obj eq 'DBIx::Custom::Where';

    $w->{clause} = $obj->to_string;
    $w->{param} = keys %$where_param
      ? $self->merge_param($where_param, $obj->param)
      : $obj->param;
  }
  elsif ($where) {
    $w->{clause} = "where $where";
    $w->{param} = $where_param;
  }
  
  return $w;
}

sub _apply_filter {
  my ($self, $table, @cinfos) = @_;

  # Initialize filters
  $self->{filter} ||= {};
  $self->{filter}{on} = 1;
  $self->{filter}{out} ||= {};
  $self->{filter}{in} ||= {};
  $self->{filter}{end} ||= {};
  
  # Usage
  my $usage = "Usage: \$dbi->apply_filter(" .
    "TABLE, COLUMN1, {in => INFILTER1, out => OUTFILTER1, end => ENDFILTER1}, " .
    "COLUMN2, {in => INFILTER2, out => OUTFILTER2, end => ENDFILTER2}, ...)";
  
  # Apply filter
  for (my $i = 0; $i < @cinfos; $i += 2) {
    
    # Column
    my $column = $cinfos[$i];
    if (ref $column eq 'ARRAY') {
      for my $c (@$column) { push @cinfos, $c, $cinfos[$i + 1] }
      next;
    }
    
    # Filter infomation
    my $finfo = $cinfos[$i + 1] || {};
    croak "$usage (table: $table) " . _subname
      unless  ref $finfo eq 'HASH';
    for my $ftype (keys %$finfo) {
      croak "$usage (table: $table) " . _subname
        unless $ftype eq 'in' || $ftype eq 'out' || $ftype eq 'end'; 
    }
    
    # Set filters
    for my $way (qw/in out end/) {
  
      # Filter
      my $filter = $finfo->{$way};
      
      # Filter state
      my $state = !exists $finfo->{$way} ? 'not_exists'
        : !defined $filter        ? 'not_defined'
        : ref $filter eq 'CODE'   ? 'code'
        : 'name';
      
      # Filter is not exists
      next if $state eq 'not_exists';
      
      # Check filter name
      croak qq{Filter "$filter" is not registered } . _subname
        if  $state eq 'name' && ! exists $self->filters->{$filter};
      
      # Set filter
      my $f = $state eq 'not_defined' ? undef
        : $state eq 'code' ? $filter
        : $self->filters->{$filter};
      $self->{filter}{$way}{$table}{$column} = $f;
      $self->{filter}{$way}{$table}{"$table.$column"} = $f;
      $self->{filter}{$way}{$table}{"${table}__$column"} = $f;
      $self->{filter}{$way}{$table}{"${table}-$column"} = $f;
    }
  }
  
  return $self;
}

# DEPRECATED!
has 'data_source';
has dbi_options => sub { {} };
has filter_check  => 1;
has 'reserved_word_quote';
has dbi_option => sub { {} };
has default_dbi_option => sub {
  warn "default_dbi_option is DEPRECATED! use default_option instead";
  return shift->default_option;
};

# DEPRECATED
sub tag_parse {
 my $self = shift;
 warn "tag_parse is DEPRECATED! use \$ENV{DBIX_CUSTOM_TAG_PARSE} " .
   "environment variable";
  if (@_) {
    $self->{tag_parse} = $_[0];
    return $self;
  }
  return $self->{tag_parse};
}

# DEPRECATED!
sub method {
  warn "method is DEPRECATED! use helper instead";
  return shift->helper(@_);
}

# DEPRECATED!
sub assign_param {
  my $self = shift;
  warn "assing_param is DEPRECATED! use assign_clause instead";
  return $self->assign_clause(@_);
}

# DEPRECATED
sub update_param {
  my ($self, $param, $opts) = @_;
  
  warn "update_param is DEPRECATED! use assign_clause instead.";
  
  # Create update parameter tag
  my $tag = $self->assign_clause($param, $opts);
  $tag = "set $tag" unless $opts->{no_set};

  return $tag;
}

# DEPRECATED!
sub create_query {
  warn "create_query is DEPRECATED! use query option of each method";
  shift->_create_query(@_);
}

# DEPRECATED!
sub apply_filter {
  my $self = shift;
  
  warn "apply_filter is DEPRECATED!";
  return $self->_apply_filter(@_);
}

# DEPRECATED!
sub select_at {
  my ($self, %opt) = @_;

  warn "select_at is DEPRECATED! use select method id option instead";

  # Options
  my $primary_keys = delete $opt{primary_key};
  my $where = delete $opt{where};
  my $param = delete $opt{param};
  
  # Table
  croak qq{"table" option must be specified } . _subname
    unless $opt{table};
  my $table = ref $opt{table} ? $opt{table}->[-1] : $opt{table};
  
  # Create where parameter
  my $where_param = $self->_id_to_param($where, $primary_keys);
  
  return $self->select(where => $where_param, %opt);
}

# DEPRECATED!
sub delete_at {
  my ($self, %opt) = @_;

  warn "delete_at is DEPRECATED! use delete method id option instead";
  
  # Options
  my $primary_keys = delete $opt{primary_key};
  my $where = delete $opt{where};
  
  # Create where parameter
  my $where_param = $self->_id_to_param($where, $primary_keys);
  
  return $self->delete(where => $where_param, %opt);
}

# DEPRECATED!
sub update_at {
  my $self = shift;

  warn "update_at is DEPRECATED! use update method id option instead";
  
  # Options
  my $param;
  $param = shift if @_ % 2;
  my %opt = @_;
  my $primary_keys = delete $opt{primary_key};
  my $where = delete $opt{where};
  my $p = delete $opt{param} || {};
  $param  ||= $p;
  
  # Create where parameter
  my $where_param = $self->_id_to_param($where, $primary_keys);
  
  return $self->update(where => $where_param, param => $param, %opt);
}

# DEPRECATED!
sub insert_at {
  my $self = shift;
  
  warn "insert_at is DEPRECATED! use insert method id option instead";
  
  # Options
  my $param;
  $param = shift if @_ % 2;
  my %opt = @_;
  my $primary_key = delete $opt{primary_key};
  $primary_key = [$primary_key] unless ref $primary_key;
  my $where = delete $opt{where};
  my $p = delete $opt{param} || {};
  $param  ||= $p;
  
