--- trunk/parse_format.pm 2003/02/23 00:01:08 22 +++ trunk/parse_format.pm 2003/07/05 23:39:04 78 @@ -1,55 +1,191 @@ #------------------------------------------------------------- # -# parse_format('format',$isis_row); +# parse_format(...) # -use isis_sf; sub parse_format { + my $type = shift || die "parset_format must be called with type!"; + my $format = shift || die "parse_format must be called with format!"; + my $row = shift || die "parse_format must be called with row!"; + my $i = shift || 0; # isis repeatable number + my $codepage = shift || die "parse_format must be called with codepage!"; + if ($type eq "isis") { + return parse_iso_format($format,$row,$i,$codepage,'isis_sf'); + } elsif ($type eq "excel") { + return parse_excel_format($format,$row,$i,$codepage); + } elsif ($type eq "marc") { + return parse_iso_format($format,$row,$i,$codepage,'marc_sf'); + } elsif ($type eq "feed") { + return parse_feed_format($format,$row,$i,$codepage); + } +} + +#------------------------------------------------------------- + +sub parse_iso_format { + my $format = shift; my $row = shift; + my $i = shift; + my $codepage = shift; + + my $func = shift || die "need to know which sub-field function to use"; + + require $func.".pm"; my $out; my $out_swish; my $prefix = ""; if ($format =~ s/^([^\d]+)//) { - $prefix = "pre: $1"; + $prefix = $1; } my $display; my $swish; + sub cnv_cp { + my $codepage = shift; + my $tmp = shift || return; + if ($codepage) { + $tmp = $codepage->convert($tmp) || print STDERR "iso: '$tmp' can't convert\n"; + } + return $tmp; + } + while ($format) { -#print STDERR "#### $format\n"; - if ($format =~ s/^(\d\d\d)(\w?)//) { - my $isis_tmp = isis_sf($row,$1,$2); - if ($isis_tmp) { - $display .= $prefix . $isis_tmp; - $swish .= $isis_tmp." "; +#print STDERR "\n#### $format"; + # this is EBSCO special to support numeric subfield in + # form of 856#3 + if ($format =~ s/^(\d\d\d)#*(\w?)//) { + my $tmp = cnv_cp($codepage,get_sf($row,$1,$2,$i)); + if ($tmp) { + $display .= $prefix.$tmp; + $swish .= $tmp." "; +#print STDERR " == $tmp"; } $prefix = ""; # this might be our local scpeciality -- fields 10 and 11 # (as opposed to 010 and 011) so they are strictly listed # here } elsif ($format =~ s/^(1[01])//) { - my $isis_tmp = isis_sf($row,$1,$2); - if ($isis_tmp) { - $display .= $prefix . $isis_tmp; - $swish .= $isis_tmp." "; + my $tmp = cnv_cp($codepage,get_sf($row,$1,undef,$i)); + if ($tmp) { + $display .= $prefix.$tmp; + $swish .= $tmp." "; } $prefix = ""; + } elsif ($format =~ s/^mfn//i) { + $display .= $prefix . $row->{mfn}; + $prefix = ""; } elsif ($format =~ s/^([^\d]+)(\d{0,3})/$2/) { - $prefix .= $1; + $prefix .= $1 if ($display); } elsif ($format =~ s/^([^\d]+\d{0,2})//) { - $prefix .= $1; + $prefix .= $1 if ($display); } elsif ($format =~ s/^(\d{1,2})//) { - $prefix .= $1; + $prefix .= $1 if ($display); + } else { + print STDERR "unparsed format: $format\n"; + $prefix .= $format; + $format = ""; + } + } + # add suffix + $display .= $prefix if ($display); + + return ($swish,$display); +} + +#------------------------------------------------------------- + +sub parse_excel_format { + my $format = shift; + my $row = shift; + my $i = shift; + my $codepage = shift; + + my $out; + my $out_swish; + + my $prefix = ""; + if ($format =~ s/^([^A-Z\|]{1,3})//) { + $prefix = $1; + } + + my $display; + my $swish; + + while ($format && length($format) > 0) { +#print STDERR "\n#### $format #"; + if ($format =~ s/^\|([A-Z]{1,2})\|//) { +#print STDERR "--$1-> $format -[",length($format),"] "; + if ($row->{$1}) { + my $tmp = $row->{$1}; + if ($codepage) { + $tmp = $codepage->convert($tmp) || warn "excel: $1 '$tmp' can't convert"; + } + $display .= $prefix . $tmp; + $swish .= $tmp." "; +#print STDERR " == $tmp"; + } + $prefix = ""; + } elsif ($format =~ s/^([^A-Z\|]+)(\|[A-Z]{1,2}\|)/$2/) { + $prefix .= $1 if ($display); + } else { + print STDERR "unparsed format: $format\n"; + $prefix .= $format; + $format = ""; + } +#print STDERR " display: $display swish: $swish [format: $format]"; + } + # add suffix + $display .= $prefix if ($display); + + return ($swish,$display); +} + +#------------------------------------------------------------- + +sub parse_feed_format { + my $format = shift; + my $data = shift; + my $i = shift; + my $codepage = shift; + + my $out; + my $out_swish; + + my $prefix = ""; + if ($format =~ s/^([^\d\|]{1,3})//) { + $prefix = $1; + } + + my $display; + my $swish; + + while ($format && length($format) > 0) { +#print STDERR "\n#### $format #"; + if ($format =~ s/^\|(\d+)\|//) { +#print STDERR "--$1-> $format -[",length($format),"] "; + if ($data->{$1}) { + my $tmp = $data->{$1}; + if ($codepage) { + $tmp = $codepage->convert($tmp) || warn "feed: $1 '$tmp' can't convert\n"; + } + $display .= $prefix . $tmp; + $swish .= $tmp." "; +#print STDERR " == $tmp"; + } + $prefix = ""; + } elsif ($format =~ s/^([^\d\|]+)(\|\d+\|)/$2/) { + $prefix .= $1 if ($display); } else { print STDERR "unparsed format: $format\n"; $prefix .= $format; $format = ""; } +#print STDERR " display: $display swish: $swish [format: $format]"; } # add suffix $display .= $prefix if ($display); @@ -58,4 +194,5 @@ } #------------------------------------------------------------- + 1;