|
0
|
1
|
|
|
2 =head1 LICENSE
|
|
|
3
|
|
|
4 Copyright (c) 1999-2012 The European Bioinformatics Institute and
|
|
|
5 Genome Research Limited. All rights reserved.
|
|
|
6
|
|
|
7 This software is distributed under a modified Apache license.
|
|
|
8 For license details, please see
|
|
|
9
|
|
|
10 http://www.ensembl.org/info/about/code_licence.html
|
|
|
11
|
|
|
12 =head1 CONTACT
|
|
|
13
|
|
|
14 Please email comments or questions to the public Ensembl
|
|
|
15 developers list at <dev@ensembl.org>.
|
|
|
16
|
|
|
17 Questions may also be sent to the Ensembl help desk at
|
|
|
18 <helpdesk@ensembl.org>.
|
|
|
19
|
|
|
20 =cut
|
|
|
21
|
|
|
22 =head1 NAME
|
|
|
23
|
|
|
24 Bio::EnsEMBL::Utils::CliHelper
|
|
|
25
|
|
|
26 =head1 VERSION
|
|
|
27
|
|
|
28 $Revision: 1.6 $
|
|
|
29
|
|
|
30 =head1 SYNOPSIS
|
|
|
31
|
|
|
32 use Bio::EnsEMBL::Utils::CliHelper;
|
|
|
33
|
|
|
34 my $cli = Bio::EnsEMBL::Utils::CliHelper->new();
|
|
|
35
|
|
|
36 # get the basic options for connecting to a database server
|
|
|
37 my $optsd = $cli->get_dba_opts();
|
|
|
38
|
|
|
39 # add the print option
|
|
|
40 push(@$optsd,"print|p");
|
|
|
41
|
|
|
42 # process the command line with the supplied options plus a reference to a help subroutine
|
|
|
43 my $opts = $cli->process_args($optsd,\&usage);
|
|
|
44
|
|
|
45 # use the command line options to get an array of database details
|
|
|
46 for my $db_args (@{$cli->get_dba_args_for_opts($opts)}) {
|
|
|
47 # use the args to create a DBA
|
|
|
48 my $dba = new Bio::EnsEMBL::DBSQL::DBAdaptor(%{$db_args});
|
|
|
49 ...
|
|
|
50 }
|
|
|
51
|
|
|
52 For adding secondary databases, a prefix can be supplied. For instance, to add a second set of
|
|
|
53 db params prefixed with dna (-dnahost -dbport etc.) use the prefix argument with get_dba_opts and
|
|
|
54 get_dba_args_for_opts:
|
|
|
55 # get the basic options for connecting to a database server
|
|
|
56 my $optsd =
|
|
|
57 [ @{ $cli_helper->get_dba_opts() }, @{ $cli_helper->get_dba_opts('gc') } ];
|
|
|
58 # process the command line with the supplied options plus a help subroutine
|
|
|
59 my $opts = $cli_helper->process_args( $optsd, \&usage );
|
|
|
60 # get the dna details
|
|
|
61 my ($dna_dba_details) =
|
|
|
62 @{ $cli_helper->get_dba_args_for_opts( $opts, 1, 'dna' ) };