  # Create where parameter
  my $where_param = $self->_id_to_param($where, $primary_key);
  $param = $self->merge_param($where_param, $param);
  
  return $self->insert(param => $param, %opt);
}

# DEPRECATED!
sub register_tag {
  my $self = shift;
  
  warn "register_tag is DEPRECATED!";
  
  # Merge tag
  my $tags = ref $_[0] eq 'HASH' ? $_[0] : {@_};
  $self->{_tags} = {%{$self->{_tags} || {}}, %$tags};
  
  return $self;
}

# DEPRECATED!
sub register_tag_processor {
  my $self = shift;
  warn "register_tag_processor is DEPRECATED!";
  # Merge tag
  my $tag_processors = ref $_[0] eq 'HASH' ? $_[0] : {@_};
  $self->{_tags} = {%{$self->{_tags} || {}}, %{$tag_processors}};
  return $self;
}

# DEPRECATED!
sub default_bind_filter {
  my $self = shift;
  
  warn "default_bind_filter is DEPRECATED!";
  
  if (@_) {
    my $fname = $_[0];
    
    if (@_ && !$fname) {
      $self->{default_out_filter} = undef;
    }
    else {
      croak qq{Filter "$fname" is not registered}
        unless exists $self->filters->{$fname};
  
      $self->{default_out_filter} = $self->filters->{$fname};
    }
    return $self;
  }
  
  return $self->{default_out_filter};
}

# DEPRECATED!
sub default_fetch_filter {
  my $self = shift;

  warn "default_fetch_filter is DEPRECATED!";
  
  if (@_) {
    my $fname = $_[0];

    if (@_ && !$fname) {
      $self->{default_in_filter} = undef;
    }
    else {
      croak qq{Filter "$fname" is not registered}
        unless exists $self->filters->{$fname};
  
      $self->{default_in_filter} = $self->filters->{$fname};
    }
    
    return $self;
  }
  
  return $self->{default_in_filter};
}

# DEPRECATED!
sub insert_param {
  my $self = shift;
  warn "insert_param is DEPRECATED! use values_clause instead";
  return $self->values_clause(@_);
}

# DEPRECATED!
sub insert_param_tag {
  warn "insert_param_tag is DEPRECATED! " .
    "use insert_param instead!";
  return shift->insert_param(@_);
}

# DEPRECATED!
sub update_param_tag {
  warn "update_param_tag is DEPRECATED! " .
    "use update_param instead";
  return shift->update_param(@_);
}
# DEPRECATED!
sub _push_relation {
  my ($self, $sql, $tables, $relation, $need_where) = @_;
  
  if (keys %{$relation || {}}) {
    $$sql .= $need_where ? 'where ' : 'and ';
    for my $rcolumn (keys %$relation) {
      my $table1 = (split (/\./, $rcolumn))[0];
      my $table2 = (split (/\./, $relation->{$rcolumn}))[0];
      push @$tables, ($table1, $table2);
      $$sql .= "$rcolumn = " . $relation->{$rcolumn} .  'and ';
    }
  }
  $$sql =~ s/and $/ /;
}

# DEPRECATED!
sub _add_relation_table {
  my ($self, $tables, $relation) = @_;
  
  if (keys %{$relation || {}}) {
    for my $rcolumn (keys %$relation) {
      my $table1 = (split (/\./, $rcolumn))[0];
      my $table2 = (split (/\./, $relation->{$rcolumn}))[0];
      my $table1_exists;
      my $table2_exists;
      for my $table (@$tables) {
        $table1_exists = 1 if $table eq $table1;
        $table2_exists = 1 if $table eq $table2;
      }
      unshift @$tables, $table1 unless $table1_exists;
      unshift @$tables, $table2 unless $table2_exists;
    }
  }
}

1;

=head1 NAME

DBIx::Custom - DBI extension to execute insert, update, delete, and select easily

=head1 SYNOPSIS

  use DBIx::Custom;
  
  # Connect
  my $dbi = DBIx::Custom->connect(
    dsn => "dbi:mysql:database=dbname",
    user => 'ken',
    password => '!LFKD%$&',
    option => {mysql_enable_utf8 => 1}
  );

  # Insert 
  $dbi->insert({title => 'Perl', author => 'Ken'}, table  => 'book');
  
  # Update 
  $dbi->update({title => 'Perl', author => 'Ken'}, table  => 'book',
    where  => {id => 5});
  
  # Delete
  $dbi->delete(table  => 'book', where => {author => 'Ken'});

  # Select
  my $result = $dbi->select(table  => 'book',
    column => ['title', 'author'], where  => {author => 'Ken'});

  # Select, more complex
  my $result = $dbi->select(
    table  => 'book',
    column => [
      {book => [qw/title author/]},
      {company => ['name']}
    ],
    where  => {'book.author' => 'Ken'},
    join => ['left outer join company on book.company_id = company.id'],
    append => 'order by id limit 5'
  );
  
  # Fetch
  while (my $row = $result->fetch) {
      
  }
  
  # Fetch as hash
  while (my $row = $result->fetch_hash) {
      
  }
  
  # Execute SQL with parameter.
  $dbi->execute(
    "select id from book where author = :author and title like :title",
    {author => 'ken', title => '%Perl%'}
  );
  
=head1 DESCRIPTION

L<DBIx::Custom> is L<DBI> wrapper module to execute SQL easily.
This module have the following features.

=over 4

=item *

Execute C<insert>, C<update>, C<delete>, or C<select> statement easily

=item *

Create C<where> clause flexibly

=item *

Named place holder support

=item *

Model support

=item *

Connection manager support

=item *

Choice your favorite relational database management system,
C<MySQL>, C<SQLite>, C<PostgreSQL>, C<Oracle>,
C<Microsoft SQL Server>, C<Microsoft Access>, C<DB2> or anything, 

=item *

Filtering by data type or column name

=item *

Create C<order by> clause flexibly

=back

=head1 DOCUMENTATION

L<DBIx::Custom::Guide> - How to use L<DBIx::Custom>

L<DBIx::Custom Wiki|https://github.com/yuki-kimoto/DBIx-Custom/wiki>
- Theare are various examples.

