From e34beb3b6829d82fb02ea9f1b4ba803e090d81fc Mon Sep 17 00:00:00 2001 From: tonnesen Date: Fri, 5 Oct 2001 21:27:46 +0000 Subject: [PATCH] Added rudimentary marc importing tool. Just takes a marc record and converts to koha database. Does not keep data in marc format. --- acqui.simple/marcimport.pl | 847 +++++++++++++++++++++++++++++++++++++ 1 file changed, 847 insertions(+) create mode 100755 acqui.simple/marcimport.pl diff --git a/acqui.simple/marcimport.pl b/acqui.simple/marcimport.pl new file mode 100755 index 0000000000..bd98f7ce37 --- /dev/null +++ b/acqui.simple/marcimport.pl @@ -0,0 +1,847 @@ +#!/usr/bin/perl + + +my $lc1='#dddddd'; +my $lc2='#ddaaaa'; + + +use C4::Database; +use CGI; +use DBI; +#use strict; +use C4::Acquisitions; +use C4::Output; +my $dbh=C4Connect; +my $userid=$ENV{'REMOTE_USER'}; +%tagtext = ( + '001' => 'Control number', + '003' => 'Control number identifier', + '005' => 'Date and time of latest transaction', + '006' => 'Fixed-length data elements -- additional material characteristics', + '007' => 'Physical description fixed field', + '008' => 'Fixed length data elements', + '010' => 'LCCN', + '015' => 'LCCN Cdn', + '020' => 'ISBN', + '022' => 'ISSN', + '037' => 'Source of acquisition', + '040' => 'Cataloging source', + '041' => 'Language code', + '043' => 'Geographic area code', + '050' => 'Library of Congress call number', + '060' => 'National Library of Medicine call number', + '082' => 'Dewey decimal call number', + '100' => 'Main entry -- Personal name', + '110' => 'Main entry -- Corporate name', + '130' => 'Main entry -- Uniform title', + '240' => 'Uniform title', + '245' => 'Title statement', + '246' => 'Varying form of title', + '250' => 'Edition statement', + '256' => 'Computer file characteristics', + '260' => 'Publication, distribution, etc.', + '263' => 'Projected publication date', + '300' => 'Physical description', + '306' => 'Playing time', + '440' => 'Series statement / Added entry -- Title', + '490' => 'Series statement', + '500' => 'General note', + '504' => 'Bibliography, etc. note', + '505' => 'Formatted contents note', + '508' => 'Creation/production credits note', + '510' => 'Citation/references note', + '511' => 'Participant or performer note', + '520' => 'Summary, etc. note', + '521' => 'Target audience note (ie age)', + '530' => 'Additional physical form available note', + '538' => 'System details note', + '586' => 'Awards note', + '600' => 'Subject added entry -- Personal name', + '610' => 'Subject added entry -- Corporate name', + '650' => 'Subject added entry -- Topical term', + '651' => 'Subject added entry -- Geographic name', + '656' => 'Index term -- Occupation', + '700' => 'Added entry -- Personal name', + '710' => 'Added entry -- Corporate name', + '730' => 'Added entry -- Uniform title', + '740' => 'Added entry -- Uncontrolled related/analytical title', + '800' => 'Series added entry -- Personal name', + '830' => 'Series added entry -- Uniform title', + '852' => 'Location', + '856' => 'Electronic location and access', +); + + +my $input = new CGI; +my $dbh=C4Connect; + +print $input->header; +print startpage(); +print startmenu('acquisitions'); +my $file=$input->param('file'); + +if ($input->param('insertnewrecord')) { + my $isbn=$input->param('isbn'); + my $issn=$input->param('issn'); + my $lccn=$input->param('lccn'); + my $q_origisbn=$dbh->quote($input->param('origisbn')); + my $q_origissn=$dbh->quote($input->param('origissn')); + my $q_origlccn=$dbh->quote($input->param('origlccn')); + my $q_isbn=$dbh->quote((($isbn) || ('NIL'))); + my $q_issn=$dbh->quote((($issn) || ('NIL'))); + my $q_lccn=$dbh->quote((($lccn) || ('NIL'))); + $sth=$dbh->prepare("insert into marcrecorddone values ($q_origisbn, $q_origissn, $q_origlccn)"); + $sth->execute; + my $sth=$dbh->prepare("select biblionumber,biblioitemnumber from biblioitems where issn=$q_issn or isbn=$q_isbn or lccn=$q_lccn"); + $sth->execute; + my $biblionumber=0; + my $biblioitemnumber=0; + print "