|
|
|
63 my $dna_db =
|
|
|
64 Bio::EnsEMBL::DBSQL::DBAdaptor->new( %{$dna_dba_details} ) );
|
|
|
65
|
|
|
66 =head1 DESCRIPTION
|
|
|
67
|
|
|
68 Utilities for a more consistent approach to parsing and handling EnsEMBL script command lines
|
|
|
69
|
|
|
70 =head1 METHODS
|
|
|
71
|
|
|
72 See subroutines.
|
|
|
73
|
|
|
74 =cut
|
|
|
75
|
|
|
76 package Bio::EnsEMBL::Utils::CliHelper;
|
|
|
77
|
|
|
78 use warnings;
|
|
|
79 use strict;
|
|
|
80
|
|
|
81 use Carp;
|
|
|
82 use Data::Dumper;
|
|
|
83 use Getopt::Long qw(:config auto_version no_ignore_case);
|
|
|
84
|
|
|
85 use Bio::EnsEMBL::Registry;
|
|
|
86 use Bio::EnsEMBL::DBSQL::DBConnection;
|
|
|
87 use Bio::EnsEMBL::DBSQL::DBAdaptor;
|
|
|
88
|
|
|
89 my $dba_opts =
|
|
|
90 [ { args => [ 'host', 'dbhost', 'h' ], type => '=s' },
|
|
|
91 { args => [ 'port', 'dbport', 'P' ], type => ':i' },
|
|
|
92 { args => [ 'user', 'dbuser', 'u' ], type => '=s' },
|
|
|
93 { args => [ 'pass', 'dbpass', 'p' ], type => ':s' },
|
|
|
94 { args => [ 'dbname', 'D' ], type => ':s' },
|
|
|
95 { args => [ 'pattern', 'dbpattern' ], type => ':s' },
|
|
|
96 { args => [ 'driver' ], type => ':s' },
|
|
|
97 { args => [ 'species_id' ], type => ':i' },
|
|
|
98 { args => [ 'species' ], type => ':i' },
|
|
|
99 ];
|
|
|
100
|
|
|
101 =head2 new()
|
|
|
102
|
|
|
103 Description : Construct a new instance of a CliHelper object
|
|
|
104 Returntype : Bio::EnsEMBL::Utils:CliHelper
|
|
|
105 Status : Under development
|
|
|
106
|
|
|
107 =cut
|
|
|
108
|
|
|
109 sub new {
|
|
|
110 my ( $class, @args ) = @_;
|
|
|
111 my $self = bless( {}, ref($class) || $class );
|
|
|
112 return $self;
|
|
|
113 }
|
|
|
114
|
|
|
115 =head2 get_dba_opts()
|
|
|
116
|
|
|
117 Arg [1] : Optional prefix for dbnames e.g. dna
|
|
|
118 Description : Retrieves the standard options for connecting to one or more Ensembl databases
|
|
|
119 Returntype : Arrayref of option definitions
|
|
|
120 Status : Under development
|
|
|
121
|
|
|
122 =cut
|
|
|
123
|
|
|
124 sub get_dba_opts {
|
|
|
125 my ( $self, $prefix ) = @_;
|
|
|
126 $prefix ||= '';
|
|
|
127 my @dba_opts = map {
|
|
|
128 my $opt = join '|', map { $prefix . $_ } @{ $_->{args} };
|
|
|
129 $opt . $_->{type};
|
|
|
130 } @{$dba_opts};
|
|
|
131 return \@dba_opts;
|
|
|
132 }
|
|
|
133
|
|
|
134 =head2 process_args()
|
|
|
135
|
|
|
136 Arg [1] : Arrayref of supported command line options (e.g. from get_dba_opts)
|
|
|
137 Arg [2] : Ref to subroutine to be invoked when -help or -? is supplied
|
|
|
138 Description : Retrieves the standard options for connecting to one or more Ensembl databases
|
|
|
139 Returntype : Hashref of parsed options
|
|
|
140 Status : Under development
|
|
|
141
|
|
|
142 =cut
|
|
|
143
|
|
|
144 sub process_args {
|
|
|
145 my ( $self, $opts_def, $usage_sub ) = @_;
|
|
|
146 my $opts = {};
|
|
|
147 push @{$opts_def}, q/help|?/ => $usage_sub;
|
|
|
148 GetOptions( $opts, @{$opts_def} )
|
|
|
149 || croak 'Could not parse command line arguments';
|
|
|
150 return $opts;
|
|
|
151 }
|
|
|
152
|
|
|
153 =head2 get_dba_args_for_opts()