Module documentations - 
L<DBIx::Custom::Result>,
L<DBIx::Custom::Query>,
L<DBIx::Custom::Where>,
L<DBIx::Custom::Model>,
L<DBIx::Custom::Order>

=head1 ATTRIBUTES

=head2 C<async_conf> EXPERIMENTAL

  my $async_conf = $dbi->async_conf;
  $dbi = $dbi->async_conf($conf);

Setting when C<async> option is used.

  # MySQL
  $dbi->async_conf({
    prepare_attr => {async => 1},
    fh => sub { shift->dbh->mysql_fd }
  })

C<prepare_attr> is DBI's C<prepare> method second argument,
C<fh> is callback that return file handle to watch.

=head2 C<connector>

  my $connector = $dbi->connector;
  $dbi = $dbi->connector($connector);

Connection manager object. if C<connector> is set, you can get C<dbh>
through connection manager. Conection manager object must have C<dbh> mehtod.

This is L<DBIx::Connector> example. Please pass
C<default_option> to L<DBIx::Connector> C<new> method.

  my $connector = DBIx::Connector->new(
    "dbi:mysql:database=$database",
    $user,
    $password,
    DBIx::Custom->new->default_option
  );
  
  my $dbi = DBIx::Custom->connect(connector => $connector);

If C<connector> is set to 1 when connect method is called,
L<DBIx::Connector> is automatically set to C<connector>

  my $dbi = DBIx::Custom->connect(
    dsn => $dsn, user => $user, password => $password, connector => 1);
  
  my $connector = $dbi->connector; # DBIx::Connector

Note that L<DBIx::Connector> must be installed.

=head2 C<dsn>

  my $dsn = $dbi->dsn;
  $dbi = $dbi->dsn("DBI:mysql:database=dbname");

Data source name, used when C<connect> method is executed.

=head2 C<default_option>

  my $default_option = $dbi->default_option;
  $dbi = $dbi->default_option($default_option);

L<DBI> default option, used when C<connect> method is executed,
default to the following values.

  {
    RaiseError => 1,
    PrintError => 0,
    AutoCommit => 1,
  }

=head2 C<exclude_table>

  my $exclude_table = $dbi->exclude_table;
  $dbi = $dbi->exclude_table(qr/pg_/);

Excluded table regex.
C<each_column>, C<each_table>, C<type_rule>,
and C<setup_model> methods ignore matching tables.

=head2 C<filters>

  my $filters = $dbi->filters;
  $dbi = $dbi->filters(\%filters);

Filters, registered by C<register_filter> method.

=head2 C<last_sql>

  my $last_sql = $dbi->last_sql;
  $dbi = $dbi->last_sql($last_sql);

Get last successed SQL executed by C<execute> method.

=head2 C<now>

  my $now = $dbi->now;
  $dbi = $dbi->now($now);

Code reference which return current time, default to the following code reference.

  sub {
    my ($sec, $min, $hour, $mday, $mon, $year) = localtime;
    $mon++;
    $year += 1900;
    return sprintf("%04d-%02d-%02d %02d:%02d:%02d");
  }

This return the time like C<2011-10-14 05:05:27>.

This is used by C<insert> method's C<created_at> option and C<updated_at> option,
and C<update> method's C<updated_at> option.

=head2 C<models>

  my $models = $dbi->models;
  $dbi = $dbi->models(\%models);

Models, included by C<include_model> method.

=head2 C<option>

  my $option = $dbi->option;
  $dbi = $dbi->option($option);

L<DBI> option, used when C<connect> method is executed.
Each value in option override the value of C<default_option>.

=head2 C<password>

  my $password = $dbi->password;
  $dbi = $dbi->password('lkj&le`@s');

Password, used when C<connect> method is executed.

=head2 C<query_builder>

  my $builder = $dbi->query_builder;

Creat query builder. This is L<DBIx::Custom::QueryBuilder>.

=head2 C<quote>

  my quote = $dbi->quote;
  $dbi = $dbi->quote('"');

Reserved word quote.
Default to double quote '"' except for mysql.
In mysql, default to back quote '`'

You can set quote pair.

  $dbi->quote('[]');

=head2 C<result_class>

  my $result_class = $dbi->result_class;
  $dbi = $dbi->result_class('DBIx::Custom::Result');

Result class, default to L<DBIx::Custom::Result>.

=head2 C<safety_character>

  my $safety_character = $dbi->safety_character;
  $dbi = $dbi->safety_character($character);

Regex of safety character for table and column name, default to '\w'.
Note that you don't have to specify like '[\w]'.

=head2 C<separator>

  my $separator = $dbi->separator;
  $dbi = $dbi->separator('-');

Separator which join table name and column name.
This have effect to C<column> and C<mycolumn> method,
and C<select> method's column option.

Default to C<.>.

=head2 C<tag_parse>

  my $tag_parse = $dbi->tag_parse(0);
  $dbi = $dbi->tag_parse;

Enable DEPRECATED tag parsing functionality, default to 1.
If you want to disable tag parsing functionality, set to 0.

=head2 C<user>

  my $user = $dbi->user;
  $dbi = $dbi->user('Ken');

User name, used when C<connect> method is executed.

=head2 C<user_column_info>

  my $user_column_info = $dbi->user_column_info;
  $dbi = $dbi->user_column_info($user_column_info);

You can set the date like the following one.

  [
    {table => 'book', column => 'title', info => {...}},
    {table => 'author', column => 'name', info => {...}}
  ]

Usually, you set return value of C<get_column_info>.

  my $user_column_info
    = $dbi->get_column_info(exclude_table => qr/^system/);
  $dbi->user_column_info($user_column_info);

If C<user_column_info> is set, C<each_column> use C<user_column_info>
to find column info. this is very fast.

=head2 C<user_table_info>

  my $user_table_info = $dbi->user_table_info;
  $dbi = $dbi->user_table_info($user_table_info);

You can set the following data.

  [
    {table => 'book', info => {...}},
    {table => 'author', info => {...}}
  ]

Usually, you can set return value of C<get_table_info>.

  my $user_table_info = $dbi->get_table_info(exclude => qr/^system/);
  $dbi->user_table_info($user_table_info);

If C<user_table_info> is set, C<each_table> use C<user_table_info>
to find table info.