\n"; + print "New Record | New File
\n"; + if ($sth->rows) { + ($biblionumber, $biblioitemnumber) = $sth->fetchrow; + my $title=$input->param('title'); + print << "EOF"; + + + +
Record already in database
$title is already in the database with biblionumber $biblionumber and biblioitemnumber $biblioitemnumber
+

+EOF + } else { + my $q_title=$dbh->quote($input->param('title')); + my $q_subtitle=$dbh->quote($input->param('subtitle')); + my $q_author=$dbh->quote($input->param('author')); + my $q_copyrightdate=$dbh->quote($input->param('copyrightdate')); + my $q_seriestitle=$dbh->quote($input->param('seriestitle')); + $sth=$dbh->prepare("select biblionumber from biblio where title=$q_title and author=$q_author and copyrightdate=$q_copyrightdate and seriestitle=$q_seriestitle"); + $sth->execute; + if ($sth->rows) { + ($biblionumber) = $sth->fetchrow; + } else { + $sth=$dbh->prepare("select max(biblionumber) from biblio"); + $sth->execute; + ($biblionumber) = $sth->fetchrow; + $biblionumber++; + $sth=$dbh->prepare("insert into biblio (biblionumber, title, author, copyrightdate, seriestitle) values ($biblionumber, $q_title, $q_author, $q_copyrightdate, $q_seriestitle)"); + $sth->execute; + $sth=$dbh->prepare("insert into bibliosubtitle (biblionumber,subtitle) values ($biblionumber, $q_subtitle)"); + $sth->execute; + } + $sth=$dbh->prepare("select max(biblioitemnumber) from biblioitems"); + $sth->execute; + ($biblioitemnumber) = $sth->fetchrow; + $biblioitemnumber++; + my $q_isbn=$dbh->quote($isbn); + my $q_issn=$dbh->quote($issn); + my $q_lccn=$dbh->quote($lccn); + my $q_volume=$dbh->quote($input->param('volume')); + my $q_number=$dbh->quote($input->param('number')); + my $q_itemtype=$dbh->quote($input->param('itemtype')); + my $q_dewey=$dbh->quote($input->param('dewey')); + my $q_subclass=$dbh->quote($input->param('subclass')); + my $q_publicationyear=$dbh->quote($input->param('publicationyear')); + my $q_publishercode=$dbh->quote($input->param('publishercode')); + my $q_volumedate=$dbh->quote($input->param('volumedate')); + my $q_volumeddesc=$dbh->quote($input->param('volumeddesc')); + my $q_illus=$dbh->quote($input->param('illustrator')); + my $q_pages=$dbh->quote($input->param('pages')); + my $q_notes=$dbh->quote($input->param('note')); + my $q_size=$dbh->quote($input->param('size')); + my $q_place=$dbh->quote($input->param('place')); + my $q_marc=$dbh->quote($input->param('marc')); + + $sth=$dbh->prepare("insert into biblioitems (biblioitemnumber, biblionumber, volume, number, itemtype, isbn, issn, dewey, subclass, publicationyear, publishercode, volumedate, volumeddesc, illus, pages, notes, size, place, lccn, marc) values ($biblioitemnumber, $biblionumber, $q_volume, $q_number, $q_itemtype, $q_isbn, $q_issn, $q_dewey, $q_subclass, $q_publicationyear, $q_publishercode, $q_volumedate, $q_volumeddesc, $q_illus, $q_pages, $q_notes, $q_size, $q_place, $q_lccn, $q_marc)"); + $sth->execute; + my $subjectheadings=$input->param('subject'); + my $additionalauthors=$input->param('additionalauthors'); + my @subjectheadings=split(/\n/,$subjectheadings); + my $subjectheading; + foreach $subjectheading (@subjectheadings) { + # remove any line ending characters (Ctrl-J or M) + $subjectheading=~s/\013//g; + $subjectheading=~s/\010//g; + # convert