104 |
my $tag; |
my $tag; |
105 |
my $rec; |
my $rec; |
106 |
|
|
107 |
$self->{size} = 0; |
my $pos = 0; |
108 |
|
my $offset = $self->{offset} || 0; |
109 |
|
my $end_pos = 0; |
110 |
|
$end_pos = $offset + $self->{limit} if $self->{limit}; |
111 |
|
|
112 |
|
warn "# range: $offset - $end_pos"; |
113 |
|
|
114 |
while( $line = <$fh> ) { |
while( $line = <$fh> ) { |
115 |
chomp($line); |
chomp($line); |
|
|
|
116 |
my $v; |
my $v; |
117 |
|
|
118 |
if ( $line =~ /^(\S\S)\s(.+)$/ ) { |
if ( $line eq 'EF' ) { |
119 |
|
last; |
120 |
|
} elsif ( $line eq 'ER' ) { |
121 |
|
last if $end_pos && $pos >= $end_pos; |
122 |
|
$pos++; |
123 |
|
next if $offset && $pos < $offset; |
124 |
|
|
125 |
|
# join tags |
126 |
|
foreach ( qw/AB DE ID TI SO RP SC FU FX PA JI/ ) { |
127 |
|
$rec->{$_} = join(' ', @{ $rec->{$_} }) if defined $rec->{$_}; |
128 |
|
} |
129 |
|
# split on ; |
130 |
|
foreach ( qw/ID SC DE/ ) { |
131 |
|
$rec->{$_} = [ split(/;\s/, $rec->{$_}) ] if defined $rec->{$_}; |
132 |
|
} |
133 |
|
$rec->{'000'} = [ $pos ]; |
134 |
|
push @{ $self->{_rec} }, $rec; |
135 |
|
|
136 |
|
$rec = {}; |
137 |
|
$line = <$fh>; |
138 |
|
chomp $line; |
139 |
|
$log->logdie("expected blank like in ",$arg->{path}, " +$.: $line") unless ( $line eq '' ); |
140 |
|
} elsif ( $offset && $pos < $offset ) { |
141 |
|
next; |
142 |
|
} elsif ( $line =~ /^(\S\S)\s(.+)$/ ) { |
143 |
$tag = $1; |
$tag = $1; |
144 |
$v = $2; |
$v = $2; |
145 |
} elsif ( $line =~ /^\s{3}(.+)$/ ) { |
} elsif ( $line =~ /^\s{3}(.+)$/ ) { |
150 |
$doi =~ s{^\s{3}}{ } || die "can't find DOI in: $doi"; |
$doi =~ s{^\s{3}}{ } || die "can't find DOI in: $doi"; |
151 |
$v .= $doi; |
$v .= $doi; |
152 |
} |
} |
153 |
} elsif ( $line eq 'ER' ) { |
|
154 |
# join tags |
} elsif ( $line =~ m{^(\S\S)\s*$} ) { |
155 |
foreach ( qw/AB DE ID TI SO RP SC/ ) { |
warn "# $arg->{path} +$. empty |$line|\n"; |
|
$rec->{$_} = join(' ', @{ $rec->{$_} }) if defined $rec->{$_}; |
|
|
} |
|
|
$rec->{'000'} = [ ++$self->{size} ]; |
|
|
push @{ $self->{_rec} }, $rec; |
|
|
$rec = {}; |
|
|
$line = <$fh>; |
|
|
chomp $line; |
|
|
$log->logdie("expected blank like in ",$arg->{path}, " +$.: $line") unless ( $line eq '' ); |
|
|
} elsif ( $line eq 'EF' ) { |
|
|
last; |
|
156 |
} else { |
} else { |
157 |
$log->logdie("can't parse +$. $arg->{path} : $line"); |
$log->logdie("can't parse +$. $arg->{path} |$line|"); |
158 |
} |
} |
159 |
|
|
160 |
if ( defined $v ) { |
if ( defined $v ) { |
166 |
|
|
167 |
} |
} |
168 |
|
|
169 |
|
$self->{size} = $pos - $offset; |
170 |
|
|
171 |
$log->debug("loaded ", $self->size, " records"); |
$log->debug("loaded ", $self->size, " records"); |
172 |
|
|
173 |
$self ? return $self : return undef; |
$self ? return $self : return undef; |