=head1 METHODS

L<DBIx::Custom> inherits all methods from L<Object::Simple>
and use all methods of L<DBI>
and implements the following new ones.

=head2 C<available_datatype>

  print $dbi->available_datatype;

Get available data types. You can use these data types
in C<type rule>'s C<from1> and C<from2> section.

=head2 C<available_typename>

  print $dbi->available_typename;

Get available type names. You can use these type names in
C<type_rule>'s C<into1> and C<into2> section.

=head2 C<assign_clause>

  my $assign_clause = $dbi->assign_clause({title => 'a', age => 2});

Create assign clause

  title = :title, author = :author

This is used to create update clause.

  "update book set " . $dbi->assign_clause({title => 'a', age => 2});

=head2 C<async> EXPERIMENTAL

  async => sub {
    my ($dbi, $result) = @_;
    ...
  };

Database async access. L<AnyEvent> is required.

This is C<mysql> async access example.

  use AnyEvent;

  my $cond = AnyEvent->condvar;

  my $timer = AnyEvent->timer(
    interval => 1,
    cb => sub { 1 }
  );

  my $count = 0;

  $dbi->execute('SELECT SLEEP(1), 3', undef,
    prepare_attr => {async => 1}, statement => 'select',
    async => sub {
      my ($dbi, $result) = @_;
      my $row = $result->fetch_one;
      is($row->[1], 3, 'before');
      $cond->send if ++$count == 2;
    }
  );

  $dbi->select('key1', table => 'table1', prepare_attr => {async => 1},
    async => sub {
      my ($dbi, $result) = @_;
      my $row = $result->fetch_one;
      is($row->[0], 1, 'after1');
      $dbi->select('key1', table => 'table1', prepare_attr => {async => 1},
        async => sub {
          my ($dbi, $result) = @_;
          my $row = $result->fetch_one;
          is($row->[0], 1, 'after2');
          $cond->send if ++$count == 2;
        }
      )
    }
  );

  $cond->recv;

=head2 C<column>

  my $column = $dbi->column(book => ['author', 'title']);

Create column clause. The follwoing column clause is created.

  book.author as "book.author",
  book.title as "book.title"

You can change separator by C<separator> attribute.

  # Separator is hyphen
  $dbi->separator('-');
  
  book.author as "book-author",
  book.title as "book-title"
  
=head2 C<connect>

  my $dbi = DBIx::Custom->connect(
    dsn => "dbi:mysql:database=dbname",
    user => 'ken',
    password => '!LFKD%$&',
    option => {mysql_enable_utf8 => 1}
  );

Connect to the database and create a new L<DBIx::Custom> object.

L<DBIx::Custom> is a wrapper of L<DBI>.
C<AutoCommit> and C<RaiseError> options are true, 
and C<PrintError> option is false by default.

=head2 C<count>

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

Get rows count.

Options is same as C<select> method's ones.

=head2 C<create_model>

  my $model = $dbi->create_model(
    table => 'book',
    primary_key => 'id',
    join => [
      'inner join company on book.comparny_id = company.id'
    ],
  );

Create L<DBIx::Custom::Model> object and initialize model.
the module is also used from C<model> method.

 $dbi->model('book')->select(...);

=head2 C<dbh>

  my $dbh = $dbi->dbh;

Get L<DBI> database handle. if C<connector> is set, you can get
database handle through C<connector> object.

=head2 C<delete>

  $dbi->delete(table => 'book', where => {title => 'Perl'});

Execute delete statement.

The following opitons are available.

B<OPTIONS>

C<delete> method use all of C<execute> method's options,
and use the following new ones.

=over 4

=item C<id>

  id => 4
  id => [4, 5]

ID corresponding to C<primary_key>.
You can delete rows by C<id> and C<primary_key>.

  $dbi->delete(
    primary_key => ['id1', 'id2'],
    id => [4, 5],
    table => 'book',
  );

The above is same as the followin one.

  $dbi->delete(where => {id1 => 4, id2 => 5}, table => 'book');

=item C<prefix>

  prefix => 'some'

prefix before table name section.

  delete some from book

=item C<table>

  table => 'book'

Table name.

=item C<where>

Same as C<select> method's C<where> option.

=back

=head2 C<delete_all>

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

Execute delete statement for all rows.
Options is same as C<delete>.

=head2 C<each_column>

  $dbi->each_column(
    sub {
      my ($dbi, $table, $column, $column_info) = @_;
      
      my $type = $column_info->{TYPE_NAME};
      
      if ($type eq 'DATE') {
          # ...
      }
    }
  );

Iterate all column informations in database.
Argument is callback which is executed when one column is found.
Callback receive four arguments. C<DBIx::Custom object>, C<table name>,
C<column name>, and C<column information>.

If C<user_column_info> is set, C<each_column> method use C<user_column_info>
infromation, you can improve the performance of C<each_column> in
the following way.

  my $column_infos = $dbi->get_column_info(exclude_table => qr/^system_/);
  $dbi->user_column_info($column_info);
  $dbi->each_column(sub { ... });

=head2 C<each_table>

  $dbi->each_table(
    sub {
      my ($dbi, $table, $table_info) = @_;
      
      my $table_name = $table_info->{TABLE_NAME};
    }
  );

Iterate all table informationsfrom in database.
Argument is callback which is executed when one table is found.
Callback receive three arguments, C<DBIx::Custom object>, C<table name>,
C<table information>.

If C<user_table_info> is set, C<each_table> method use C<user_table_info>
infromation, you can improve the performance of C<each_table> in
the following way.

  my $table_infos = $dbi->get_table_info(exclude => qr/^system_/);
  $dbi->user_table_info($table_info);
  $dbi->each_table(sub { ... });

=head2 C<execute>

  my $result = $dbi->execute(
    "select * from book where title = :title and author like :author",
    {title => 'Perl', author => '%Ken%'}
  );

  my $result = $dbi->execute(
    "select * from book where title = :book.title and author like :book.author",
    {'book.title' => 'Perl', 'book.author' => '%Ken%'}
  );

Execute SQL. SQL can contain column parameter such as :author and :title.
You can append table name to column name such as :book.title and :book.author.
Second argunet is data, embedded into column parameter.
Return value is L<DBIx::Custom::Result> object when select statement is executed,
or the count of affected rows when insert, update, delete statement is executed.