to upper case + $subjectheading=uc($subjectheading); + chomp ($subjectheading); + while (ord(substr($subjectheading, length($subjectheading)-1, 1))<14) { + chop $subjectheading; + } + # quote value + my $q_subjectheading=$dbh->quote($subjectheading); + $sth=$dbh->prepare("insert into bibliosubject (biblionumber,subject) + values ($biblionumber, $q_subjectheading)"); + $sth->execute; + } + my @additionalauthors=split(/\n/,$additionalauthors); + my $additionalauthor; + foreach $additionalauthor (@additionalauthors) { + # remove any line ending characters (Ctrl-L or Ctrl-M) + $additionalauthor=~s/\013//g; + $additionalauthor=~s/\010//g; + # convert to upper case + $additionalauthor=uc($additionalauthor); + # quote value + my $q_additionalauthor=$dbh->quote($additionalauthor); + $sth=$dbh->prepare("insert into additionalauthors (biblionumber,author) values ($biblionumber, $q_additionalauthor)"); + $sth->execute; + } + + my $title=$input->param('title'); + print << "EOF"; + + + +
Record entered into database
$title has been entered into the database with biblionumber + $biblionumber and biblioitemnumber $biblioitemnumber
+EOF + } + my $title=$input->param('title'); + $sth=$dbh->prepare("select max(barcode) from items"); + $sth->execute; + my ($barcode) = $sth->fetchrow; + $barcode++; + print << "EOF"; + + + +
+Add a New Item for $title + +
+
+ + + + + + + + +
BARCODE Home Branch:
Replacement Price:
Notes
+
+

+ + +EOF +print endmenu(); +print endpage(); + +exit; +} + +if ($input->param('newitem')) { + my $barcode=$input->param('barcode'); + my $q_barcode=$dbh->quote($barcode); + my $q_notes=$dbh->quote($input->param('notes')); + my $q_homebranch=$dbh->quote($input->param('homebranch')); + my $biblionumber=$input->param('biblionumber'); + my $biblioitemnumber=$input->param('biblioitemnumber'); + my $replacementprice=($input->param('replacementprice') || 0); + my $sth=$dbh->prepare("select max(itemnumber) from items"); + $sth->execute; + my ($itemnumber) = $sth->fetchrow; + $itemnumber++; + my @datearr=localtime(time); + my $date=(1900+$datearr[5])."-".($datearr[4]+1)."-".$datearr[3]; + $sth=$dbh->prepare("insert into items (itemnumber, biblionumber, biblioitemnumber, barcode, itemnotes, homebranch, holdingbranch, dateaccessioned, replacementprice) values ($itemnumber, $biblionumber, $biblioitemnumber, $q_barcode, $q_notes, $q_homebranch, 'STWE', '$date', $replacementprice)"); + $sth->execute; +} + + +if ($file) { + my $qisbn=$input->param('isbn'); + my $qissn=$input->param('issn'); + my $qlccn=$input->param('lccn'); + my $qcontrolnumber=$input->param('controlnumber'); + if ($qisbn || $qissn || $qlccn || $qcontrolnumber) { + print "New File

\n"; + open (F, "$file"); + my $data=; + + $splitchar=chr(29); + my @records; + foreach $record (split(/$splitchar/, $data)) { + my $marctext="\n"; + $marctext.="\n"; + $leader=substr($record,0,24); + $marctext.="\n"; + $record=substr($record,24); + $splitchar2=chr(30); + my $directory=0; + my $tagcounter=0; + my %tag; + my @record; + foreach $field (split(/$splitchar2/, $record)) { + my %field; + ($color eq $lc1) ? ($color=$lc2) : ($color=$lc1); + unless ($directory) { + $directory=$field; + my $itemcounter=1; + $counter=0; + while ($item=substr($directory,0,12)) { + $tag=substr($directory,0,3); + $length=substr($directory,3,4); + $start=substr($directory,7,6); + $directory=substr($directory,12); + $tag{$counter}=$tag; + $counter++; + } + $directory=1; + next; + } + $tag=$tag{$tagcounter}; + $tagcounter++; + $field{'tag'}=$tag; + $marctext.=""; + $splitchar3=chr(31); + my @subfields=split(/$splitchar3/, $field); + $indicator=$subfields[0]; + $field{'indicator'}=$indicator; + my $firstline=1; + if ($#subfields==0) { + $marctext.=""; + } else { + my %subfields; + $marctext.="\n"; + $field{'subfields'}=\%subfields; + } + push (@record, \%field); + } + $marctext.="
