--- trunk/IsisDB.pm 2004/12/30 17:16:34 16 +++ trunk/lib/Biblio/Isis.pm 2006/07/09 12:18:44 58 @@ -1,13 +1,13 @@ -package IsisDB; +package Biblio::Isis; use strict; use Carp; -use Data::Dumper; +use File::Glob qw(:globally :nocase); BEGIN { use Exporter (); use vars qw ($VERSION @ISA @EXPORT @EXPORT_OK %EXPORT_TAGS); - $VERSION = 0.06; + $VERSION = 0.21; @ISA = qw (Exporter); #Give a hoot don't pollute, do not export more than needed by default @EXPORT = qw (); @@ -18,24 +18,25 @@ =head1 NAME -IsisDB - Read CDS/ISIS, WinISIS and IsisMarc database +Biblio::Isis - Read CDS/ISIS, WinISIS and IsisMarc database =head1 SYNOPSIS - use IsisDB; + use Biblio::Isis; - my $isis = new IsisDB( + my $isis = new Biblio::Isis( isisdb => './cds/cds', ); - for(my $mfn = 1; $mfn <= $isis->{'maxmfn'}; $mfn++) { + for(my $mfn = 1; $mfn <= $isis->count; $mfn++) { print $isis->to_ascii($mfn),"\n"; } =head1 DESCRIPTION This module will read ISIS databases created by DOS CDS/ISIS, WinIsis or -IsisMarc. It can be used as perl-only alternative to OpenIsis module. +IsisMarc. It can be used as perl-only alternative to OpenIsis module which +seems to depriciate it's old C bindings for perl. It can create hash values from data in ISIS database (using C), ASCII dump (using C) or just hash with field names and packed @@ -48,7 +49,7 @@ It also has support for identifiers (only if ISIS database is created by IsisMarc), see C. -This will module will always be slower than OpenIsis module which use C +This module will always be slower than OpenIsis module which use C library. However, since it's written in perl, it's platform independent (so you don't need C compiler), and can be easily modified. I hope that it creates data structures which are easier to use than ones created by @@ -78,7 +79,7 @@ Open ISIS database - my $isis = new IsisDB( + my $isis = new Biblio::Isis( isisdb => './cds/cds', read_fdt => 1, include_deleted => 1, @@ -87,6 +88,7 @@ $v =~ s#foo#bar#g; }, debug => 1, + join_subfields_with => ' ; ', ); Options are described below: @@ -96,7 +98,8 @@ =item isisdb This is full or relative path to ISIS database files which include -common prefix of C<.FDT>, C<.MST>, C<.CNT>, C<.XRF> and C<.MST> files. +common prefix of C<.MST>, and C<.XRF> and optionally C<.FDT> (if using +C option) files. In this example it uses C<./cds/cds.MST> and related files. @@ -115,11 +118,15 @@ =item debug -Dump a B of debugging output. +Dump a B of debugging output even at level 1. For even more increase level. -=back +=item join_subfields_with + +Define delimiter which will be used to join repeatable subfields. This +option is included to support lagacy application written against version +older than 0.21 of this module. By default, it disabled. See L. -It will also set C<$isis-E{'maxmfn'}> which is maximum MFN stored in database. +=back =cut @@ -134,15 +141,44 @@ $self->{$v} = {@_}->{$v}; } + my @isis_files = grep(/\.(FDT|MST|XRF|CNT)$/i,glob($self->{isisdb}."*")); + + foreach my $f (@isis_files) { + my $ext = $1 if ($f =~ m/\.(\w\w\w)$/); + $self->{lc($ext)."_file"} = $f; + } + + my @must_exist = qw(mst xrf); + push @must_exist, "fdt" if ($self->{read_fdt}); + + foreach my $ext (@must_exist) { + unless ($self->{$ext."