Named placeholder such as C<:title> is replaced by placeholder C<?>.
  
  # Original
  select * from book where title = :title and author like :author
  
  # Replaced
  select * from where title = ? and author like ?;

You can specify operator with named placeholder
by C<name{operator}> syntax.

  # Original
  select * from book where :title{=} and :author{like}
  
  # Replaced
  select * from where title = ? and author like ?;

Note that colons in time format such as 12:13:15 is exeption,
it is not parsed as named placeholder.
If you want to use colon generally, you must escape it by C<\\>

  select * from where title = "aa\\:bb";

B<OPTIONS>

The following opitons are available.

=over 4

=item C<after_build_sql> 

You can filter sql after the sql is build.

  after_build_sql => $code_ref

The following one is one example.

  $dbi->select(
    table => 'book',
    column => 'distinct(name)',
    after_build_sql => sub {
      "select count(*) from ($_[0]) as t1"
    }
  );

The following SQL is executed.

  select count(*) from (select distinct(name) from book) as t1;

=item C<append>

  append => 'order by name'

Append some statement after SQL.

=item C<prepare_attr> EXPERIMENTAL

  prepare_attr => {async => 1}

Statemend handle attributes,
this is L<DBI>'s C<prepare> method second argument.

=item C<bind_type>

Specify database bind data type.

  bind_type => [image => DBI::SQL_BLOB]
  bind_type => [[qw/image audio/] => DBI::SQL_BLOB]

This is used to bind parameter by C<bind_param> of statment handle.

  $sth->bind_param($pos, $value, DBI::SQL_BLOB);

=item C<filter>
  
  filter => {
    title  => sub { uc $_[0] }
    author => sub { uc $_[0] }
  }

  # Filter name
  filter => {
    title  => 'upper_case',
    author => 'upper_case'
  }
      
  # At once
  filter => [
    [qw/title author/]  => sub { uc $_[0] }
  ]

Filter. You can set subroutine or filter name
registered by by C<register_filter>.
This filter is executed before data is saved into database.
and before type rule filter is executed.

=item C<query>

  query => 1

C<execute> method return hash reference which contain SQL and column
infromation

  my $sql = $query->{sql};
  my $columns = $query->{columns};
  
=item C<reuse>
  
  reuse => $hash_ref

Reuse query object if the hash reference variable is set.
  
  my $queries = {};
  $dbi->execute($sql, $param, reuse => $queries);

This will improved performance when you want to execute same query repeatedly
because generally creating query object is slow.

=item C<primary_key>

  primary_key => 'id'
  primary_key => ['id1', 'id2']

Priamry key. This is used for C<id> option.

=item C<select> EXPERIMETAL

  select => 1

If you set C<select> to 1, this statement become select statement
and return value is always L<DBIx::Custom::Result> object.

=item C<table>
  
  table => 'author'

If you want to omit table name in column name
and enable C<into1> and C<into2> type filter,
You must set C<table> option.

  $dbi->execute("select * from book where title = :title and author = :author",
    {title => 'Perl', author => 'Ken', table => 'book');

  # Same
  $dbi->execute(
    "select * from book where title = :book.title and author = :book.author",
    {title => 'Perl', author => 'Ken');

=item C<table_alias>

  table_alias => {user => 'worker'}

Table alias. Key is real table name, value is alias table name.
If you set C<table_alias>, you can enable C<into1> and C<into2> type rule
on alias table name.

=item C<type_rule_off>

  type_rule_off => 1

Turn C<into1> and C<into2> type rule off.

=item C<type_rule1_off>

  type_rule1_off => 1

Turn C<into1> type rule off.

=item C<type_rule2_off>

  type_rule2_off => 1

Turn C<into2> type rule off.

=back

=head2 C<get_column_info>

  my $column_infos = $dbi->get_column_info(exclude_table => qr/^system_/);

get column infomation except for one which match C<exclude_table> pattern.

  [
    {table => 'book', column => 'title', info => {...}},
    {table => 'author', column => 'name' info => {...}}
  ]

=head2 C<get_table_info>

  my $table_infos = $dbi->get_table_info(exclude => qr/^system_/);

get table infomation except for one which match C<exclude> pattern.

  [
    {table => 'book', info => {...}},
    {table => 'author', info => {...}}
  ]

You can set this value to C<user_table_info>.

=head2 C<helper>

  $dbi->helper(
    find_or_create   => sub {
      my $self = shift;
      
      # Process
    },
    ...
  );

Register helper. These helper is called directly from L<DBIx::Custom> object.

  $dbi->find_or_create;

=head2 C<insert>

  $dbi->insert({title => 'Perl', author => 'Ken'}, table  => 'book');

Execute insert statement. First argument is row data. Return value is
affected row count.

If you want to set constant value to row data, use scalar reference
as parameter value.

  {date => \"NOW()"}

You can pass multiple parameters, this is very fast.

  $dbi->insert(
    [
      {title => 'Perl', author => 'Ken'},
      {title => 'Ruby', author => 'Tom'}
    ],
    table  => 'book'
  );

In multiple insert, you can't use C<id> option.
and only first parameter is used to create sql.

B<options>

C<insert> method use all of C<execute> method's options,
and use the following new ones.

=over 4

=item C<bulk_insert>

  bulk_insert => 1

bulk insert is executed if database support bulk insert and 
multiple parameters is passed to C<insert>.
The SQL like the following one is executed.

  insert into book (id, title) values (?, ?), (?, ?);

=item C<ctime>

  ctime => 'created_time'

Created time column name. time when row is created is set to the column.
default time format is "YYYY-mm-dd HH:MM:SS", which can be changed by
C<now> attribute.

=item C<id>

  id => 4
  id => [4, 5]

ID corresponding to C<primary_key>.
You can insert a row by C<id> and C<primary_key>.

  $dbi->insert(
    {title => 'Perl', author => 'Ken'}
    primary_key => ['id1', 'id2'],
    id => [4, 5],
    table => 'book'
  );

The above is same as the followin one.

  $dbi->insert(
    {id1 => 4, id2 => 5, title => 'Perl', author => 'Ken'},
    table => 'book'
  );

=item C<prefix>

  prefix => 'or replace'

prefix before table name section

  insert or replace into book

=item C<table>

  table => 'book'

Table name.