MARC RECORD
Leader:$leader
$tagtext{$tag}$tag$indicator
\n"; + my $color2=$color; + for ($i=1; $i<=$#subfields; $i++) { + ($color2 eq $lc1) ? ($color2=$lc2) : ($color2=$lc1); + my $text=$subfields[$i]; + my $subfieldcode=substr($text,0,1); + my $subfield=substr($text,1); + $marctext.="\n"; + if ($subfields{$subfieldcode}) { + my $subfieldlist=$subfields{$subfieldcode}; + my @subfieldlist=@$subfieldlist; + if ($#subfieldlist>=0) { + push (@subfieldlist, $subfield); + } else { + @subfieldlist=($subfields{$subfieldcode}, $subfield); + } + $subfields{$subfieldcode}=\@subfieldlist; + } else { + $subfields{$subfieldcode}=$subfield; + } + } + $marctext.="
$subfieldcode
$subfield
\n"; + $marctext{\@record}=$marctext; + $marc{\@record}=$record; + push (@records, \@record); + $counter++; + } +RECORD: + foreach $record (@records) { + my ($lccn, $isbn, $issn, $dewey, $author, $title, $place, $publisher, $publicationyear, $volume, $number, @subjects, $note, $additionalauthors, $illustrator, $copyrightdate, $seriestitle); + my $marctext=$marctext{$record}; + my $marc=$marc{$record}; + foreach $field (@$record) { + if ($field->{'tag'} eq '001') { + $controlnumber=$field->{'indicator'}; + } + if ($field->{'tag'} eq '010') { + $lccn=$field->{'subfields'}->{'a'}; + $lccn=~s/^\s*//; + ($lccn) = (split(/\s+/, $lccn))[0]; + } + if ($field->{'tag'} eq '015') { + $lccn=$field->{'subfields'}->{'a'}; + $lccn=~s/^\s*//; + $lccn=~s/^C//; + ($lccn) = (split(/\s+/, $lccn))[0]; + } + if ($field->{'tag'} eq '020') { + $isbn=$field->{'subfields'}->{'a'}; + ($isbn=~/^ARRAY/) && ($isbn=$$isbn[0]); + $isbn=~s/[^\d]*//g; + } + if ($field->{'tag'} eq '022') { + $issn=$field->{'subfields'}->{'a'}; + $issn=~s/^\s*//; + ($issn) = (split(/\s+/, $issn))[0]; + } + if ($field->{'tag'} eq '082') { + $dewey=$field->{'subfields'}->{'a'}; + $dewey=~s/\///g; + if (@$dewey) { + $dewey=$$dewey[0]; + } + #$dewey=~s/\///g; + } + if ($field->{'tag'} eq '100') { + $author=$field->{'subfields'}->{'a'}; + } + if ($field->{'tag'} eq '245') { + $title=$field->{'subfields'}->{'a'}; + $title=~s/ \/$//; + $subtitle=$field->{'subfields'}->{'b'}; + $subtitle=~s/ \/$//; + } + if ($field->{'tag'} eq '260') { + $place=$field->{'subfields'}->{'a'}; + if (@$place) { + $place=$$place[0]; + } + $place=~s/\s*:$//g; + $publisher=$field->{'subfields'}->{'b'}; + if (@$publisher) { + $publisher=$$publisher[0]; + } + $publisher=~s/\s*:$//g; + $publicationyear=$field->{'subfields'}->{'c'}; + if ($publicationyear=~/c(\d\d\d\d)/) { + $copyrightdate=$1; + } + if ($publicationyear=~/[^c](\d\d\d\d)/) { + $publicationyear=$1; + } elsif ($copyrightdate) { + $publicationyear=$copyrightdate; + } else { + $publicationyear=~/(\d\d\d\d)/; + $publicationyear=$1; + } + } + if ($field->{'tag'} eq '300') { + $pages=$field->{'subfields'}->{'a'}; + $pages=~s/ \;$//; + $size=$field->{'subfields'}->{'c'}; + $pages=~s/\s*:$//g; + $size=~s/\s*:$//g; + } + if ($field->{'tag'} eq '362') { + if ($field->{'subfields'}->{'a'}=~/(\d+).*(\d+)/) { + $volume=$1; + $number=$2; + } + } + if ($field->{'tag'} eq '440') { + $seriestitle=$field->{'subfields'}->{'a'}; + if ($field->{'subfields'}->{'v'}=~/(\d+).*(\d+)/) { + $volume=$1; + $number=$2; + } + } + if ($field->{'tag'} eq '700') { + my $name=$field->{'subfields'}->{'a'}; + if ($field->{'subfields'}->{'c'}=~/ill/) { + $additionalauthors.