--- trunk/IsisDB.pm 2004/12/29 15:10:34 7 +++ trunk/lib/Biblio/Isis.pm 2005/01/07 20:57:56 37 @@ -1,13 +1,15 @@ -package IsisDB; +package Biblio::Isis; use strict; use Carp; +use File::Glob qw(:globally :nocase); + use Data::Dumper; BEGIN { use Exporter (); use vars qw ($VERSION @ISA @EXPORT @EXPORT_OK %EXPORT_TAGS); - $VERSION = 0.02; + $VERSION = 0.11; @ISA = qw (Exporter); #Give a hoot don't pollute, do not export more than needed by default @EXPORT = qw (); @@ -18,19 +20,44 @@ =head1 NAME -IsisDB - Read CDS/ISIS database +Biblio::Isis - Read CDS/ISIS, WinISIS and IsisMarc database =head1 SYNOPSIS - use IsisDB - my $isis = new IsisDB( + use Biblio::Isis; + + my $isis = new Biblio::Isis( isisdb => './cds/cds', ); + for(my $mfn = 1; $mfn <= $isis->count; $mfn++) { + print $isis->to_ascii($mfn),"\n"; + } + =head1 DESCRIPTION -This module will read CDS/ISIS databases and create hash values out of it. -It can be used as perl-only alternative to OpenIsis module. +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 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 +values (like C<^asomething^belse>). + +Unique feature of this module is ability to C records. +It will also skip zero sized fields (OpenIsis has a bug in XS bindings, so +fields which are zero sized will be filled with random junk from memory). + +It also has support for identifiers (only if ISIS database is created by +IsisMarc), see 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 +OpenIsis, so reduced time in other parts of the code should compensate for +slower performance of this module (speed of reading ISIS database is +rarely an issue). =head1 METHODS @@ -50,21 +77,18 @@ # some binary reads # -sub Read32 { - my $self = shift; - - my $f = shift || die "Read32 needs file handle"; - read($$f,$b,4) || die "can't read 4 bytes from $$f from position ".tell($f); - return unpack("l",$b); -} - =head2 new -Open CDS/ISIS database +Open ISIS database - my $isis = new IsisDB( + my $isis = new Biblio::Isis( isisdb => './cds/cds', read_fdt => 1, + include_deleted => 1, + hash_filter => sub { + my $v = shift; + $v =~ s#foo#bar#g; + }, debug => 1, ); @@ -74,22 +98,31 @@ =item isisdb -Prefix path to CDS/ISIS. It should contain full or relative path to database -and common prefix of C<.FDT>, C<.MST>, C<.CNT>, C<.XRF> and C<.MST> files. +This is full or relative path to ISIS database files which include +common prefix of C<.MST>, and C<.XRF> and optionally C<.FDT> (if using +C option) files. + +In this example it uses C<./cds/cds.MST> and related files. =item read_fdt Boolean flag to specify if field definition table should be read. It's off by default. +=item include_deleted + +Don't skip logically deleted records in ISIS. + +=item hash_filter + +Filter code ref which will be used before data is converted to hash. + =item debug -Dump a C of debugging output. +Dump a B of debugging output. =back -It will also set C<$isis-E{'maxmfn'}> which is maximum MFN stored in database. - =cut sub new { @@ -97,19 +130,38 @@ my $self = {}; bless($self, $class); - $self->{isisdb} = {@_}->{isisdb} || croak "new needs database name as argument!"; + croak "new needs database name (isisdb) as argument!" unless ({@_}->{isisdb}); + + foreach my $v (qw{isisdb debug include_deleted hash_filter}) { + $self->{$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}); - $self->{debug} = {@_}->{debug}; + foreach my $ext (@must_exist) { + croak "missing ",uc($ext)," file in ",$self->{isisdb} unless ($self->{$ext."