|
|
|
154
|
|
|
155 Arg [1] : Hash of options (e.g. parsed from command line options by process_args())
|
|
|
156 Arg [2] : If set to 1, the databases are assumed to have a single species only. Default is 0.
|
|
|
157 Arg [3] : Optional prefix to use when parsing e.g. dna
|
|
|
158 Description : Uses the parsed command line options to generate an array of DBAdaptor arguments
|
|
|
159 : (e.g. expands dbpattern, finds all species_ids for multispecies databases)
|
|
|
160 : These can then be passed directly to Bio::EnsEMBL::DBSQL::DBAdaptor->new()
|
|
|
161 Returntype : Arrayref of DBA argument hash refs
|
|
|
162 Status : Under development
|
|
|
163
|
|
|
164 =cut
|
|
|
165
|
|
|
166 sub get_dba_args_for_opts {
|
|
|
167 my ( $self, $opts, $single_species, $prefix ) = @_;
|
|
|
168 $prefix ||= '';
|
|
|
169 $single_species ||= 0;
|
|
|
170 my ( $host, $port, $user, $pass, $dbname, $pattern, $driver ) =
|
|
|
171 map { $prefix . $_ } qw(host port user pass dbname pattern driver);
|
|
|
172 my @db_args;
|
|
|
173 if ( defined $opts->{$host} ) {
|
|
|
174 my $dbc =
|
|
|
175 Bio::EnsEMBL::DBSQL::DBConnection->new( -USER => $opts->{$user},
|
|
|
176 -PASS => $opts->{$pass},
|
|
|
177 -HOST => $opts->{$host},
|
|
|
178 -PORT => $opts->{$port},
|
|
|
179 -DRIVER => $opts->{$driver} );
|
|
|
180 my @dbnames;
|
|
|
181 if ( defined $opts->{$dbname} ) {
|
|
|
182 push @dbnames, $opts->{$dbname};
|
|
|
183 } elsif ( defined $opts->{$pattern} ) {
|
|
|
184 # get a basic DBConnection and use to find out which dbs are involved
|
|
|
185 @dbnames =
|
|
|
186 grep { m/$opts->{pattern}/smx }
|
|
|
187 @{ $dbc->sql_helper()->execute_simple(q/SHOW DATABASES/) };
|
|
|
188 } else {
|
|
|
189 print Dumper($opts);
|
|
|
190 croak 'dbname or dbpattern arguments required';
|
|
|
191 }
|
|
|
192 for my $dbname (@dbnames) {
|
|
|
193
|
|
|
194 #Decipher group of DBAdaptor by capturing the name_name(_name?)_core_ code. Otherwise we don't know
|
|
|
195 my ($group) = $dbname =~ /^[a-z]+_[a-z0-9]+(?:_[a-z0-9]+)?_([a-z]+)(?:_\d+)?_\d+/;
|
|
|
196
|
|
|
197 my $multi = 0;
|
|
|
198 my $species_ids = [ [ 1, undef ] ];
|
|
|
199 if ( !$single_species ) {
|
|
|
200 $species_ids =
|
|
|
201 $dbc->sql_helper()
|
|
|
202 ->execute(
|
|
|
203 "SELECT species_id,meta_value FROM $dbname.meta WHERE meta_key='species.production_name'"
|
|
|
204 );
|
|
|
205 if ( scalar( @{$species_ids} ) > 1 ) {
|
|
|
206 $multi = 1;
|
|
|
207 }
|
|
|
208 if ( defined $opts->{species_id} ) {
|
|
|
209 $species_ids =
|
|
|
210 [ [ $opts->{species_id}, $opts->{species} ] ];
|
|
|
211 }
|
|
|
212 }
|
|
|
213 for my $species_id ( @{$species_ids} ) {
|
|
|
214 my $args = {
|
|
|
215 -HOST => $opts->{$host},
|
|
|
216 -USER => $opts->{$user},
|
|
|
217 -PORT => $opts->{$port},
|
|
|
218 -PASS => $opts->{$pass},
|
|
|
219 -DBNAME => $dbname,
|
|
|
220 -DRIVER => $opts->{$driver},
|
|
|
221 -SPECIES_ID => $species_id->[0],
|
|
|
222 -SPECIES => $species_id->[1],
|
|
|
223 -MULTISPECIES_DB => $multi };
|
|
|
224 $args->{-GROUP} = $group if $group;
|
|
|
225 push(@db_args, $args);
|
|
|
226 }
|
|
|
227 }
|
|
|
228 } ## end if ( defined $opts->{$host...})
|
|
|
229 else {
|
|
|
230 croak '(db)host arguments required';
|
|
|
231 }
|
|
|
232 return \@db_args;
|
|
|
233 } ## end sub get_dba_args_for_opts
|
|
|
234
|
|
|
235 =head2 get_dba_args_for_opts()
