9 use Text::Tradition::Directory;
10 use Text::Tradition::StemmaUtil qw/ character_input phylip_pars newick_to_svg /;
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( $informat =~ /^(CTE|KUL|Self|TEI|CollateX|tab(ular)?)|xlsx?|db$/i ) {
46 help( "Input format must be one of CollateX, CTE, Self, TEI, Tabular, XLS(X), or DB" );
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( $outformat =~ /^(graphml|svg|dot|adj(acency)?|stemma(svg)?|(c|t)sv|db)$/i ) {
59 help( "Output format must be one of db, graphml, svg, csv, tsv, stemma, adjacency, or dot" );
61 $outformat = 'adjacency_list' if $outformat =~ /^adj/i;
64 help( "Subgraphs only supported in GraphML, dot, adjacency, or SVG format" )
65 unless $outformat =~ /^(graphml|dot|svg|adjacency_list)$/;
68 # Do we have a base if we need it?
69 if( $informat =~ /^(KUL|CollateText)$/ && !$inbase ) {
70 help( "$informat input needs a base text" );
72 $sep = "\t" if $sep eq 'tab';
77 if( $informat eq 'db' ) {
78 my $dbargs = { dsn => $dsn };
79 $dbargs->{'extra_args'}->{'user'} = $dbuser if $dbuser;
80 $dbargs->{'extra_args'}->{'password'} = $dbpass if $dbpass;
81 $dir = Text::Tradition::Directory->new( $dbargs );
82 my $scope = $dir->new_scope();
83 $tradition = $dir->lookup( $input );
85 # First: read the base. Make a graph, but also note which
86 # nodes represent line beginnings.
87 my %args = ( 'input' => $informat,
89 $args{'linear'} = 0 if $nonlinear;
90 $args{'base'} = $inbase if $inbase;
91 $args{'language'} = $language if $language;
92 $args{'name'} = $name if $name;
93 $args{'nocalc'} = 1 if $nocalc;
94 if( $informat eq 'Tabular' ) {
96 $args{'excel'} = $excel;
98 $args{'sep_char'} = $sep;
101 # If we are writing to the database, use that DB as the userstore.
102 if( $outformat eq 'db' ) {
104 my $extra_args = { 'create' => 1 };
105 $extra_args->{'user'} = $dbuser if $dbuser;
106 $extra_args->{'password'} = $dbpass if $dbpass;
107 $dir = Text::Tradition::Directory->new( 'dsn' => $dsn,
108 'extra_args' => $extra_args );
110 $args{'userstore'} = $dir;
112 ### Custom hacking for Stone
113 if( $informat eq 'CollateText' ) {
114 $args{'sigla'} = [ qw/ S M X V Z Bb B K W L / ];
116 my $scope = $dir->new_scope() if $dir;
117 $tradition = Text::Tradition->new( %args );
120 my $stemma = $tradition->add_stemma( dotfile => $stemmafile );
121 print STDERR "Saved stemma at $stemmafile\n" if $stemma;
124 # Now output what we have been asked to.
125 if( $outformat =~ /^stemma(.*)$/ ) {
126 my $type = $1 || 'newick';
127 my $cdata = character_input( $tradition );
130 $newick = phylip_pars( $cdata );
131 } catch( Text::Tradition::Error $e ) {
132 print STDERR "Bad result from pars: " . $e->message;
135 if( $type eq 'newick' ) {
137 } elsif( $type eq 'svg' ) {
138 print newick_to_svg( $newick );
140 } elsif( $outformat eq 'db' ) {
142 my $extra_args = { 'create' => 1 };
143 $extra_args->{'user'} = $dbuser if $dbuser;
144 $extra_args->{'password'} = $dbpass if $dbpass;
145 $dir = Text::Tradition::Directory->new( 'dsn' => $dsn,
146 'extra_args' => $extra_args );
148 my $scope = $dir->new_scope;
151 $uuid = $dir->store( $dbid => $tradition );
153 $uuid = $dir->store( $tradition );
155 print STDERR "Saved tradition to database with ID $uuid\n";
157 my $output = "as_$outformat";
159 $opts->{'from'} = $from if $from;
160 $opts->{'to'} = $to if $to;
161 $opts->{'nocalc'} = 1 if $nocalc;
162 print $tradition->collation->$output( $opts );
167 print STDERR << "EOF"
168 Usage: $0 -i [format] -o [format] (--base [filename]) (--(no)linear) [inputfile]
169 i, input: Format of the input file. Must be one of CollateX, CSV, CTE, Self, TEI.
170 o, output: Format of the output. Must be one of svg, dot, graphml, csv, stemma.
171 b, base: Filename that contains a base text. Needed for CSV input.
172 l, linear: Treat transposed readings separately, producing a linear graph.
173 If nolinear, treat transposed readings as the same node.
174 h, help: Print this message.
178 print STDERR "$msg\n";
180 exit ($msg ? 1 : 0 );