="$name\n"; + } else { + $illustrator=$name; + } + } + if ($field->{'tag'} =~/^5/) { + $note.="$field->{'subfields'}->{'a'}\n"; + } + if ($field->{'tag'} =~/65\d/) { + my $subject=$field->{'subfields'}->{'a'}; + $subject=~s/\.$//; + if ($gensubdivision=$field->{'subfields'}->{'x'}) { + my @sub=@$gensubdivision; + if ($#sub>=0) { + foreach $s (@sub) { + $s=~s/\.$//; + $subject.=" -- $s"; + } + } else { + $gensubdivision=~s/\.$//; + $subject.=" -- $gensubdivision"; + } + } + if ($chronsubdivision=$field->{'subfields'}->{'y'}) { + my @sub=@$chronsubdivision; + if ($#sub>=0) { + foreach $s (@sub) { + $s=~s/\.$//; + $subject.=" -- $s"; + } + } else { + $chronsubdivision=~s/\.$//; + $subject.=" -- $chronsubdivision"; + } + } + if ($geosubdivision=$field->{'subfields'}->{'z'}) { + my @sub=@$geosubdivision; + if ($#sub>=0) { + foreach $s (@sub) { + $s=~s/\.$//; + $subject.=" -- $s"; + } + } else { + $geosubdivision=~s/\.$//; + $subject.=" -- $geosubdivision"; + } + } + push @subjects, $subject; + } + } + $titleinput=$input->textfield(-name=>'title', -default=>$title, -size=>40); + $marcinput=$input->hidden(-name=>'marc', -default=>$marc); + $subtitleinput=$input->textfield(-name=>'subtitle', -default=>$subtitle, -size=>40); + $authorinput=$input->textfield(-name=>'author', -default=>$author); + $illustratorinput=$input->textfield(-name=>'illustrator', -default=>$illustrator); + $additionalauthorsinput=$input->textarea(-name=>'additionalauthors', -default=>$additionalauthors, -rows=>4, -cols=>20); + my $subject=''; + foreach (@subjects) { + $subject.="$_\n"; + } + $subjectinput=$input->textarea(-name=>'subject', -default=>$subject, -rows=>4, -cols=>40); + $noteinput=$input->textarea(-name=>'note', -default=>$note, -rows=>4, -cols=>40, -wrap=>'physical'); + $copyrightinput=$input->textfield(-name=>'copyrightdate', -default=>$copyrightdate); + $seriestitleinput=$input->textfield(-name=>'seriestitle', -default=>$seriestitle); + $volumeinput=$input->textfield(-name=>'volume', -default=>$volume); + $volumedateinput=$input->textfield(-name=>'volumedate', -default=>$volumedate); + $volumeddescinput=$input->textfield(-name=>'volumeddesc', -default=>$volumeddesc); + $numberinput=$input->textfield(-name=>'number', -default=>$number); + $isbninput=$input->textfield(-name=>'isbn', -default=>$isbn); + $issninput=$input->textfield(-name=>'issn', -default=>$issn); + $lccninput=$input->textfield(-name=>'lccn', -default=>$lccn); + $isbninput=$input->textfield(-name=>'isbn', -default=>$isbn); + $deweyinput=$input->textfield(-name=>'dewey', -default=>$dewey); + $cleanauthor=$author; + $cleanauthor=~s/[^A-Za-z]//g; + $subclassinput=$input->textfield(-name=>'subclass', -default=>uc(substr($cleanauthor,0,3))); + $publisherinput=$input->textfield(-name=>'publishercode', -default=>$publisher); + $pubyearinput=$input->textfield(-name=>'publicationyear', -default=>$publicationyear); + $placeinput=$input->textfield(-name=>'place', -default=>$place); + $pagesinput=$input->textfield(-name=>'pages', -default=>$pages); + $sizeinput=$input->textfield(-name=>'size', -default=>$size); + $fileinput=$input->hidden(-name=>'file', -default=>$file); + $origisbn=$input->hidden(-name=>'origisbn', -default=>$isbn); + $origissn=$input->hidden(-name=>'origissn', -default=>$issn); + $origlccn=$input->hidden(-name=>'origlccn', -default=>$lccn); + $origcontrolnumber=$input->hidden(-name=>'origcontrolnumber', -default=>$controlnumber); + + my $itemtypeselect=''; + $sth=$dbh->prepare("select itemtype,description from itemtypes"); + $sth->execute; + while (my ($itemtype, $description) = $sth->fetchrow) { + $itemtypeselect.="