add facilities for removing 'collated' relationship, and for annotation
[scpubgit/stemmatology.git] / script / strip_punctuation.pl
CommitLineData
6f0ec5df 1#!/usr/bin/env perl
2
3use lib 'lib';
4use strict;
5use warnings;
6use Text::Tradition::Directory;
7
8my( $dsn, $user, $pass ) = @ARGV;
9
10my $connect_args = { dsn => $dsn };
11$connect_args->{'extra_args'} = { user => $user, password => $pass }
12 if $user && $pass;
13my $dir = Text::Tradition::Directory->new( $connect_args );
14
15foreach my $id ( $dir->tradition_ids ) {
16 my $scope = $dir->new_scope;
17 my $tradition = $dir->lookup( $id );
18 print STDERR "Processing tradition " . $tradition->name . "\n";
19 foreach my $reading ( $tradition->collation->readings ) {
20 $reading->alter_text( strip_punct( $reading->text ) );
21 }
22 $tradition->collation->flatten_ranks;
23 $dir->save( $tradition );
24}
25
26print STDERR "Done\n";
27
28sub strip_punct {
29 my( $rtext ) = @_;
30 my $orig_r = $rtext;
31 return $rtext unless $rtext =~ /\w/;
32 $rtext =~ s/^\W+//;
33 $rtext =~ s/\W+$//;
34 print STDERR "Altering $orig_r to $rtext\n"
35 unless $orig_r eq $rtext;
36 return $rtext;
37}
38