_file"}); + } + + print STDERR "## using files: ",join(" ",@isis_files),"\n" if ($self->{debug}); # 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); @@ -126,12 +178,13 @@ } } - 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 @@ -139,64 +192,33 @@ # 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); - $self->{'NXTMFN'}=$self->Read32(\*fileMST) || 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': $!"; + seek($self->{'fileMST'},4,0) || croak "can't seek to offset 0 in MST: $!"; - # There is two 26 Bytes fixed lenght records + my $buff; - # 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); - - my $IDTYPE = shift @arr; - foreach (@flds) { - $self->{$IDTYPE}->{$_} = abs(shift @arr); - } - } + read($self->{'fileMST'}, $buff, 4) || croak "can't read NXTMFN from MST: $!"; + $self->{'NXTMFN'}=unpack("V",$buff) || croak "NXTNFN is zero"; - my $buff; - read(fileCNT, $buff, 26); - $self->unpack_cnt($buff); + print STDERR Dumper($self),"\n" if ($self->{debug}); - 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 $self : return undef; +} - close(fileCNT); +=head2 count - print Dumper($self) 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 @@ -206,7 +228,12 @@ my $rec = $isis->fetch(55); Returns hash with keys which are field names and values are unpacked values -for that field. +for that field like this: + + $rec = { + '210' => [ '^aNew York^cNew York University press^dcop. 1988' ], + '990' => [ '2140', '88', 'HAY' ], + }; =cut @@ -215,60 +242,77 @@ my $mfn = shift || croak "fetch needs MFN as argument!"; - print "fetch: $mfn\n" if ($self->{debug}); + # is mfn allready in memory? + my $old_mfn = $self->{'current_mfn'} || -1; + 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 "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 - my $pointer=$self->Read32(\*{$self->{'fileXRF'}}); + read($self->{'fileXRF'}, $buff, 4); + my $pointer=unpack("V",$buff) || croak "pointer is null"; + + # 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); - print "XRFMFB: $XRFMFB XRFMFP: $XRFMFP\n" if ($self->{debug}); + # (XRFMFB - 1) * 512 + XRFMFP + # why do i have to do XRFMFP % 1024 ? - # XXX fix this to be more readable!! - # e.g. (XRFMFB - 1) * 512 + XRFMFP - - my $offset = $pointer; - my $offset2=int($offset/2048)-1; - my $offset22=int($offset/4096); - my $offset3=$offset-($offset22*4096); - if ($offset3>512) { - $offset3=$offset3-2048; - } - my $offset4=($offset2*512)+$offset3; + my $blk_off = (($XRFMFB - 1) * 512) + ($XRFMFP % 512); - print "$offset - $offset2 - $offset3 - $offset4\n" if ($self->{debug}); + print STDERR "## pointer: $pointer XRFMFB: $XRFMFB XRFMFP: $XRFMFP offset: $blk_off\n" if ($self->{'debug'}); # Get Record Information - seek($self->{'fileMST'},$offset4,0); + seek($self->{'fileMST'},$blk_off,0) || croak "can't seek to $blk_off: $!"; - my $value=$self->Read32(\*{$self->{'fileMST'}}); + read($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) { -print ("Error: The MFN:".$mfn." is not found in MST(".$value.")"); - return -1; # 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; + } - my $buff; read($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 "MFRL: $MFRL MFBWB: $MFBWB MFBWP: $MFBWP BASE: $BASE NVF: $NVF STATUS: $STATUS\n" if ($self->{debug}); + print STDERR "## MFRL: $MFRL MFBWB: $MFBWB MFBWP: $MFBWP BASE: $BASE NVF: $NVF STATUS: $STATUS\n" if ($self->{debug}); + + warn "MFRL $MFRL is not even number" unless ($MFRL % 2 == 0); + + warn "BASE is not 18+6*NVF" unless ($BASE == 18 + 6 * $NVF); # Get Directory Format @@ -276,16 +320,15 @@ my @FieldLEN; my @FieldTAG; - for (my $i = 0 ; $i < $NVF ; $i++) { + read($self->{'fileMST'}, $buff, 6 * $NVF); -# $TAG=$self->Read16($fileMST); -# $POS=$self->Read16($fileMST); -# $LEN=$self->Read16($fileMST); + my $rec_len = 0; - read($self->{'fileMST'}, $buff, 6); - my ($TAG,$POS,$LEN) = unpack("sss", $buff); + for (my $i = 0 ; $i < $NVF ; $i++) { - print "TAG: $TAG POS: $POS LEN: $LEN\n" if ($self->{debug}); + my ($TAG,$POS,$LEN) = unpack("vvv", substr($buff,$i * 6, 6)); + + print STDERR "## TAG: $TAG POS: $POS LEN: $LEN\n" if ($self->{debug}); # The TAG does not exists in .FDT so we set it to 0. # @@ -300,34 +343,45 @@ push @FieldTAG,$TAG; push @FieldPOS,$POS; push @FieldLEN,$LEN; + + $rec_len += $LEN; } # Get Variable Fields - delete $self->{record}; + read($self->{'fileMST'},$buff,$rec_len); + + print STDERR "## rec_len: $rec_len poc: ",tell($self->{'fileMST'})."