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 = 0.05;
+ $VERSION = 0.08;
@ISA = qw (Exporter);
#Give a hoot don't pollute, do not export more than needed by default
@EXPORT = qw ();
=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<read_fdt> option) files.
In this example it uses C<./cds/cds.MST> and related files.
$self->{$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 $ext (@must_exist) {
+ croak "missing ",uc($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(fileFDT, $self->{fdt_file}) || croak "can't read '$self->{fdt_file}': $!";
while (<fileFDT>) {
chomp;
# 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}': $!";
# MST format: (* = 32 bit signed)
# CTLMFN* always 0
# NXTMFB* last block allocated to master file
# NXTMFP offset to next available position in last block
# MFTYPE always 0 for user db file (1 for system)
- seek(fileMST,4,0);
+ seek($self->{'fileMST'},4,0);
my $buff;
- read(fileMST, $buff, 4);
+ read($self->{'fileMST'}, $buff, 4);
$self->{'NXTMFN'}=unpack("l",$buff) || carp "NXTNFN is zero";
# save maximum MFN
$self->{'maxmfn'} = $self->{'NXTMFN'} - 1;
- close(fileMST);
+
+
+
+ print STDERR Dumper($self),"\n" if ($self->{debug});
+
+ # open files for later
+ open($self->{'fileXRF'}, $self->{xrf_file}) || croak "can't open '$self->{xrf_file}': $!";
+
+ $self ? return $self : return undef;
+}
+
+=head2 read_cnt
+
+This function is not really used by module, but can be useful to find info
+about your index (if debugging it for example).
+
+ print Dumper($isis->read_cnt);
+
+=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(fileCNT, $self->{isisdb}.".CNT") || croak "can't read '$self->{isisdb}.CNT': $!";
+ open(fileCNT, $self->{cnt_file}) || croak "can't read '$self->{cnt_file}': $!";
# There is two 26 Bytes fixed lenght records
my $buff = shift || return;
my @arr = unpack("ssssssllls", $buff);
- print "unpack_cnt: ",join(" ",@arr),"\n" if ($self->{'debug'});
+ print STDERR "unpack_cnt: ",join(" ",@arr),"\n" if ($self->{'debug'});
my $IDTYPE = shift @arr;
foreach (@flds) {
- $self->{$IDTYPE}->{$_} = abs(shift @arr);
+ $self->{cnt}->{$IDTYPE}->{$_} = abs(shift @arr);
}
}
+ my $buff;
+
read(fileCNT, $buff, 26);
$self->unpack_cnt($buff);
read(fileCNT, $buff, 26);
$self->unpack_cnt($buff);
-
close(fileCNT);
- print Dumper($self),"\n" if ($self->{debug});
-
- # open files for later
- open($self->{'fileXRF'}, $self->{isisdb}.".XRF") || croak "can't open '$self->{isisdb}.XRF': $!";
-
- open($self->{'fileMST'}, $self->{isisdb}.".MST") || croak "can't open '$self->{isisdb}.MST': $!";
-
- $self ? return $self : return undef;
+ return $self->{cnt};
}
=head2 fetch
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 $self->{record};
+
# read XRFMFB abd XRFMFP
read($self->{'fileXRF'}, $buff, 4);
my $pointer=unpack("l",$buff) || carp "pointer is null";
+ # check for logically deleted record
+ if ($pointer < 0) {
+ print STDERR "## record $mfn is logically deleted\n" if ($self->{debug});
+ $self->{deleted} = $mfn;
+
+ return unless $self->{include_deleted};
+
+ $pointer = abs($pointer);
+ }
+
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 $blk_off = (($XRFMFB - 1) * 512) + ($XRFMFP % 512);
- 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)+$offset3;
-
- print "$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);
read($self->{'fileMST'}, $buff, 4);
my $value=unpack("l",$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 $value";
+ return;
+ }
read($self->{'fileMST'}, $buff, 14);
my ($MFRL,$MFBWB,$MFBWP,$BASE,$NVF,$STATUS) = unpack("slssss", $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});
- # delete old record
- delete $self->{record};
+ warn "MFRL $MFRL is not even number" unless ($MFRL % 2 == 0);
- if (! $self->{'include_deleted'} && $MFRL < 0) {
- print "## logically deleted record $mfn, skipping...\n" if ($self->{debug});
- return;
- }
+ warn "BASE is not 18+6*NVF" unless ($BASE == 18 + 6 * $NVF);
# Get Directory Format
read($self->{'fileMST'}, $buff, 6 * $NVF);
- my $fld_len = 0;
+ my $rec_len = 0;
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));
- print "TAG: $TAG POS: $POS LEN: $LEN\n" if ($self->{debug});
+ 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.
#
push @FieldPOS,$POS;
push @FieldLEN,$LEN;
- $fld_len += $LEN;
+ $rec_len += $LEN;
}
# Get Variable Fields
- read($self->{'fileMST'},$buff,$fld_len);
+ 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 ; $i++) {
# skip zero-sized fields
push @{$self->{record}->{$FieldTAG[$i]}}, substr($buff,$FieldPOS[$i],$FieldLEN[$i]);
}
- close(fileMST);
- print Dumper($self),"\n" if ($self->{debug});
+ $self->{'current_mfn'} = $mfn;
+
+ print STDERR Dumper($self),"\n" if ($self->{debug});
return $self->{'record'};
}
my $mfn = shift || confess "need mfn!";
# init record to include MFN as field 000
- my $rec = { '000' => $mfn };
+ my $rec = { '000' => [ $mfn ] };
my $row = $self->fetch($mfn);
my $val;
# has identifiers?
- ($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/\^/) {
projects / Biblio-Isis / blobdiff