11 |
|
|
12 |
=head1 NAME |
=head1 NAME |
13 |
|
|
14 |
WebPAC::Input::Ovid - support for Ovid citation export |
WebPAC::Input::Ovid - support for Ovid citation export (BRS/Tagged and Direct Export) |
15 |
|
|
16 |
=head1 VERSION |
=head1 VERSION |
17 |
|
|
30 |
path => '/path/to/ovid-cites.txt', |
path => '/path/to/ovid-cites.txt', |
31 |
); |
); |
32 |
|
|
33 |
|
You can also specify file glob: |
34 |
|
|
35 |
|
my $input = WebPAC::Input::Ovid->new( path => '/path/to/ovid.*.txt' ); |
36 |
|
|
37 |
=head1 FUNCTIONS |
=head1 FUNCTIONS |
38 |
|
|
39 |
=head2 new |
=head2 new |
70 |
|
|
71 |
my $log = $self->_get_logger(); |
my $log = $self->_get_logger(); |
72 |
|
|
73 |
open( my $fh, '<', $arg->{path} ) || $log->logconfess("can't open $arg->{path}: $!"); |
my @paths; |
74 |
|
|
75 |
$log->info("reading '$arg->{path}'"); |
if ( $arg->{path} =~ m/\*/ ) { |
76 |
|
@paths = glob $arg->{path}; |
77 |
|
} else { |
78 |
|
@paths = ( $arg->{path} ); |
79 |
|
} |
80 |
|
|
|
my $rec; |
|
81 |
my $size = 0; |
my $size = 0; |
|
my $tag; |
|
|
|
|
82 |
$self->{_rec} = []; |
$self->{_rec} = []; |
83 |
|
|
84 |
while( my $line = <$fh> ) { |
my $max_size; |
85 |
$line =~ s{[\r\n]+$}{}; |
$max_size = ( $self->{offset} || 0 ) + $self->{limit} if $self->{limit}; |
86 |
next if $line eq ''; |
|
87 |
|
foreach my $path ( @paths ) { |
88 |
warn "<< $line\n" if $debug; |
|
89 |
|
open( my $fh, '<', $path ) || $log->logconfess("can't open $path: $!"); |
90 |
if ( $line =~ m/^<(\d+)>$/ ) { |
$log->info("reading '$path'"); |
91 |
push @{ $self->{_rec} }, $rec if $rec; |
|
92 |
warn "## rec = ",dump( $rec ),$/ if $debug; |
my $tag; |
93 |
my $expect_rec = $#{ $self->{_rec} } + 2; |
my $rec; |
94 |
warn "wrong Ovid record number: $1 != $expect_rec" if $debug && $1 != $expect_rec; |
|
95 |
$rec = { '000' => [ $1 ] }; |
while( my $line = <$fh> ) { |
96 |
} elsif ( $line =~ /^(\w.+)/ ) { |
$line =~ s{[\r\n]+$}{}; |
97 |
$tag = $1; |
next if $line eq ''; |
98 |
warn "++ $tag\n" if $debug; |
|
99 |
} elsif ( $line =~ /^\s\s(.+)/ ) { |
warn "<< $line\n" if $debug; |
100 |
my $v = $1; |
|
101 |
$v =~ s{[\s\.]+$}{}; |
if ( $line =~ m/^<(\d+)>$/ ) { |
102 |
$rec->{$tag} = [ $v ]; |
last if $max_size && $size > $max_size; |
103 |
} else { |
|
104 |
warn "### skip: '$line'\n" if $debug; |
push @{ $self->{_rec} }, $rec if $rec; |
105 |
|
warn "## rec = ",dump( $rec ),$/ if $debug; |
106 |
|
my $expect_rec = $#{ $self->{_rec} } + 2; |
107 |
|
warn "wrong Ovid record number: $1 != $expect_rec" if $debug && $1 != $expect_rec; |
108 |
|
$rec = { '000' => [ ++$size ] }; |
109 |
|
} elsif ( $line =~ /^(\w+)\s+(-\s)?(.*)/ ) { |
110 |
|
$tag = $1; |
111 |
|
warn "++ $tag\n" if $debug; |
112 |
|
$rec->{$tag} = [ $3 ] if $3; |
113 |
|
} elsif ( $line =~ /^\s+(-\s)?(.+)/ ) { |
114 |
|
push @{ $rec->{$tag} }, $2; |
115 |
|
} else { |
116 |
|
warn "### skip: '$line'\n" if $debug; |
117 |
|
} |
118 |
|
|
119 |
} |
} |
120 |
|
|
121 |
} |
# save last rec |
122 |
|
push @{ $self->{_rec} }, $rec if $rec; |
123 |
|
warn "### rec ",dump $rec if $debug; |
124 |
|
|
125 |
# save last rec |
$log->debug("loaded ", $self->size, " records from $path"); |
|
push @{ $self->{_rec} }, $rec if $rec; |
|
126 |
|
|
127 |
$log->debug("loaded ", $self->size, " records"); |
} |
128 |
|
|
129 |
$self ? return $self : return undef; |
$self ? return $self : return undef; |
130 |
} |
} |