From f0b6e8f49d756121a755f661d7e162feb5699ba8 Mon Sep 17 00:00:00 2001
From: Gerhard Gonter <ggonter@gmail.com>
Date: Mon, 27 Mar 2023 19:29:10 +0200
Subject: [PATCH] improved commandline argument handling to provide options for
"light" and "heavy" versions of the output
---
filter_eod_report.pl | 53 +++++++++++++++++++++++++++++++++++++++-----
1 file changed, 47 insertions(+), 6 deletions(-)
diff --git a/filter_eod_report.pl b/filter_eod_report.pl
index a083170..3b95735 100755
--- a/filter_eod_report.pl
+++ b/filter_eod_report.pl
@@ -7,11 +7,46 @@ $Data::Dumper::Indent= 1;
use Util::Simple_CSV;
-my $fnm_in= shift(@ARGV) || 'eod_data.tsv';
-my $fnm_out= shift(@ARGV) || 'report_bib.tsv';
+my $detail_level= 1;
+# 0 .. only the most necessary things (corrections and new phaira urls)
+# 1 .. plus everything that is missing, e.g. added NBNs (urn) or Handles (hdl)
-my @columns_copy= qw(ac_number mms_id ticket val_hdl);
-my @columns_out= qw(ac_number mms_id alma_fetched ticket url nbn hdl val_hdl);
+my @pars= ();
+while (defined (my $arg= shift (@ARGV)))
+{
+ if ($arg eq '-') { push (@pars, '-'); }
+ elsif ($arg eq '--') { push (@pars, @ARGV); @ARGV= (); }
+ elsif ($arg =~ /^--(.+)/)
+ {
+ my ($opt, $val)= split ('=', $1, 2);
+ if ($opt eq 'help') { usage(); }
+ elsif ($opt eq 'heavy') { $detail_level= 1; }
+ elsif ($opt eq 'light') { $detail_level= 0; }
+ else { usage(); }
+ }
+ elsif ($arg =~ /^-(.+)/)
+ {
+ foreach my $opt (split ('', $1))
+ {
+ if ($opt eq 'h') { usage(); exit (0); }
+ elsif ($opt eq 'H') { $detail_level= 1; }
+ elsif ($opt eq 'L') { $detail_level= 0; }
+ else { usage(); }
+ }
+ }
+ else
+ {
+ push (@pars, $arg);
+ }
+}
+
+print join (' ', __FILE__, __LINE__, 'caller=['. caller() . ']'), "\n";
+
+my $fnm_in= shift(@pars) || 'eod_data.tsv';
+my $fnm_out= shift(@pars) || 'report_bib_'. (($detail_level) ? 'heavy' : 'light') .'.tsv';
+
+my @columns_copy= qw(ac_number mms_id alma_notes ticket val_hdl);
+my @columns_out= qw(ac_number mms_id alma_fetched alma_notes ticket url nbn hdl);
my $tsv_in= Util::Simple_CSV->new(load => $fnm_in, separator => "\t");
@@ -21,13 +56,14 @@ print TSV_OUT join("\t", @columns_out), "\n";
my $data= $tsv_in->{data};
# print __LINE__, " data: ", Dumper ($data);
+my $count= 0;
foreach my $row (@$data)
{
# print __LINE__, " row: ", Dumper ($row);
my ($update_url, $update_urn, $update_hdl, $update_doi)= map { $row->{$_} } qw( update_phaidra_url update_urn update_hdl update_doi );
# printf ("update_url=[%s] update_urn=[%s] update_hdl=[%s] update_doi=[%s]\n", $update_url, $update_urn, $update_hdl, $update_doi);
- if ($update_url || (0 && $update_urn) || $update_hdl)
+ if ($update_url || ($detail_level && ($update_urn || $update_hdl)))
{
my %out= map { $_ => $row->{$_} } @columns_copy;
$out{url}= $update_url if (defined ($update_url));
@@ -35,11 +71,16 @@ foreach my $row (@$data)
$out{hdl}= $update_hdl if (defined ($update_hdl));
$out{alma_fetched}= $row->{ts_fetched};
- $out{hdl}= $row->{hdl} unless ($row->{val_hdl} eq $row->{hdl}); # set handle only when not alreadys set...
+ $out{hdl}= $row->{hdl} if (($detail_level == 0 && $row->{val_hdl} ne $row->{hdl}) || $detail_level > 0); # set handle only when not alreadys set...
# print __LINE__, " out: ", Dumper(\%out);
print TSV_OUT join("\t", map { $out{$_} } @columns_out), "\n";
+ $count++;
}
}
+close(TSV);
+
+print __LINE__, " saved $count records to $fnm_out\n";
+
--
GitLab