Commit | Line | Data |
910a0a6d |
1 | #!/usr/bin/env perl |
2 | |
3 | use lib 'lib'; |
4 | use strict; |
5 | use warnings; |
6 | use Getopt::Long; |
56c56f0d |
7 | use TryCatch; |
910a0a6d |
8 | use Text::Tradition; |
861c3e27 |
9 | use Text::Tradition::Directory; |
56c56f0d |
10 | use Text::Tradition::StemmaUtil qw/ character_input phylip_pars /; |
910a0a6d |
11 | |
12 | binmode STDERR, ":utf8"; |
13 | binmode STDOUT, ":utf8"; |
14 | eval { no warnings; binmode $DB::OUT, ":utf8"; }; |
15 | |
10943ab0 |
16 | # Variables with defaults |
17 | my( $informat, $outformat, $language, $name, $sep, $dsn ) = ( '', '', 'Default', |
18 | 'Tradition', "\t", "dbi:SQLite:dbname=stemmaweb/db/traditions.db" ); |
19 | # Variables with no default |
20 | my( $inbase, $help, $stemmafile, $dbuser, $dbpass, $from, $to, $dbid, $debug ); |
910a0a6d |
21 | |
408449b7 |
22 | GetOptions( 'i|in=s' => \$informat, |
23 | 'b|base=s' => \$inbase, |
24 | 'o|out=s' => \$outformat, |
6a1c434d |
25 | 'l|language=s' => \$language, |
861c3e27 |
26 | 'n|name=s' => \$name, |
408449b7 |
27 | 'h|help' => \$help, |
861c3e27 |
28 | 's|stemma=s' => \$stemmafile, |
7d99d254 |
29 | 'u|user=s' => \$dbuser, |
30 | 'p|pass=s' => \$dbpass, |
fd7014c4 |
31 | 'f|from=s' => \$from, |
32 | 't|to=s' => \$to, |
a7fb3133 |
33 | 'sep=s' => \$sep, |
861c3e27 |
34 | 'dsn=s' => \$dsn, |
28333e88 |
35 | 'dbid=s' => \$dbid, |
10943ab0 |
36 | 'debug' => \$debug |
910a0a6d |
37 | ); |
38 | |
39 | if( $help ) { |
40 | help(); |
41 | } |
42 | |
fa954f4c |
43 | unless( $informat =~ /^(CSV|CTE|KUL|Self|TEI|CollateX|tab(ular)?)|stone$/i ) { |
910a0a6d |
44 | help( "Input format must be one of CollateX, CSV, CTE, Self, TEI" ); |
45 | } |
46 | $informat = 'CollateX' if $informat =~ /^c(ollate)?x$/i; |
47 | $informat = 'KUL' if $informat =~ /^kul$/i; |
48 | $informat = 'CTE' if $informat =~ /^cte$/i; |
49 | $informat = 'Self' if $informat =~ /^self$/i; |
50 | $informat = 'TEI' if $informat =~ /^tei$/i; |
d9e873d0 |
51 | $informat = 'Tabular' if $informat =~ /^tab$/i; |
fa954f4c |
52 | $informat = 'CollateText' if $informat =~ /^stone$/i; |
910a0a6d |
53 | |
861c3e27 |
54 | unless( $outformat =~ /^(graphml|svg|dot|stemma|csv|db)$/ ) { |
55 | help( "Output format must be one of db, graphml, svg, csv, stemma, or dot" ); |
910a0a6d |
56 | } |
57 | |
fd7014c4 |
58 | if( $from || $to ) { |
59 | help( "Subgraphs only supported in GraphML format" ) |
60 | unless $outformat eq 'graphml'; |
61 | } |
62 | |
910a0a6d |
63 | # Do we have a base if we need it? |
fa954f4c |
64 | if( $informat =~ /^(KUL|CollateText)$/ && !$inbase ) { |
910a0a6d |
65 | help( "$informat input needs a base text" ); |
66 | } |
aa71409f |
67 | $sep = "\t" if $sep eq 'tab'; |
fa954f4c |
68 | |
910a0a6d |
69 | my $input = $ARGV[0]; |
910a0a6d |
70 | |
71 | # First: read the base. Make a graph, but also note which |
