Commit | Line | Data |
910a0a6d |
1 | #!/usr/bin/env perl |
2 | |
3 | use lib 'lib'; |
4 | use strict; |
5 | use warnings; |
6 | use Getopt::Long; |
7 | use Text::Tradition; |
8 | use Text::Tradition::Stemma; |
9 | |
10 | binmode STDERR, ":utf8"; |
11 | binmode STDOUT, ":utf8"; |
12 | eval { no warnings; binmode $DB::OUT, ":utf8"; }; |
13 | |
a7fb3133 |
14 | my( $informat, $inbase, $outformat, $help, $linear, $name, $HACK, $sep ) |
15 | = ( '', '', '', '', 1, 'Tradition', 0, ',' ); |
910a0a6d |
16 | |
408449b7 |
17 | GetOptions( 'i|in=s' => \$informat, |
18 | 'b|base=s' => \$inbase, |
19 | 'o|out=s' => \$outformat, |
910a0a6d |
20 | 'l|linear!' => \$linear, |
aa71409f |
21 | 'n|name=s' => \$name, |
408449b7 |
22 | 'h|help' => \$help, |
a7fb3133 |
23 | 'sep=s' => \$sep, |
408449b7 |
24 | 'hack' => \$HACK, |
910a0a6d |
25 | ); |
26 | |
27 | if( $help ) { |
28 | help(); |
29 | } |
30 | |
fa954f4c |
31 | unless( $informat =~ /^(CSV|CTE|KUL|Self|TEI|CollateX|tab(ular)?)|stone$/i ) { |
910a0a6d |
32 | help( "Input format must be one of CollateX, CSV, CTE, Self, TEI" ); |
33 | } |
34 | $informat = 'CollateX' if $informat =~ /^c(ollate)?x$/i; |
35 | $informat = 'KUL' if $informat =~ /^kul$/i; |
36 | $informat = 'CTE' if $informat =~ /^cte$/i; |
37 | $informat = 'Self' if $informat =~ /^self$/i; |
38 | $informat = 'TEI' if $informat =~ /^tei$/i; |
d9e873d0 |
39 | $informat = 'Tabular' if $informat =~ /^tab$/i; |
fa954f4c |
40 | $informat = 'CollateText' if $informat =~ /^stone$/i; |
910a0a6d |
41 | |
42 | unless( $outformat =~ /^(graphml|svg|dot|stemma|csv)$/ ) { |
43 | help( "Output format must be one of graphml, svg, csv, stemma, or dot" ); |
44 | } |
45 | |
46 | # Do we have a base if we need it? |
fa954f4c |
47 | if( $informat =~ /^(KUL|CollateText)$/ && !$inbase ) { |
910a0a6d |
48 | help( "$informat input needs a base text" ); |
49 | } |
aa71409f |
50 | $sep = "\t" if $sep eq 'tab'; |
fa954f4c |
51 | |
910a0a6d |
52 | my $input = $ARGV[0]; |
910a0a6d |
53 | |
54 | # First: read the base. Make a graph, but also note which |
55 | # nodes represent line beginnings. |
dfc37e38 |
56 | my %args = ( 'input' => $informat, |
57 | 'file' => $input, |
910a0a6d |
58 | 'linear' => $linear ); |
59 | $args{'base'} = $inbase if $inbase; |
408449b7 |
60 | $args{'name'} = $name if $name; |
a7fb3133 |
61 | $args{'sep_char'} = $sep if $informat eq 'Tabular'; |
fa954f4c |
62 | ### Custom hacking for Stone |
63 | if( $informat eq 'CollateText' ) { |
64 | $args{'sigla'} = [ qw/ S M X V Z Bb B K W L / ]; |
65 | } |
910a0a6d |
66 | my $tradition = Text::Tradition->new( %args ); |
67 | |
68 | ### Custom hacking |
69 | # Remove witnesses C, E, G in the Matthew text |
70 | if( $HACK ) { |
71 | foreach( $tradition->collation->paths() ) { |
72 | $tradition->collation->del_path( $_ ) if $_->label =~ /^[ceg]$/i; |
73 | } |
74 | foreach( $tradition->collation->readings() ) { |
75 | if( !$_->outgoing() && !$_->incoming() ) { |
76 | print STDERR "Deleting reading " . $_->label . "\n"; |
77 | $tradition->collation->del_reading( $_ ); |
78 | } |
79 | } |
80 | } |
81 | |
82 | # Now output what we have been asked to. |
83 | if( $outformat eq 'stemma' ) { |
84 | my $stemma = Text::Tradition::Stemma->new( |
85 | 'collation' => $tradition->collation ); |
40f19742 |
86 | my( $result, $tree ) = $stemma->run_phylip_pars(); |
910a0a6d |
87 | if( $result ) { |
88 | print $tree; |
89 | } else { |
90 | print STDERR "Bad result: $tree"; |
91 | } |
92 | } else { |
93 | my $output = "as_$outformat"; |
94 | print $tradition->collation->$output(); |
95 | } |
96 | |
97 | sub help { |
98 | my( $msg ) = @_; |
99 | print STDERR << "EOF" |
100 | Usage: $0 -i [format] -o [format] (--base [filename]) (--(no)linear) [inputfile] |
101 | i, input: Format of the input file. Must be one of CollateX, CSV, CTE, Self, TEI. |
102 | o, output: Format of the output. Must be one of svg, dot, graphml, csv, stemma. |
103 | b, base: Filename that contains a base text. Needed for CSV input. |
104 | l, linear: Treat transposed readings separately, producing a linear graph. |
105 | If nolinear, treat transposed readings as the same node. |
106 | h, help: Print this message. |
107 | EOF |
108 | ; |
109 | if( $msg ) { |
110 | print STDERR "$msg\n"; |
111 | } |
112 | exit ($msg ? 1 : 0 ); |
113 | } |