=item C<mtime>

This option is same as C<update> method C<mtime> option.

=item C<wrap>

  wrap => {price => sub { "max($_[0])" }}

placeholder wrapped string.

If the following statement

  $dbi->insert({price => 100}, table => 'book',
    {price => sub { "$_[0] + 5" }});

is executed, the following SQL is executed.

  insert into book price values ( ? + 5 );

=back

=over 4

=head2 C<include_model>

  $dbi->include_model('MyModel');

Include models from specified namespace,
the following layout is needed to include models.

  lib / MyModel.pm
      / MyModel / book.pm
                / company.pm

Name space module, extending L<DBIx::Custom::Model>.

B<MyModel.pm>

  package MyModel;
  use DBIx::Custom::Model -base;
  
  1;

Model modules, extending name space module.

B<MyModel/book.pm>

  package MyModel::book;
  use MyModel -base;
  
  1;

B<MyModel/company.pm>

  package MyModel::company;
  use MyModel -base;
  
  1;
  
MyModel::book and MyModel::company is included by C<include_model>.

You can get model object by C<model>.

  my $book_model = $dbi->model('book');
  my $company_model = $dbi->model('company');

See L<DBIx::Custom::Model> to know model features.

=head2 C<like_value>

  my $like_value = $dbi->like_value

Code reference which return a value for the like value.

  sub { "%$_[0]%" }

=head2 C<mapper>

  my $mapper = $dbi->mapper(param => $param);

Create a new L<DBIx::Custom::Mapper> object.

=head2 C<merge_param>

  my $param = $dbi->merge_param({key1 => 1}, {key1 => 1, key2 => 2});

Merge parameters. The following new parameter is created.

  {key1 => [1, 1], key2 => 2}

If same keys contains, the value is converted to array reference.

=head2 C<model>

  my $model = $dbi->model('book');

Get a L<DBIx::Custom::Model> object
create by C<create_model> or C<include_model>

=head2 C<mycolumn>

  my $column = $dbi->mycolumn(book => ['author', 'title']);

Create column clause for myself. The follwoing column clause is created.

  book.author as author,
  book.title as title

=head2 C<new>

  my $dbi = DBIx::Custom->new(
    dsn => "dbi:mysql:database=dbname",
    user => 'ken',
    password => '!LFKD%$&',
    option => {mysql_enable_utf8 => 1}
  );

Create a new L<DBIx::Custom> object.

=head2 C<not_exists>

  my $not_exists = $dbi->not_exists;

DBIx::Custom::NotExists object, indicating the column is not exists.
This is used in C<param> of L<DBIx::Custom::Where> .

=head2 C<order>

  my $order = $dbi->order;

Create a new L<DBIx::Custom::Order> object.

=head2 C<q>

  my $quooted = $dbi->q("title");

Quote string by value of C<quote>.

=head2 C<register_filter>

  $dbi->register_filter(
    # Time::Piece object to database DATE format
    tp_to_date => sub {
      my $tp = shift;
      return $tp->strftime('%Y-%m-%d');
    },
    # database DATE format to Time::Piece object
    date_to_tp => sub {
      my $date = shift;
      return Time::Piece->strptime($date, '%Y-%m-%d');
    }
  );
  
Register filters, used by C<filter> option of many methods.

=head2 C<select>

  my $result = $dbi->select(
    column => ['author', 'title'],
    table  => 'book',
    where  => {author => 'Ken'},
  );
  
Execute select statement.

You can pass odd number arguments. first argument is C<column>.

  my $result = $dbi->select(['author', 'title'], table => 'book');

B<OPTIONS>

C<select> method use all of C<execute> method's options,
and use the following new ones.

=over 4

=item C<column>
  
  column => 'author'
  column => ['author', 'title']

Column clause.
  
if C<column> is not specified, '*' is set.

  column => '*'

You can specify hash of array reference.

  column => [
    {book => [qw/author title/]},
    {person => [qw/name age/]}
  ]

This is expanded to the following one by using C<colomn> method.

  book.author as "book.author",
  book.title as "book.title",
  person.name as "person.name",
  person.age as "person.age"

You can specify array of array reference, first argument is
column name, second argument is alias.

  column => [
    ['date(book.register_datetime)' => 'book.register_date']
  ];

Alias is quoted properly and joined.

  date(book.register_datetime) as "book.register_date"

=item C<id>

  id => 4
  id => [4, 5]

ID corresponding to C<primary_key>.
You can select rows by C<id> and C<primary_key>.

  $dbi->select(
    primary_key => ['id1', 'id2'],
    id => [4, 5],
    table => 'book'
  );

The above is same as the followin one.

  $dbi->select(
    where => {id1 => 4, id2 => 5},
    table => 'book'
  );
  
=item C<param>

  param => {'table2.key3' => 5}

Parameter shown before where clause.
  
For example, if you want to contain named placeholder in join clause, 
you can pass parameter by C<param> option.

  join  => ['inner join (select * from table2 where table2.key3 = :table2.key3)' . 
            ' as table2 on table1.key1 = table2.key1']

=itme C<prefix>

  prefix => 'SQL_CALC_FOUND_ROWS'

Prefix of column cluase

  select SQL_CALC_FOUND_ROWS title, author from book;

=item C<join>

  join => [
    'left outer join company on book.company_id = company_id',
    'left outer join location on company.location_id = location.id'
  ]
      
Join clause. If column cluase or where clause contain table name like "company.name",
join clausees needed when SQL is created is used automatically.

  $dbi->select(
    table => 'book',
    column => ['company.location_id as location_id'],
    where => {'company.name' => 'Orange'},
    join => [
      'left outer join company on book.company_id = company.id',
      'left outer join location on company.location_id = location.id'
    ]
  );

In above select, column and where clause contain "company" table,
the following SQL is created

  select company.location_id as location_id
  from book
    left outer join company on book.company_id = company.id
  where company.name = ?;

You can specify two table by yourself. This is useful when join parser can't parse
the join clause correctly.