_file"}) { + carp "missing ",uc($ext)," file in ",$self->{isisdb}; + return; + } + } + + if ($self->{debug}) { + print STDERR "## using files: ",join(" ",@isis_files),"\n"; + eval "use Data::Dump"; + + if (! $@) { + *Dumper = *Data::Dump::dump; + } else { + use Data::Dumper; + } + } + # if you want to read .FDT file use read_fdt argument when creating class! - if ({@_}->{read_fdt} && -e $self->{isisdb}.".FDT") { + if ($self->{read_fdt} && -e $self->{fdt_file}) { # read the $db.FDT file for tags my $fieldzone=0; - open(fileFDT, $self->{isisdb}.".FDT") || croak "can't read '$self->{isisdb}.FDT': $!"; + open(my $fileFDT, $self->{fdt_file}) || croak "can't read '$self->{fdt_file}': $!"; + binmode($fileFDT); - while () { + while (<$fileFDT>) { chomp; if ($fieldzone) { my $name=substr($_,0,30); @@ -159,12 +195,13 @@ } } - close(fileFDT); + close($fileFDT); } # Get the Maximum MFN from $db.MST - open(fileMST,$self->{isisdb}.".MST") || croak "can't read '$self->{isisdb}.MST': $!"; + open($self->{'fileMST'}, $self->{mst_file}) || croak "can't open '$self->{mst_file}': $!"; + binmode($self->{'fileMST'}); # MST format: (* = 32 bit signed) # CTLMFN* always 0 @@ -172,69 +209,33 @@ # NXTMFB* last block allocated to master file # NXTMFP offset to next available position in last block # MFTYPE always 0 for user db file (1 for system) - seek(fileMST,4,0); + seek($self->{'fileMST'},4,0) || croak "can't seek to offset 0 in MST: $!"; my $buff; - read(fileMST, $buff, 4); - $self->{'NXTMFN'}=unpack("l",$buff) || carp "NXTNFN is zero"; - - # save maximum MFN - $self->{'maxmfn'} = $self->{'NXTMFN'} - 1; - - close(fileMST); - - # Get the index information from $db.CNT - - open(fileCNT, $self->{isisdb}.".CNT") || croak "can't read '$self->{isisdb}.CNT': $!"; - - # There is two 26 Bytes fixed lenght records + read($self->{'fileMST'}, $buff, 4) || croak "can't read NXTMFN from MST: $!"; + $self->{'NXTMFN'}=unpack("V",$buff) || croak "NXTNFN is zero"; - # 0: IDTYPE BTree type 16 - # 2: ORDN Nodes Order 16 - # 4: ORDF Leafs Order 16 - # 6: N Number of Memory buffers for nodes 16 - # 8: K Number of buffers for first level index 16 - # 10: LIV Current number of Index Levels 16 - # 12: POSRX* Pointer to Root Record in N0x 32 - # 16: NMAXPOS* Next Available position in N0x 32 - # 20: FMAXPOS* Next available position in L0x 32 - # 24: ABNORMAL Formal BTree normality indicator 16 - # length: 26 bytes - - sub unpack_cnt { - my $self = shift; - - my @flds = qw(ORDN ORDF N K LIV POSRX NMAXPOS FMAXPOS ABNORMAL); - - my $buff = shift || return; - my @arr = unpack("ssssssllls", $buff); - - print STDERR "unpack_cnt: ",join(" ",@arr),"\n" if ($self->{'debug'}); - - my $IDTYPE = shift @arr; - foreach (@flds) { - $self->{$IDTYPE}->{$_} = abs(shift @arr); - } - } + print STDERR "## self ",Dumper($self),"\n" if ($self->{debug}); - read(fileCNT, $buff, 26); - $self->unpack_cnt($buff); - - read(fileCNT, $buff, 26); - $self->unpack_cnt($buff); + # open files for later + open($self->{'fileXRF'}, $self->{xrf_file}) || croak "can't open '$self->{xrf_file}': $!"; + binmode($self->{'fileXRF'}); + $self ? return $self : return undef; +} - close(fileCNT); +=head2 count - print STDERR Dumper($self),"\n" if ($self->{debug}); +Return number of records in database - # open files for later - open($self->{'fileXRF'}, $self->{isisdb}.".XRF") || croak "can't open '$self->{isisdb}.XRF': $!"; + print $isis->count; - open($self->{'fileMST'}, $self->{isisdb}.".MST") || croak "can't open '$self->{isisdb}.MST': $!"; +=cut - $self ? return $self : return undef; +sub count { + my $self = shift; + return $self->{'NXTMFN'} - 1; } =head2 fetch @@ -260,68 +261,81 @@ # is mfn allready in memory? my $old_mfn = $self->{'current_mfn'} || -1; - return if ($mfn == $old_mfn); + return $self->{record} if ($mfn == $old_mfn); print STDERR "## fetch: $mfn\n" if ($self->{debug}); # XXX check this? my $mfnpos=($mfn+int(($mfn-1)/127))*4; - print STDERR "## seeking to $mfnpos in file '$self->{isisdb}.XRF'\n" if ($self->{debug}); + print STDERR "## seeking to $mfnpos in file '$self->{xrf_file}'\n" if ($self->{debug}); seek($self->{'fileXRF'},$mfnpos,0); my $buff; + # delete old record + delete $self->{record}; + # read XRFMFB abd XRFMFP read($self->{'fileXRF'}, $buff, 4); - my $pointer=unpack("l",$buff) || carp "pointer is null"; + my $pointer=unpack("V",$buff); + if (! $pointer) { + if ($self->{include_deleted}) { + return; + } else { + warn "pointer for MFN $mfn is null\n"; + return; + } + } + + # check for logically deleted record + if ($pointer & 0x80000000) { + print STDERR "## record $mfn is logically deleted\n" if ($self->{debug}); + $self->{deleted} = $mfn; + + return unless $self->{include_deleted}; + + # abs + $pointer = ($pointer ^ 0xffffffff) + 1; + } my $XRFMFB = int($pointer/2048); my $XRFMFP = $pointer - ($XRFMFB*2048); - # (XRFMFB - 1) * 512 + XRFMFP # why do i have to do XRFMFP % 1024 ? - my $blk_off = (($XRFMFB - 1) * 512) + ($XRFMFP % 1024); + my $blk_off = (($XRFMFB - 1) * 512) + ($XRFMFP % 512); print STDERR "## pointer: $pointer XRFMFB: $XRFMFB XRFMFP: $XRFMFP offset: $blk_off\n" if ($self->{'debug'}); # Get Record Information - seek($self->{'fileMST'},$blk_off,0); + seek($self->{'fileMST'},$blk_off,0) || croak "can't seek to $blk_off: $!"; - read($self->{'fileMST'}, $buff, 4); - my $value=unpack("l",$buff); + read($self->{'fileMST'}, $buff, 4) || croak "can't read 4 bytes at offset $blk_off from MST file: $!"; + my $value=unpack("V",$buff); print STDERR "## offset for rowid $value is $blk_off (blk $XRFMFB off $XRFMFP)\n" if ($self->{debug}); if ($value!=$mfn) { - carp "Error: MFN ".$mfn." not found in MST(".$value.")"; - #return; # XXX deleted record? - } + if ($value == 0) { + print STDERR "## record $mfn is physically deleted\n" if ($self->{debug}); + $self->{deleted} = $mfn; + return; + } -# $MFRL=$self->Read16($fileMST); -# $MFBWB=$self->Read32($fileMST); -# $MFBWP=$self->Read16($fileMST); -# $BASE=$self->Read16($fileMST); -# $NVF=$self->Read16($fileMST); -# $STATUS=$self->Read16($fileMST); + carp "Error: MFN ".$mfn." not found in MST file, found $value"; + return; + } read($self->{'fileMST'}, $buff, 14); - my ($MFRL,$MFBWB,$MFBWP,$BASE,$NVF,$STATUS) = unpack("slssss", $buff); + my ($MFRL,$MFBWB,$MFBWP,$BASE,$NVF,$STATUS) = unpack("vVvvvv", $buff); print STDERR "## MFRL: $MFRL MFBWB: $MFBWB MFBWP: $MFBWP BASE: $BASE NVF: $NVF STATUS: $STATUS\n" if ($self->{debug}); - # delete old record - delete $self->{record}; - - ## FIXME this is a bug - if (! $self->{'include_deleted'} && $MFRL < 0) { - print "## logically deleted record $mfn, skipping...\n" if ($self->{debug}); - return; - } + warn "MFRL $MFRL is not even number" unless ($MFRL % 2 == 0); warn "BASE is not 18+6*NVF" unless ($BASE == 18 + 6 * $NVF); @@ -337,11 +351,7 @@ for (my $i = 0 ; $i < $NVF ; $i++) { -# $TAG=$self->Read16($fileMST); -# $POS=$self->Read16($fileMST); -# $LEN=$self->Read16($fileMST); - - my ($TAG,$POS,$LEN) = unpack("sss", substr($buff,$i * 6, 6)); + my ($TAG,$POS,$LEN) = unpack("vvv", substr($buff,$i * 6, 6)); print STDERR "## TAG: $TAG POS: $POS LEN: $LEN\n" if ($self->{debug}); @@ -374,22 +384,39 @@ push @{$self->{record}->{$FieldTAG[$i]}}, substr($buff,$FieldPOS[$i],$FieldLEN[$i]); } - close(fileMST); $self->{'current_mfn'} = $mfn; - print Dumper($self),"\n" if ($self->{debug}); + print STDERR Dumper($self),"\n" if ($self->{debug}); return $self->{'record'}; } +=head2 mfn + +Returns current MFN position + + my $mfn = $isis->mfn; + +=cut + +# This function should be simple return $self->{current_mfn}, +# but if new is called with _hack_mfn it becomes setter. +# It's useful in tests when setting $isis->{record} directly + +sub mfn { + my $self = shift; + return $self->{current_mfn}; +}; + + =head2 to_ascii -Dump ASCII output of record with specified MFN +Returns ASCII output of record with specified MFN print $isis->to_ascii(42); -It outputs something like this: +This outputs something like this: 210 ^aNew York^cNew York University press^dcop. 1988 990 2140 @@ -406,7 +433,7 @@ my $mfn = shift || croak "need MFN"; - my $rec = $self->fetch($mfn); + my $rec = $self->fetch($mfn) || return; my $out = "0\t$mfn"; @@ -426,7 +453,7 @@ my $hash = $isis->to_hash($mfn); -It has ability to convert characters (using C from ISIS +It has ability to convert characters (using C) from ISIS database before creating structures enabling character re-mapping or quick fix-up of data. @@ -462,42 +489,144 @@ } ], +In case there are repeatable subfields in record, this will create +following structure: + + '900' => [ { + 'a' => [ 'foo', 'bar', 'baz' ], + }] + +Or in more complex example of + + 902 ^aa1^aa2^aa3^bb1^aa4^bb2^cc1^aa5 + +it will create + + 902 => [ + { a => ["a1", "a2", "a3", "a4", "a5"], b => ["b1", "b2"], c => "c1" }, + ], + +This behaviour can be changed using C option to L, +in which case C will always create single value for each subfield. +This will change result to: + + + This method will also create additional field C<000> with MFN. +There is also more elaborative way to call C like this: + + my $hash = $isis->to_hash({ + mfn => 42, + include_subfields => 1, + }); + +Each option controll creation of hash: + +=over 4 + +=item mfn + +Specify MFN number of record + +=item include_subfields + +This option will create additional key in hash called C which will +have original record subfield order and index to that subfield like this: + + 902 => [ { + a => ["a1", "a2", "a3", "a4", "a5"], + b => ["b1", "b2"], + c => "c1", + subfields => ["a", 0, "a", 1, "a", 2, "b", 0, "a", 3, "b", 1, "c", 0, "a", 4], + } ], + +=item join_subfields_with + +Define delimiter which will be used to join repeatable subfields. You can +specify option here instead in L if you want to have per-record control. + +=back + =cut sub to_hash { my $self = shift; + my $mfn = shift || confess "need mfn!"; + my $arg; + + if (ref($mfn) eq 'HASH') { + $arg = $mfn; + $mfn = $arg->{mfn} || confess "need mfn in arguments"; + } # init record to include MFN as field 000 my $rec = { '000' => [ $mfn ] }; - my $row = $self->fetch($mfn); + my $row = $self->fetch($mfn) || return; + + my $j_rs = $arg->{join_subfields_with}; + $j_rs = $self->{join_subfields_with} unless(defined($j_rs)); + my $i_sf = $arg->{include_subfields}; - foreach my $k (keys %{$row}) { - foreach my $l (@{$row->{$k}}) { + foreach my $f_nr (keys %{$row}) { + foreach my $l (@{$row->{$f_nr}}) { # filter output - $l = $self->{'hash_filter'}->($l) if ($self->{'hash_filter'}); + if ($self->{'hash_filter'}) { + $l = $self->{'hash_filter'}->($l); + next unless defined($l); + } my $val; + my $r_sf; # repeatable subfields in this record # has identifiers? - ($val->{'i1'},$val->{'i2'}) = ($1,$2) if ($l =~ s/^([01 #])([01 #])//); + ($val->{'i1'},$val->{'i2'}) = ($1,$2) if ($l =~ s/^([01 #])([01 #])\^/\^/); # has subfields? if ($l =~ m/\^/) { foreach my $t (split(/\^/,$l)) { next if (! $t); - $val->{substr($t,0,1)} = substr($t,1); + my ($sf,$v) = (substr($t,0,1), substr($t,1)); + # XXX this might be option, but why? + next unless ($v); +# warn "### $f_nr^$sf:$v",$/ if ($self->{debug} > 1); + + if (ref( $val->{$sf} ) eq 'ARRAY') { + + push @{ $val->{$sf} }, $v; + + # record repeatable subfield it it's offset + push @{ $val->{subfields} }, ( $sf, $#{ $val->{$sf} } ) if (! $j_rs && $i_sf); + $r_sf->{$sf}++; + + } elsif (defined( $val->{$sf} )) { + + # convert scalar field to array + $val->{$sf} = [ $val->{$sf}, $v ]; + + push @{ $val->{subfields} }, ( $sf, 1 ) if (! $j_rs && $i_sf); + $r_sf->{$sf}++; + + } else { + $val->{$sf} = $v; + push @{ $val->{subfields} }, ( $sf, 0 ) if ($i_sf); + } } } else { $val = $l; } - push @{$rec->{$k}}, $val; + if ($j_rs) { + map { + $val->{$_} = join($j_rs, @{ $val->{$_} }); + } keys %$r_sf + } + + push @{$rec->{$f_nr}}, $val; } } @@ -518,11 +647,122 @@ return $self->{'TagName'}->{$tag} || $tag; } + +=head2 read_cnt + +Read content of C<.CNT> file and return hash containing it. + + print Dumper($isis->read_cnt); + +This function is not used by module (C<.CNT> files are not required for this +module to work), but it can be useful to examine your index (while debugging +for example). + +=cut + +sub read_cnt { + my $self = shift; + + croak "missing CNT file in ",$self->{isisdb} unless ($self->{cnt_file}); + + # Get the index information from $db.CNT + + open(my $fileCNT, $self->{cnt_file}) || croak "can't read '$self->{cnt_file}': $!"; + binmode($fileCNT); + + my $buff; + + read($fileCNT, $buff, 26) || croak "can't read first table from CNT: $!"; + $self->unpack_cnt($buff); + + read($fileCNT, $buff, 26) || croak "can't read second table from CNT: $!"; + $self->unpack_cnt($buff); + + close($fileCNT); + + return $self->{cnt}; +} + +=head2 unpack_cnt + +Unpack one of two 26 bytes fixed length record in C<.CNT> file. + +Here is definition of record: + + off key description size + 0: IDTYPE BTree type s + 2: ORDN Nodes Order s + 4: ORDF Leafs Order s + 6: N Number of Memory buffers for nodes s + 8: K Number of buffers for first level index s + 10: LIV Current number of Index Levels s + 12: POSRX Pointer to Root Record in N0x l + 16: NMAXPOS Next Available position in N0x l + 20: FMAXPOS Next available position in L0x l + 24: ABNORMAL Formal BTree normality indicator s + length: 26 bytes + +This will fill C<$self> object under C with hash. It's used by C. + +=cut + +sub unpack_cnt { + my $self = shift; + + my @flds = qw(ORDN ORDF N K LIV POSRX NMAXPOS FMAXPOS ABNORMAL); + + my $buff = shift || return; + my @arr = unpack("vvvvvvVVVv", $buff); + + print STDERR "unpack_cnt: ",join(" ",@arr),"\n" if ($self->{'debug'}); + + my $IDTYPE = shift @arr; + foreach (@flds) { + $self->{cnt}->{$IDTYPE}->{$_} = abs(shift @arr); + } +} + 1; =head1 BUGS -This module has been very lightly tested. Use with caution and report bugs. +Some parts of CDS/ISIS documentation are not detailed enough to exmplain +some variations in input databases which has been tested with this module. +When I was in doubt, I assumed that OpenIsis's implementation was right +(except for obvious bugs). + +However, every effort has been made to test this module with as much +databases (and programs that create them) as possible. + +I would be very greatful for success or failure reports about usage of this +module with databases from programs other than WinIsis and IsisMarc. I had +tested this against ouput of one C-based application, but I don't +know any details about it's version. + +=head1 VERSIONS + +As this is young module, new features are added in subsequent version. It's +a good idea to specify version when using this module like this: + + use Biblio::Isis 0.21 + +Below is list of changes in specific version of module (so you can target +older versions if you really have to): + +=over 8 + +=item 0.21 + +Added C to L and L. + +Added C to L. + +=item 0.20 + +Added C<< $isis->mfn >>, support for repeatable subfields and +C<< $isis->to_hash({ mfn => 42, ... }) >> calling convention + +=back =head1 AUTHOR