|
|
|
236
|
|
|
237 Arg [1] : Hash of options (e.g. parsed from command line options by process_args())
|
|
|
238 Arg [2] : If set to 1, the databases are assumed to have a single species only. Default is 0.
|
|
|
239 Arg [3] : Optional prefix to use when parsing e.g. dna
|
|
|
240 Description : Uses the parsed command line options to generate an array DBAdaptors.
|
|
|
241 : Note this can overload connections on a server
|
|
|
242 Returntype : Arrayref of Bio::EnsEMBL::DBSQL::DBAdaptor
|
|
|
243 Status : Under development
|
|
|
244
|
|
|
245 =cut
|
|
|
246
|
|
|
247 sub get_dbas_for_opts {
|
|
|
248 my ( $self, $opts, $single_species, $prefix ) = @_;
|
|
|
249
|
|
|
250 # get all the DBA details that we want to work with and create DBAs for each in turn
|
|
|
251 my $dbas;
|
|
|
252 for my $args (
|
|
|
253 @{ $self->get_dba_args_for_opts( $opts, $single_species, $prefix ) } )
|
|
|
254 {
|
|
|
255 push @{$dbas}, Bio::EnsEMBL::DBSQL::DBAdaptor->new( %{$args} );
|
|
|
256 }
|
|
|
257 return $dbas;
|
|
|
258 }
|
|
|
259
|
|
|
260 =head2 load_registry_for_opts
|
|
|
261
|
|
|
262 Arg [1] : Hash of options (e.g. parsed from command line options by process_args())
|
|
|
263 Arg [2] : Optional prefix to use when parsing e.g. dna or master
|
|
|
264 Description : Loads a Registry from the given options hash. If a C<registry>
|
|
|
265 option is given then the code will call C<load_all>. Otherwise
|
|
|
266 we use the database parameters given to call
|
|
|
267 C<load_registry_from_db()>.
|
|
|
268 Returntype : Integer of the number of DBAdaptors loaded
|
|
|
269 Status : Under development
|
|
|
270
|
|
|
271 =cut
|
|
|
272
|
|
|
273 sub load_registry_for_opts {
|
|
|
274 my ($self, $opts, $prefix) = @_;
|
|
|
275 $prefix ||= q{};
|
|
|
276 if($opts->{registry}) {
|
|
|
277 my $location = $opts->{registry};
|
|
|
278 return Bio::EnsEMBL::Registry->load_all($location);
|
|
|
279 }
|
|
|
280 my ( $host, $port, $user, $pass ) = map { $prefix . $_ } qw(host port user pass);
|
|
|
281 my %args = (
|
|
|
282 -HOST => $opts->{$host},
|
|
|
283 -PORT => $opts->{$port},
|
|
|
284 -USER => $opts->{$user},
|
|
|
285 );
|
|
|
286 $args{-PASS} = $opts->{$pass};
|
|
|
287 return Bio::EnsEMBL::Registry->load_registry_from_db(%args);
|
|
|
288 }
|
|
|
289
|
|
|
290 1;
|
