work on tagged OGNT
This commit is contained in:
parent
7c454c4ce3
commit
2f76e751b4
|
@ -7,7 +7,7 @@ use utf8;
|
|||
#use open IN => ":utf8", OUT => ":utf8";
|
||||
use open IO => ":utf8";
|
||||
binmode STDOUT, ":encoding(UTF-8)";
|
||||
mkdir "OGNT";
|
||||
mkdir "OGNT_for_tagging";
|
||||
mkdir "Logs";
|
||||
|
||||
open LOG, ">Logs/log.log";
|
||||
|
@ -34,7 +34,7 @@ while (<IN>) {
|
|||
Separate();
|
||||
}
|
||||
|
||||
say OUT " </verse>\n </chapter>\n </book>\n</xml>";
|
||||
say OUT " </verse>\n </chapter>\n </div>\n </book>\n</xml>";
|
||||
|
||||
say "Closing input and output files ...";
|
||||
|
||||
|
@ -54,23 +54,23 @@ sub Separate {
|
|||
my ($this_bk) = ($bk{$bn});
|
||||
$bklc = lc $bk{$bn};
|
||||
if (OUT-> opened()) {
|
||||
say OUT " </verse>\n </chapter>\n </book>\n</xml>";
|
||||
say OUT " </verse>\n </chapter>\n </div>\n </book>\n</xml>";
|
||||
close OUT;
|
||||
}
|
||||
open OUT, ">:utf8", "OGNT/$bn-$bk{$bn}.xml" or die "$! $bn-$bk{$bn}.xml";
|
||||
say OUT "\n<xml>\n <div type=\"book\" osisID=\"$bklc\">\n <chapter osisID=\"$bklc.$ch\">\n <verse osisID=\"$bklc.$ch.$vs\">";
|
||||
open OUT, ">:utf8", "OGNT_for_tagging/$bn-$bk{$bn}.xml" or die "$! $bn-$bk{$bn}.xml";
|
||||
say OUT "\n<xml>\n <book>\n <div type=\"book\" osisID=\"$bklc\">\n <chapter osisID=\"$bklc.$ch\">\n <verse osisID=\"$bklc.$ch.$vs\">";
|
||||
($last_bn, $last_ch, $last_vs) = ($bn, $ch, $vs)
|
||||
}
|
||||
elsif ($ch ne $last_ch) {
|
||||
say OUT " </verse>\n </chapter>\n <chapter osisID=\"$bklc.$ch\">\n <verse osisID=\"$bklc.$ch.$vs\">";
|
||||
say OUT " </verse>\n </chapter>\n <chapter osisID=\"$bklc.$ch\">\n <verse osisID=\"$bklc.$ch.$vs\">";
|
||||
($last_ch, $last_vs) = ($ch, $vs)
|
||||
}
|
||||
elsif ($vs ne $last_vs) {
|
||||
my ($this_bk, $bklc) = ($bk{$bn}, lc $bk{$bn});
|
||||
say OUT " </verse>\n <verse osisID=\"$bklc.$ch.$vs\">";
|
||||
say OUT " </verse>\n <verse osisID=\"$bklc.$ch.$vs\">";
|
||||
$last_vs = $vs;
|
||||
}
|
||||
say OUT "\t\t\t\t<w OGNTsort=\"$OGNTSort\" ULBorder=\"\<##\>\"lemma=\"$sn\" morph=\"$gram\" lexeme=\"$lexeme\">$word</w>"
|
||||
say OUT "\t\t\t\t\t<w OGNTsort=\"$OGNTSort\" ULBorder=\"\<##\>\"lemma=\"G$sn\" morph=\"$gram\" lexeme=\"$lexeme\">$word</w>"
|
||||
}
|
||||
}
|
||||
|
||||
|
|
Loading…
Reference in New Issue