72 | # nodes represent line beginnings. |
dfc37e38 |
73 | my %args = ( 'input' => $informat, |
6a1c434d |
74 | 'file' => $input ); |
910a0a6d |
75 | $args{'base'} = $inbase if $inbase; |
6a1c434d |
76 | $args{'language'} = $language if $language; |
408449b7 |
77 | $args{'name'} = $name if $name; |
a7fb3133 |
78 | $args{'sep_char'} = $sep if $informat eq 'Tabular'; |
fa954f4c |
79 | ### Custom hacking for Stone |
80 | if( $informat eq 'CollateText' ) { |
81 | $args{'sigla'} = [ qw/ S M X V Z Bb B K W L / ]; |
82 | } |
910a0a6d |
83 | my $tradition = Text::Tradition->new( %args ); |
861c3e27 |
84 | if( $stemmafile ) { |
173ecc07 |
85 | my $stemma = $tradition->add_stemma( dotfile => $stemmafile ); |
861c3e27 |
86 | print STDERR "Saved stemma at $stemmafile\n" if $stemma; |
87 | } |
910a0a6d |
88 | |
910a0a6d |
89 | # Now output what we have been asked to. |
90 | if( $outformat eq 'stemma' ) { |
56c56f0d |
91 | my $cdata = character_input( $tradition->collation->alignment_table ); |
92 | try { |
93 | print phylip_pars( $cdata ); |
94 | } catch( Text::Tradition::Error $e ) { |
95 | print STDERR "Bad result: " . $e->message; |
910a0a6d |
96 | } |
861c3e27 |
97 | } elsif( $outformat eq 'db' ) { |
7d99d254 |
98 | my $extra_args = { 'create' => 1 }; |
99 | $extra_args->{'user'} = $dbuser if $dbuser; |
100 | $extra_args->{'password'} = $dbpass if $dbpass; |
861c3e27 |
101 | my $dir = Text::Tradition::Directory->new( 'dsn' => $dsn, |
7d99d254 |
102 | 'extra_args' => $extra_args ); |
861c3e27 |
103 | my $scope = $dir->new_scope; |
28333e88 |
104 | my $uuid; |
105 | if( $dbid ) { |
106 | $uuid = $dir->store( $dbid => $tradition ); |
107 | } else { |
108 | $uuid = $dir->store( $tradition ); |
109 | } |
861c3e27 |
110 | print STDERR "Saved tradition to database with ID $uuid\n"; |
910a0a6d |
111 | } else { |
112 | my $output = "as_$outformat"; |
fd7014c4 |
113 | my $opts = {}; |
114 | $opts->{'from'} = $from if $from; |
115 | $opts->{'to'} = $to if $to; |
10943ab0 |
116 | $opts->{'nocalc'} = 1 if $debug; |
fd7014c4 |
117 | print $tradition->collation->$output( $opts ); |
910a0a6d |
118 | } |
119 | |
120 | sub help { |
121 | my( $msg ) = @_; |
122 | print STDERR << "EOF" |
123 | Usage: $0 -i [format] -o [format] (--base [filename]) (--(no)linear) [inputfile] |
124 | i, input: Format of the input file. Must be one of CollateX, CSV, CTE, Self, TEI. |
125 | o, output: Format of the output. Must be one of svg, dot, graphml, csv, stemma. |
126 | b, base: Filename that contains a base text. Needed for CSV input. |
127 | l, linear: Treat transposed readings separately, producing a linear graph. |
128 | If nolinear, treat transposed readings as the same node. |
129 | h, help: Print this message. |
130 | EOF |
131 | ; |
132 | if( $msg ) { |
133 | print STDERR "$msg\n"; |
134 | } |
135 | exit ($msg ? 1 : 0 ); |
136 | } |