From: Dobrica Pavlinusic <dpavlin@rot13.org>
Date: Fri, 13 Sep 2013 10:05:12 +0000 (+0200)
Subject: re-insert commas in data
X-Git-Url: http://git.rot13.org/?p=crolist2marc;a=commitdiff_plain;h=9da83a6114ad6f08db6e1644324228a51b7ecd17;ds=sidebyside

re-insert commas in data

This requires a little bit of background knowladge about database
schema (TEXTKEY is 8 chars, TEXTREST is rest of content)
---

diff --git a/csv2tsv.pl b/csv2tsv.pl
new file mode 100755
index 0000000..fdbd848
--- /dev/null
+++ b/csv2tsv.pl
@@ -0,0 +1,44 @@
+#!/usr/bin/perl
+use warnings;
+use strict;
+
+# ls *.csv | xargs -i sh -cx "./csv2tsv.pl {} > tsv/{} 2>>/tmp/log"
+
+use Data::Dump qw(dump);
+
+my $cols = 0;
+my $delimiter;
+
+while(my $line = <>) {
+	chomp $line;
+	if ( $cols == 0 ) { # header
+		$cols = $line =~ s/([,;])/$1/g;
+		$delimiter = $1;
+		warn "# delimiter [$delimiter] columns=$cols\n";
+	}
+
+	$line =~ s/,$/,_null_/; # protect last empty column
+
+	my @v = split(/$delimiter/, $line);
+	if ( $#v > $cols ) {
+#warn "FIX $#v $cols", dump(@v);
+		my @l2 = splice( @v, -2 ); # keep last two
+		my @fix = splice( @v, 6 );
+warn "XXX before ",dump( @fix );
+		my ( $k, $v ) = ( shift(@fix), '' );
+		while ( @fix && (( length($k) + length($fix[0]) ) < 8 ) ) {
+			$k .= $k ? ',' : '';
+			$k .= shift(@fix);
+		}
+		$k .= " " x (8 - length($k)); # pad key to 8 spaces
+		$v = join(',', @fix);
+warn "YYY after  ",dump( $k, $v );
+		die "ASSERT key < 8 [$k]" if length($k) < 8;
+		die "ASSERT key > 8 [$k]" if length($k) > 8;
+		@v = ( @v, $k, $v, @l2 );
+	}
+
+	$v[-1] =~ s/_null_//;
+
+	$ENV{DEBUG} || print join("\t", @v), "\n";
+}