\n" if ($self->{debug}); for (my $i = 0 ; $i < $NVF ; $i++) { - my $rec; - read($self->{'fileMST'},$rec,$FieldLEN[$i]); - push @{$self->{record}->{$FieldTAG[$i]}}, $rec; - } - close(fileMST); + # skip zero-sized fields + next if ($FieldLEN[$i] == 0); - # The record is marked for deletion - if ($STATUS==1) { - return -1; + push @{$self->{record}->{$FieldTAG[$i]}}, substr($buff,$FieldPOS[$i],$FieldLEN[$i]); } - print Dumper($self) if ($self->{debug}); + $self->{'current_mfn'} = $mfn; + + print STDERR Dumper($self),"\n" if ($self->{debug}); return $self->{'record'}; } =head2 to_ascii -Dump ascii output of selected MFN +Returns ASCII output of record with specified MFN + + print $isis->to_ascii(42); + +This outputs something like this: - print $isis->to_ascii(55); + 210 ^aNew York^cNew York University press^dcop. 1988 + 990 2140 + 990 88 + 990 HAY + +If C is specified when calling C it will display field names +from C<.FDT> file instead of numeric tags. =cut @@ -341,7 +395,8 @@ my $out = "0\t$mfn"; foreach my $f (sort keys %{$rec}) { - $out .= "\n$f\t".join("\n$f\t",@{$self->{record}->{$f}}); + my $fn = $self->tag_name($f); + $out .= "\n$fn\t".join("\n$fn\t",@{$self->{record}->{$f}}); } $out .= "\n"; @@ -349,30 +404,195 @@ return $out; } -# -# XXX porting from php left-over: -# -# do I *REALLY* need those methods, or should I use -# $self->{something} directly? -# -# Probably direct usage is better! -# +=head2 to_hash + +Read record with specified MFN and convert it to hash + + my $hash = $isis->to_hash($mfn); + +It has ability to convert characters (using C) from ISIS +database before creating structures enabling character re-mapping or quick +fix-up of data. + +This function returns hash which is like this: + + $hash = { + '210' => [ + { + 'c' => 'New York University press', + 'a' => 'New York', + 'd' => 'cop. 1988' + } + ], + '990' => [ + '2140', + '88', + 'HAY' + ], + }; + +You can later use that hash to produce any output from ISIS data. + +If database is created using IsisMarc, it will also have to special fields +which will be used for identifiers, C and C like this: + + '200' => [ + { + 'i1' => '1', + 'i2' => ' ' + 'a' => 'Goa', + 'f' => 'Valdo D\'Arienzo', + 'e' => 'tipografie e tipografi nel XVI secolo', + } + ], + +This method will also create additional field C<000> with MFN. -sub TagName { +=cut + +sub to_hash { my $self = shift; - return $self->{TagName}; + + my $mfn = shift || confess "need mfn!"; + + # init record to include MFN as field 000 + my $rec = { '000' => [ $mfn ] }; + + my $row = $self->fetch($mfn); + + foreach my $k (keys %{$row}) { + foreach my $l (@{$row->{$k}}) { + + # filter output + $l = $self->{'hash_filter'}->($l) if ($self->{'hash_filter'}); + + my $val; + + # has identifiers? + ($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); + } + } else { + $val = $l; + } + + push @{$rec->{$k}}, $val; + } + } + + return $rec; } -sub NextMFN { +=head2 tag_name + +Return name of selected tag + + print $isis->tag_name('200'); + +=cut + +sub tag_name { my $self = shift; - return $self->{NXTMFN}; + my $tag = shift || return; + 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 AUTHOR @@ -381,8 +601,8 @@ dpavlin@rot13.org http://www.rot13.org/~dpavlin/ -This module is based heavily on code from LIBISIS.PHP - Library to read ISIS files V0.1.1 -written in php and (c) 2000 Franck Martin - released under LGPL. +This module is based heavily on code from C library to read ISIS files V0.1.1 +written in php and (c) 2000 Franck Martin and released under LGPL. =head1 COPYRIGHT @@ -395,5 +615,7 @@ =head1 SEE ALSO -L, perl(1). +OpenIsis web site L + +perl4lib site L