  $dbi->select(
    table => 'book',
    column => ['company.location_id as location_id'],
    where => {'company.name' => 'Orange'},
    join => [
      {
        clause => 'left outer join location on company.location_id = location.id',
        table => ['company', 'location']
      }
    ]
  );

=item C<table>

  table => 'book'

Table name.

=item C<where>
  
  # (1) Hash reference
  where => {author => 'Ken', 'title' => ['Perl', 'Ruby']}
  # -> where author = 'Ken' and title in ('Perl', 'Ruby')
  
  # (2) DBIx::Custom::Where object
  where => $dbi->where(
    clause => ['and', ':author{=}', ':title{like}'],
    param  => {author => 'Ken', title => '%Perl%'}
  )
  # -> where author = 'Ken' and title like '%Perl%'
  
  # (3) Array reference[Array refenrece, Hash reference]
  where => [
    ['and', ':author{=}', ':title{like}'],
    {author => 'Ken', title => '%Perl%'}
  ]
  # -> where author = 'Ken' and title like '%Perl%'
  
  # (4) Array reference[String, Hash reference]
  where => [
    ':author{=} and :title{like}',
    {author => 'Ken', title => '%Perl%'}
  ]
  #  -> where author = 'Ken' and title like '%Perl%'
  
  # (5) String
  where => 'title is null'
  #  -> where title is null

Where clause. See L<DBIx::Custom::Where>.
  
=back

=head2 C<setup_model>

  $dbi->setup_model;

Setup all model objects.
C<columns> of model object is automatically set, parsing database information.

=head2 C<type_rule>

  $dbi->type_rule(
    into1 => {
      date => sub { ... },
      datetime => sub { ... }
    },
    into2 => {
      date => sub { ... },
      datetime => sub { ... }
    },
    from1 => {
      # DATE
      9 => sub { ... },
      # DATETIME or TIMESTAMP
      11 => sub { ... },
    }
    from2 => {
      # DATE
      9 => sub { ... },
      # DATETIME or TIMESTAMP
      11 => sub { ... },
    }
  );

Filtering rule when data is send into and get from database.
This has a little complex problem.

In C<into1> and C<into2> you can specify
type name as same as type name defined
by create table, such as C<DATETIME> or C<DATE>.

Note that type name and data type don't contain upper case.
If these contain upper case charactor, you convert it to lower case.

C<into2> is executed after C<into1>.

Type rule of C<into1> and C<into2> is enabled on the following
column name.

=over 4

=item 1. column name

  issue_date
  issue_datetime

This need C<table> option in each method.

=item 2. table name and column name, separator is dot

  book.issue_date
  book.issue_datetime

=back

You get all type name used in database by C<available_typename>.

  print $dbi->available_typename;

In C<from1> and C<from2> you specify data type, not type name.
C<from2> is executed after C<from1>.
You get all data type by C<available_datatype>.

  print $dbi->available_datatype;

You can also specify multiple types at once.

  $dbi->type_rule(
    into1 => [
      [qw/DATE DATETIME/] => sub { ... },
    ],
  );

=head2 C<update>

  $dbi->update({title => 'Perl'}, table  => 'book', where  => {id => 4});

Execute update statement. First argument is update row data.

If you want to set constant value to row data, use scalar reference
as parameter value.

  {date => \"NOW()"}

B<OPTIONS>

C<update> method use all of C<execute> method's options,
and use the following new ones.

=over 4

=item C<id>

  id => 4
  id => [4, 5]

ID corresponding to C<primary_key>.
You can update rows by C<id> and C<primary_key>.

  $dbi->update(
    {title => 'Perl', author => 'Ken'}
    primary_key => ['id1', 'id2'],
    id => [4, 5],
    table => 'book'
  );

The above is same as the followin one.

  $dbi->update(
    {title => 'Perl', author => 'Ken'}
    where => {id1 => 4, id2 => 5},
    table => 'book'
  );

=item C<prefix>

  prefix => 'or replace'

prefix before table name section

  update or replace book

=item C<table>

  table => 'book'

Table name.

=item C<where>

Same as C<select> method's C<where> option.

=item C<wrap>

  wrap => {price => sub { "max($_[0])" }}

placeholder wrapped string.

If the following statement

  $dbi->update({price => 100}, table => 'book',
    {price => sub { "$_[0] + 5" }});

is executed, the following SQL is executed.

  update book set price =  ? + 5;

=item C<mtime>

  mtime => 'modified_time'

Modified time column name. time row is updated is set to the column.
default time format is C<YYYY-mm-dd HH:MM:SS>, which can be changed by
C<now> attribute.

=back

=head2 C<update_all>

  $dbi->update_all({title => 'Perl'}, table => 'book', );

Execute update statement for all rows.
Options is same as C<update> method.

=head2 C<update_or_insert>
  
  # ID
  $dbi->update_or_insert(
    {title => 'Perl'},
    table => 'book',
    id => 1,
    primary_key => 'id',
    option => {
      select => {
         append => 'for update'
      }
    }
  );

Update or insert.

C<update_or_insert> method execute C<select> method first to find row.
If the row is exists, C<update> is executed.
If not, C<insert> is executed.

C<OPTIONS>

C<update_or_insert> method use all common option
in C<select>, C<update>, C<delete>, and has the following new ones.

=over 4

=item C<option>

  option => {
    select => {
      append => '...'
    },
    insert => {
      prefix => '...'
    },
    update => {
      filter => {}
    }
  }

If you want to pass option to each method,
you can use C<option> option.

=over 4

=item C<select_option>

  select_option => {append => 'for update'}

select method option,
select method is used to check the row is already exists.

=head2 C<show_datatype>

  $dbi->show_datatype($table);

Show data type of the columns of specified table.

  book
  title: 5
  issue_date: 91

This data type is used in C<type_rule>'s C<from1> and C<from2>.

=head2 C<show_tables>

  $dbi->show_tables;

Show tables.

=head2 C<show_typename>

  $dbi->show_typename($table);

Show type name of the columns of specified table.

  book
  title: varchar
  issue_date: date

This type name is used in C<type_rule>'s C<into1> and C<into2>.

=head2 C<values_clause>

  my $values_clause = $dbi->values_clause({title => 'a', age => 2});

Create values clause.

