stop tracking bbedit stuff; first pass at Collate! parsing

[scpubgit/stemmatology.git] / make_tradition.pl

diff --git a/make_tradition.pl b/make_tradition.pl
old mode 100644 (file)
new mode 100755 (executable)
index 22b50df..f928c33
--- a/make_tradition.pl
+++ b/make_tradition.pl
@@ -11,22 +11,23 @@ binmode STDERR, ":utf8";
 binmode STDOUT, ":utf8";
 eval { no warnings; binmode $DB::OUT, ":utf8"; };
 
-my( $informat, $inbase, $outformat, $help, $linear, $HACK ) 
-    = ( '', '', '', '', 1, 0 );
+my( $informat, $inbase, $outformat, $help, $linear, $name, $HACK ) 
+    = ( '', '', '', '', 1, 'Tradition', 0 );
 
-GetOptions( 'i|in=s'   => \$informat,
-            'b|base=s' => \$inbase,
-            'o|out=s'  => \$outformat,
+GetOptions( 'i|in=s'    => \$informat,
+            'b|base=s'  => \$inbase,
+            'o|out=s'   => \$outformat,
             'l|linear!' => \$linear,
-            'h|help' => \$help,
-            'hack' => \$HACK,
+            'n|name'    => \$name,
+            'h|help'    => \$help,
+            'hack'      => \$HACK,
     );
 
 if( $help ) {
     help();
 }
 
-unless( $informat =~ /^(CSV|CTE|KUL|Self|TEI|CollateX|tab(ular)?)$/i ) {
+unless( $informat =~ /^(CSV|CTE|KUL|Self|TEI|CollateX|tab(ular)?)|stone$/i ) {
     help( "Input format must be one of CollateX, CSV, CTE, Self, TEI" );
 }
 $informat = 'CollateX' if $informat =~ /^c(ollate)?x$/i;
@@ -35,32 +36,31 @@ $informat = 'CTE' if $informat =~ /^cte$/i;
 $informat = 'Self' if $informat =~ /^self$/i;
 $informat = 'TEI' if $informat =~ /^tei$/i;
 $informat = 'Tabular' if $informat =~ /^tab$/i;
+$informat = 'CollateText' if $informat =~ /^stone$/i;
 
 unless( $outformat =~ /^(graphml|svg|dot|stemma|csv)$/ ) {
     help( "Output format must be one of graphml, svg, csv, stemma, or dot" );
 }
 
 # Do we have a base if we need it?
-if( $informat eq 'KUL' && !$inbase ) {
+if( $informat =~ /^(KUL|CollateText)$/ && !$inbase ) {
     help( "$informat input needs a base text" );
 }
 
-# CSV parsing requires a filename; XML parsing requires a string.
+
 my $input = $ARGV[0];
-unless( $informat eq 'KUL' || $informat eq 'CSV' ) {
-    my @lines;
-    open( INFILE, "$input" ) or die "Could not read $input";
-    binmode INFILE, ':utf8';
-    @lines = <INFILE>;
-    close INFILE;
-    $input = join( '', @lines );
-}
 
 # First: read the base. Make a graph, but also note which
 # nodes represent line beginnings.
-my %args = ( $informat => $input,
+my %args = ( 'input' => $informat,
+             'file' => $input,
              'linear' => $linear );
 $args{'base'} = $inbase if $inbase;
+$args{'name'} = $name if $name;
+### Custom hacking for Stone
+if( $informat eq 'CollateText' ) {
+    $args{'sigla'} = [ qw/ S M X V Z Bb B K W L / ];
+}
 my $tradition = Text::Tradition->new( %args );
 
 ### Custom hacking