X-Git-Url: http://git.equinoxoli.org/?p=migration-tools.git;a=blobdiff_plain;f=Equinox-Migration%2Flib%2FEquinox%2FMigration%2FMapDrivenMARCXMLProc.pm;h=9c698a1dd545c3921ff30bb12f2fccc04c49e452;hp=ab053bba0fb782367d4ada7750f612ac5925e935;hb=ef8cc7d38891b4fab1ce59c2c9df60ed08f9446c;hpb=a85d1ead29a140aefd716b303f0f0e67a96cee75 diff --git a/Equinox-Migration/lib/Equinox/Migration/MapDrivenMARCXMLProc.pm b/Equinox-Migration/lib/Equinox/Migration/MapDrivenMARCXMLProc.pm index ab053bb..9c698a1 100644 --- a/Equinox-Migration/lib/Equinox/Migration/MapDrivenMARCXMLProc.pm +++ b/Equinox-Migration/lib/Equinox/Migration/MapDrivenMARCXMLProc.pm @@ -4,7 +4,9 @@ use warnings; use strict; use XML::Twig; -use Equinox::Migration::SubfieldMapper 1.002; +use DBM::Deep; +use Equinox::Migration::SubfieldMapper 1.004; + =head1 NAME @@ -12,12 +14,16 @@ Equinox::Migration::MapDrivenMARCXMLProc =head1 VERSION -Version 1.000 +Version 1.002 =cut -our $VERSION = '1.000'; +our $VERSION = '1.002'; +my $dstore; +my $sfmap; +my @mods = qw( multi bib required ); +my $verbose = 0; =head1 SYNOPSIS @@ -38,48 +44,31 @@ and C (the MARC data to be processed). my $m = Equinox::Migration::MapDrivenMARCXMLProc->new( mapfile => FILE, marcfile => FILE ); -There is an optional third, argument, C, which specifies a -arrayref of datafields to "sample" by reporting on subfields which are -found in the data but not in the map. - - my $m = Equinox::Migration::MapDrivenMARCXMLProc->new( mapfile => FILE, - marcfile => FILE, - sample => \@TAGS - ); - -See L for more info. - =cut sub new { my ($class, %args) = @_; - my $self = bless { mods => { multi => {}, - once => {}, - required => {}, - }, - data => { recs => undef, # X::T record objects - rptr => 0, # next record pointer - crec => undef, # parsed record storage - stag => undef, # list of tags to sample - umap => undef, # unmapped data samples - }, + my $self = bless { }, $class; # initialize map and taglist - die "Argument 'mapfile' must be specified\n" unless (defined $args{mapfile}); - my @mods = keys %{$self->{mods}}; - $self->{map} = Equinox::Migration::SubfieldMapper->new( file => $args{mapfile}, - mods => \@mods ); - $self->{data}{tags} = $self->{map}->tags; + die "Argument 'mapfile' must be specified\n" unless ($args{mapfile}); + $sfmap = Equinox::Migration::SubfieldMapper->new( file => $args{mapfile}, + mods => \@mods ); + + # initialize datastore + $dstore = DBM::Deep->new( file => "EMMXSSTORAGE.dbmd", + data_sector_size => 256 ); + $dstore->{rcnt} = 0; # next record ptr + $dstore->{tags} = $sfmap->tags; # list of all tags + $self->{data} = $dstore; # initialize twig - die "Argument 'marcfile' must be specified\n" unless (defined $args{marcfile}); + die "Argument 'marcfile' must be specified\n" unless ($args{marcfile}); if (-r $args{marcfile}) { - $self->{twig} = XML::Twig->new; - $self->{twig}->parsefile($args{marcfile}); - my @records = $self->{twig}->root->children; - $self->{data}{recs} = \@records; + my $xmltwig = XML::Twig->new( twig_handlers => { record => \&parse_record } ); + $xmltwig->parsefile( $args{marcfile} ); } else { die "Can't open marc file: $!\n"; } @@ -87,43 +76,34 @@ sub new { return $self; } +sub DESTROY { unlink "EMMXSSTORAGE.dbmd" } =head2 parse_record -Extracts data from the next record, per the mapping file. Returns a -normalized datastructure (see L for details) on -success; returns 0 otherwise. - - while (my $rec = $m->parse_record) { - # handle extracted record data - } +Extracts data from the next record, per the mapping file. =cut sub parse_record { - my ($self) = @_; - - # get the next record and wipe current parsed record - return 0 unless defined $self->{data}{recs}[ $self->{data}{rptr} ]; - my $record = $self->{data}{recs}[ $self->{data}{rptr} ]; - $self->{data}{crec} = { egid => undef, bib => undef, tags => undef }; + my ($twig, $record) = @_; + my $crec = {}; # current record my @fields = $record->children; for my $f (@fields) - { $self->process_field($f) } + { process_field($f, $crec) } # cleanup memory and increment pointer $record->purge; - $self->{data}{rptr}++; + $dstore->{rcnt}++; - return $self->{data}{crec}; + # check for required fields + check_required(); + push @{ $dstore->{recs} }, $crec; } sub process_field { - my ($self, $field) = @_; - my $map = $self->{map}; + my ($field, $crec) = @_; my $tag = $field->{'att'}->{'tag'}; - my $crec = $self->{data}{crec}; # leader unless (defined $tag) { @@ -137,55 +117,122 @@ sub process_field { $crec->{egid} = $sub->text; return; } - if ($map->has($tag)) { + if ($sfmap->has($tag)) { push @{$crec->{tags}}, { tag => $tag, uni => undef, multi => undef }; + push @{$crec->{tmap}{$tag}}, (@{$crec->{tags}} - 1); my @subs = $field->children('subfield'); for my $sub (@subs) - { $self->process_subs($tag, $sub) } - # check map to ensure all declared subs have a value - my $mods = $map->mods($field); - for my $mappedsub ( @{ $map->subfields($tag) } ) { + { process_subs($tag, $sub, $crec) } + + # check map to ensure all declared tags and subs have a value + my $mods = $sfmap->mods($field); + for my $mappedsub ( @{ $sfmap->subfields($tag) } ) { next if $mods->{multi}; $crec->{tags}[-1]{uni}{$mappedsub} = '' unless defined $crec->{tags}[-1]{uni}{$mappedsub}; } + for my $mappedtag ( @{ $sfmap->tags }) { + $crec->{tmap}{$mappedtag} = undef + unless defined $crec->{tmap}{$mappedtag}; + } } } sub process_subs { - my ($self, $tag, $sub) = @_; - my $map = $self->{map}; + my ($tag, $sub, $crec) = @_; my $code = $sub->{'att'}->{'code'}; # handle unmapped tag/subs - unless ($map->has($tag, $code)) { - my $u = $self->{data}{umap}; - my $s = $self->{data}{stag}; - return unless (defined $s->{$tag}); - - # set a value, total-seen count and records-seen-in count - $u->{$tag}{$code}{value} = $sub->text unless defined $u->{$tag}{$code}; - $u->{$tag}{$code}{count}++; - $u->{$tag}{$code}{rcnt}++ unless ($u->{$tag}{$code}{last} == $self->{data}{rptr}); - $u->{$tag}{$code}{last} = $self->{data}{rptr}; - return; - } + return unless ($sfmap->has($tag, $code)); # fetch our datafield struct and fieldname - my $dataf = $self->{data}{crec}{tags}[-1]; - my $field = $map->field($tag, $code); + my $dataf = $crec->{tags}[-1]; + my $field = $sfmap->field($tag, $code); + $crec->{names}{$tag}{$code} = $field; - # handle modifiers, or slug data in normally - if (my $mods = $map->mods($field)) { + # test filters + for my $filter ( @{$sfmap->filters($field)} ) { + return if ($sub->text =~ /$filter/i); + } + # handle multi modifier + if (my $mods = $sfmap->mods($field)) { if ($mods->{multi}) { - my $name = $tag . $code; - push @{$dataf->{multi}{$name}}, $sub->text; + push @{$dataf->{multi}{$code}}, $sub->text; + return; } - } else { - $dataf->{uni}{$code} = $sub->text; } + + # if this were a multi field, it would be handled already. make sure its a singleton + die "Multiple occurances of a non-multi field: $tag$code at rec ", + ($dstore->{rcnt} + 1),"\n" if (defined $dataf->{uni}{$code}); + + # everything seems okay + $dataf->{uni}{$code} = $sub->text; } + +sub check_required { + my $mods = $sfmap->mods; + my $crec = $dstore->{crec}; + + for my $tag_id (keys %{$mods->{required}}) { + for my $code (@{$mods->{required}{$tag_id}}) { + my $found = 0; + + for my $tag (@{$crec->{tags}}) { + $found = 1 if ($tag->{multi}{($tag_id . $code)}); + $found = 1 if ($tag->{uni}{$code}); + } + + die "Required mapping $tag_id$code not found in rec ",$dstore->{rcnt},"\n" + unless ($found); + } + } + +} + +=head2 recno + +Returns current record number (starting from zero) + +=cut + +sub recno { my ($self) = @_; return $self->{data}{rcnt} } + +=head2 name + +Returns mapped fieldname when passed a record number, tag, and code + + my $name = $m->name(3,999,'a'); + +=cut + +sub name { my ($self, $r, $t, $c) = @_; return $dstore->{recs}[$r]{names}{$t}{$c} }; + +=head1 MODIFIERS + +MapDrivenMARCXMLProc implements the following modifiers, and passes +them to L, meaning that specifying +any other modifiers in a MDMP map file will cause a fatal error when +it is processed. + +=head2 multi + +If a mapping is declared to be C, then MDMP expects to see more +than one instance of that subfield per datafield, and the data is +handled accordingly (see L below). + +Occurring zero or one time is legal for a C mapping. + +A mapping which is not flagged as C, but which occurs more than +once per datafield will cause a fatal error. + +=head2 required + +By default, if a mapping does not occur in a datafield, processing +continues normally. if a mapping has the C modifier, +however, it must appear, or a fatal error will occur. + =head1 PARSED RECORDS Given: @@ -196,36 +243,21 @@ Given: Then C<$rec> will look like: { - egid => evergreen_record_id, - bib => { - (tag_id . sub_code)1 => value1, - (tag_id . sub_code)2 => value2, - ... - }, + egid => evergreen_record_id, tags => [ { tag => tag_id, - multi => { (tag_id . sub_code) => [ val1, val2, ... ] }, + multi => { code => [ val1, val2, ... ] }, uni => { code => value, code2 => value2, ... }, }, ... - ] + ], + tmap => { tag_id => [ INDEX_LIST ], tag_id2 => [ INDEX_LIST ], ... } } That is, there is an C key which points to the Evergreen ID of -that record, a C key which points to a hashref, and a C -key which points to an arrayref. - -=head3 C - -A reference to a hash which holds extracted data which occurs only -once per record (and is therefore "bib-level"; the default assumption -is that a tag/subfield pair can occur multiple times per record). The -keys are composed of tag id and subfield code, catenated -(e.g. 901c). The values are the contents of that subfield of that tag. - -If there are no tags defined as bib-level in the mapfile, C will -be C. +that record, a C key which points to an arrayref, and a C +key which points to a hashref. =head3 C @@ -236,9 +268,8 @@ Each tag hash holds its own id (e.g. C<998>), and two references to two more hashrefs, C and C. The C hash holds the extracted data for tag/sub mappings which -have the C modifier on them. The keys in C are -composed of the tag id and subfield code, catenated -(e.g. C<901c>). The values are arrayrefs containing the content of all +have the C modifier on them. The keys in C subfield +codes. The values are arrayrefs containing the content of all instances of that subfield in that instance of that tag. If no tags are defined as C, it will be C. @@ -253,15 +284,22 @@ datafield will be given a value of '' (the null string) in the current record struct. Oppose subfields which are not mapped, which will be C. -=head1 UNMAPPED TAGS +=head3 tmap - { tag_id => { - sub_code => { value => VALUE, count => COUNT }, - sub_code2 => { value => VALUE, count => COUNT }, - ... - }, - ... - } +A hashref, where each key (a tag id like "650") points to a listref +containing the index (or indices) of C where that tag has +extracted data. + +The intended use of this is to simplify the processing of data from +tags which can appear more than once in a MARC record, like +holdings. If your holdings data is in 852, C{852}> will be a +listref with the indices of C which hold the data from the 852 +datafields. + +Complimentarily, C prevents data from singular datafields from +having to be copied for every instance of a multiple datafield, as it +lets you get the data from that record's one instance of whichever +field you're looking for. =head1 AUTHOR