9 use Text::Tradition::Directory;
10 use Text::Tradition::StemmaUtil qw/ character_input phylip_pars /;
12 binmode STDERR, ":utf8";
13 binmode STDOUT, ":utf8";
14 eval { no warnings; binmode $DB::OUT, ":utf8"; };
16 # Variables with defaults
17 my( $informat, $outformat, $language, $name, $sep, $dsn ) = ( '', '', 'Default',
18 'Tradition', "\t", "dbi:SQLite:dbname=db/traditions.db" );
19 # Variables with no default
20 my( $inbase, $help, $stemmafile, $dbuser, $dbpass, $from, $to, $dbid,
21 $nocalc, $nonlinear );
23 GetOptions( 'i|in=s' => \$informat,
24 'b|base=s' => \$inbase,
25 'o|out=s' => \$outformat,
26 'l|language=s' => \$language,
29 's|stemma=s' => \$stemmafile,
30 'u|user=s' => \$dbuser,
31 'p|pass=s' => \$dbpass,
34 'nl|nonlinear' => \$nonlinear,
38 'nc|nocalc' => \$nocalc,
45 unless( $informat =~ /^(CSV|CTE|KUL|Self|TEI|CollateX|tab(ular)?)|xlsx?|db$/i ) {
46 help( "Input format must be one of CollateX, CSV, CTE, Self, TEI" );
48 my $excel = $informat =~ /^xls/i ? lc( $informat ) : undef;
49 $informat = 'CollateX' if $informat =~ /^c(ollate)?x$/i;
50 $informat = 'KUL' if $informat =~ /^kul$/i;
51 $informat = 'CTE' if $informat =~ /^cte$/i;
52 $informat = 'Self' if $informat =~ /^self$/i;
53 $informat = 'TEI' if $informat =~ /^tei$/i;
54 $informat = 'Tabular' if $informat =~ /^tab$/i;
55 $informat = 'CollateText' if $informat =~ /^stone$/i;
56 $informat = 'Tabular' if $informat =~ /^xls/i;
58 unless( $outformat =~ /^(graphml|svg|dot|stemma|csv|db)$/ ) {
59 help( "Output format must be one of db, graphml, svg, csv, stemma, or dot" );
63 help( "Subgraphs only supported in GraphML, dot, or SVG format" )
64 unless $outformat =~ /^(graphml|dot|svg)$/;
67 # Do we have a base if we need it?
68 if( $informat =~ /^(KUL|CollateText)$/ && !$inbase ) {
69 help( "$informat input needs a base text" );
71 $sep = "\t" if $sep eq 'tab';
76 if( $informat eq 'db' ) {
77 my $dbargs = { dsn => $dsn };
78 $dbargs->{'extra_args'}->{'user'} = $dbuser if $dbuser;
79 $dbargs->{'extra_args'}->{'password'} = $dbpass if $dbpass;
80 $dir = Text::Tradition::Directory->new( $dbargs );
81 my $scope = $dir->new_scope();
82 $tradition = $dir->lookup( $input );
84 # First: read the base. Make a graph, but also note which
85 # nodes represent line beginnings.
86 my %args = ( 'input' => $informat,
88 $args{'linear'} = 0 if $nonlinear;
89 $args{'base'} = $inbase if $inbase;
90 $args{'language'} = $language if $language;
91 $args{'name'} = $name if $name;
92 $args{'nocalc'} = 1 if $nocalc;
93 if( $informat eq 'Tabular' ) {
95 $args{'excel'} = $excel;
97 $args{'sep_char'} = $sep;
100 # If we are writing to the database, use that DB as the userstore.
101 if( $outformat eq 'db' ) {
103 my $extra_args = { 'create' => 1 };
104 $extra_args->{'user'} = $dbuser if $dbuser;
105 $extra_args->{'password'} = $dbpass if $dbpass;
106 $dir = Text::Tradition::Directory->new( 'dsn' => $dsn,
107 'extra_args' => $extra_args );
109 $args{'userstore'} = $dir;
111 ### Custom hacking for Stone
112 if( $informat eq 'CollateText' ) {
113 $args{'sigla'} = [ qw/ S M X V Z Bb B K W L / ];
115 my $scope = $dir->new_scope() if $dir;
116 $tradition = Text::Tradition->new( %args );
119 my $stemma = $tradition->add_stemma( dotfile => $stemmafile );
120 print STDERR "Saved stemma at $stemmafile\n" if $stemma;
123 # Now output what we have been asked to.
124 if( $outformat eq 'stemma' ) {
125 my $cdata = character_input( $tradition );
127 print phylip_pars( $cdata );
128 } catch( Text::Tradition::Error $e ) {
129 print STDERR "Bad result: " . $e->message;
131 } elsif( $outformat eq 'db' ) {
133 my $extra_args = { 'create' => 1 };
134 $extra_args->{'user'} = $dbuser if $dbuser;
135 $extra_args->{'password'} = $dbpass if $dbpass;
136 $dir = Text::Tradition::Directory->new( 'dsn' => $dsn,
137 'extra_args' => $extra_args );
139 my $scope = $dir->new_scope;
142 $uuid = $dir->store( $dbid => $tradition );
144 $uuid = $dir->store( $tradition );
146 print STDERR "Saved tradition to database with ID $uuid\n";
148 my $output = "as_$outformat";
150 $opts->{'from'} = $from if $from;
151 $opts->{'to'} = $to if $to;
152 $opts->{'nocalc'} = 1 if $nocalc;
153 print $tradition->collation->$output( $opts );
158 print STDERR << "EOF"
159 Usage: $0 -i [format] -o [format] (--base [filename]) (--(no)linear) [inputfile]
160 i, input: Format of the input file. Must be one of CollateX, CSV, CTE, Self, TEI.
161 o, output: Format of the output. Must be one of svg, dot, graphml, csv, stemma.
162 b, base: Filename that contains a base text. Needed for CSV input.
163 l, linear: Treat transposed readings separately, producing a linear graph.
164 If nolinear, treat transposed readings as the same node.
165 h, help: Print this message.
169 print STDERR "$msg\n";
171 exit ($msg ? 1 : 0 );