korpuse teisendused

korpus_sh/COMMANDS

+
+python3 wrapSH.py -s ./sh/cg.sh -r ./korpus.cg

korpus_sh/sh/cg.sh

 #!/bin/sh
-ELEMENT_ID=$1
+ID=$1
 SENTENCE=$2
 
-echo "$SENTENCE"
\ No newline at end of file
+RADA=/home/osboxes/bin/EstCG-master
+RADAMRF=~/bin/vabamorf-master/apps/cmdline/project/unix
+RADADCT=~/bin/vabamorf-master/dct/binary/
+
+echo $SENTENCE \
+| $RADA/rlausestaja.pl \
+| $RADA/wr2json.pl \
+| $RADAMRF/etana analyze -lex $RADADCT/et.dct -guess \
+| sed -e "s/[\\]n/\n/g" | $RADA/json2mrf.pl \
+| $RADA/rtolkija.pl \
+| $RADA/pron17.pl \
+| $RADA/tcopyrem_kaili.pl \
+| $RADA/tkms2cg3.pl \
+| vislcg3 -g $RADA/preprocess.rul\
+| vislcg3 -o -g $RADA/clo.rul \
+| vislcg3 -o -g $RADA/morfyhe17.rul \
+| vislcg3 -o -g $RADA/PhVerbs17.rul \
+| vislcg3 -o -g $RADA/pindsyn17.rul \
+| vislcg3 -o -g $RADA/strukt.rul \
+| perl ./util/cgAddId.pl $ID

korpus_sh/sh/sample.sh

+#!/bin/sh
+ELEMENT_ID=$1
+SENTENCE=$2
+
+echo "$SENTENCE"
\ No newline at end of file

korpus_sh/util/cgAddId.pl

+#!/bin/perl 
+
+# argument [0] on ID
+#std.in sisendist loeme teksti
+#asendame kõik <s> -> <s id="ID_0"> <s id="ID_1"> jne
+#<s id="ID_0"> asendame <s id="ID">
+
+use utf8;
+use strict;
+
+binmode(STDOUT, ":utf8");
+binmode(STDIN, ":utf8");
+
+our $ID = $ARGV[0];
+our $text = '';
+
+while (<STDIN>) {
+
+	$text .= $_;
+	
+}
+
+our $altered_text = '';
+
+our $i = 0;
+while ($altered_text ne $text) {
+
+	$altered_text = $text;
+	my $subst = '<s id="'.$ID.'_'.$i.'">';
+	if ($i==0)
+	{
+		$subst = '<s id="'.$ID.'">';
+	} 	
+	$text =~ s/<s>/$subst/;
+	$i++;
+}
+
+print STDOUT $text;

korpus_sh/wrapSH.py

@@ -85,7 +85,7 @@ for row in tsv_file.read().split('\n'):
 	arr = row.split('\t')
 	if len(arr)>1:
 		if len(arr[0]) and len(arr[1]):
-			print ([options['sh_script'], '"%s"'%arr[0] , '"%s"'%arr[1]] )
-			subprocess.run([options['sh_script'], '"%s"'%arr[0] , '"%s"'%arr[1]], stdout=outfile)
+			print ([options['sh_script'], '%s'%arr[0] , '%s'%arr[1]] )
+			subprocess.run([options['sh_script'], '%s'%arr[0] , '%s'%arr[1]], stdout=outfile)