  (title, author) values (title = :title, age = :age);

You can use this in insert statement.

  my $insert_sql = "insert into book $values_clause";

=head2 C<where>

  my $where = $dbi->where(
    clause => ['and', 'title = :title', 'author = :author'],
    param => {title => 'Perl', author => 'Ken'}
  );

Create a new L<DBIx::Custom::Where> object.

=head1 ENVIRONMENTAL VARIABLES

=head2 C<DBIX_CUSTOM_DEBUG>

If environment variable C<DBIX_CUSTOM_DEBUG> is set to true,
executed SQL and bind values are printed to STDERR.

=head2 C<DBIX_CUSTOM_DEBUG_ENCODING>

DEBUG output encoding. Default to UTF-8.

=head2 C<DBIX_CUSTOM_TAG_PARSE>

If you set DBIX_CUSTOM_TAG_PARSE to 0, tag parsing is off.

=head2 C<DBIX_CUSTOM_DISABLE_MODEL_EXECUTE>

If you set DBIX_CUSTOM_DISABLE_MODEL_EXECUTE to 1,
L<DBIx::Custom::Model> execute method call L<DBIx::Custom> execute.

=head1 DEPRECATED FUNCTIONALITY

L<DBIx::Custom>

  # Attribute methods
  tag_parse # will be removed 2017/1/1
  default_dbi_option # will be removed 2017/1/1
  dbi_option # will be removed 2017/1/1
  data_source # will be removed at 2017/1/1
  dbi_options # will be removed at 2017/1/1
  filter_check # will be removed at 2017/1/1
  reserved_word_quote # will be removed at 2017/1/1
  cache_method # will be removed at 2017/1/1
  
  # Methods
  update_timestamp # will be removed at 2017/1/1
  insert_timestamp # will be removed at 2017/1/1
  method # will be removed at 2017/1/1
  assign_param # will be removed at 2017/1/1
  update_param # will be removed at 2017/1/1
  insert_param # will be removed at 2017/1/1
  create_query # will be removed at 2017/1/1
  apply_filter # will be removed at 2017/1/1
  select_at # will be removed at 2017/1/1
  delete_at # will be removed at 2017/1/1
  update_at # will be removed at 2017/1/1
  insert_at # will be removed at 2017/1/1
  register_tag # will be removed at 2017/1/1
  default_bind_filter # will be removed at 2017/1/1
  default_fetch_filter # will be removed at 2017/1/1
  insert_param_tag # will be removed at 2017/1/1
  register_tag # will be removed at 2017/1/1
  register_tag_processor # will be removed at 2017/1/1
  update_param_tag # will be removed at 2017/1/1
  
  # Options
  insert method created_at option # will be removed 2017/3/1
  update method updated_at option # will be removed 2017/3/1
  select column option [COLUMN => ALIAS] syntax # will be removed 2017/1/1
  execute method id option # will be removed 2017/1/1
  update timestamp option # will be removed 2017/1/1
  insert timestamp option # will be removed 2017/1/1
  select method where_param option # will be removed 2017/1/1
  delete method where_param option # will be removed 2017/1/1
  update method where_param option # will be removed 2017/1/1
  insert method param option # will be removed at 2017/1/1
  insert method id option # will be removed at 2017/1/1
  select method relation option # will be removed at 2017/1/1
  select method column option [COLUMN, as => ALIAS] format
    # will be removed at 2017/1/1
  execute method's sqlfilter option # will be removed at 2017/1/1
  
  # Others
  execute($query, ...) # execute method receiving query object.
                       # this is removed at 2017/1/1
  execute("select * from {= title}"); # execute method's
                                      # tag parsing functionality
                                      # will be removed at 2017/1/1
  Query caching # will be removed at 2017/1/1

L<DBIx::Custom::Model>

  # Attribute methods
  execute # will be removed at 2017/1/1
  method # will be removed at 2017/1/1
  filter # will be removed at 2017/1/1
  name # will be removed at 2017/1/1
  type # will be removed at 2017/1/1

L<DBIx::Custom::Query>

This module is DEPRECATED! # will be removed at 2017/1/1
  
  # Attribute methods
  default_filter # will be removed at 2017/1/1
  table # will be removed at 2017/1/1
  filters # will be removed at 2017/1/1
  
  # Methods
  filter # will be removed at 2017/1/1

L<DBIx::Custom::QueryBuilder>

This module is DEPRECATED! # will be removed at 2017/1/1
  
  # Attribute methods
  tags # will be removed at 2017/1/1
  tag_processors # will be removed at 2017/1/1
  
  # Methods
  register_tag # will be removed at 2017/1/1
  register_tag_processor # will be removed at 2017/1/1
  
  # Others
  build_query("select * from {= title}"); # tag parsing functionality
                                          # will be removed at 2017/1/1

L<DBIx::Custom::Result>
  
  # Attribute methods
  filter_check # will be removed at 2017/1/1
  
  # Methods
  fetch_first # will be removed at 2017/2/1
  fetch_hash_first # will be removed 2017/2/1
  filter_on # will be removed at 2017/1/1
  filter_off # will be removed at 2017/1/1
  end_filter # will be removed at 2017/1/1
  remove_end_filter # will be removed at 2017/1/1
  remove_filter # will be removed at 2017/1/1
  default_filter # will be removed at 2017/1/1

L<DBIx::Custom::Tag>

  This module is DEPRECATED! # will be removed at 2017/1/1

L<DBIx::Custom::Order>

  # Other
  prepend method array reference receiving
    $order->prepend(['book', 'desc']); # will be removed 2017/1/1

=head1 BACKWARDS COMPATIBILITY POLICY

If a functionality is DEPRECATED, you can know it by DEPRECATED warnings
except for attribute method.
You can check all DEPRECATED functionalities by document.
DEPRECATED functionality is removed after five years,
but if at least one person use the functionality and tell me that thing
I extend one year each time he tell me it.

EXPERIMENTAL functionality will be changed without warnings.

=head1 BUGS

Please tell me bugs if found.

C<< <kimoto.yuki at gmail.com> >>

L<http://github.com/yuki-kimoto/DBIx-Custom>

=head1 AUTHOR

Yuki Kimoto, C<< <kimoto.yuki at gmail.com> >>

=head1 COPYRIGHT & LICENSE

Copyright 2009-2011 Yuki Kimoto, all rights reserved.

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

=cut