将ISBN文件名转换为BibTeX记录

Posted 2021-02-26

#!/usr/bin/env perl
 
use strict; 
use warnings;
 
use Encoding "utf8";
use Text::BibTeX;
use WebService::ISBNDB::API::Books;
use Getopt::Long;
use Pod::Usage;
 
my %options;
GetOptions('usage|?'  => $options{usage},
           'h|help' 	=> $options{help}
          );
pod2usage(1) if $options{usage};
pod2usage(-verbose => 2) if $options{help};
 
my $api_key = $ENV{ISBNDB_KEY} || 'TMDKWJSX';
 
my $dir = shift || '.';
my $file = shift || '&STDOUT';
my $bib = Text::BibTeX::File->new('>'.$file); 
 
 
opendir my $dh, $dir
	or die "Cannot open $dir: $!
";
my @files =  grep { -f && m{/d{9}[x|d].pdf$}i } 
		map {"$dir/$_"} 
			readdir $dh;
 
foreach my $file (@files) {
 
	# extract isbn from file name
	my $isbn = $file =~ /(d{9}[x|d]).pdf$/i ? $1 : '0000000000' ; 
	# check database for isbn number, loop if failed
	my $book = WebService::ISBNDB::API::Books->find( { api_key => $api_key, isbn => $isbn } );
  next unless $book;
 
	# set new bibtex entry
	my $entry = new Text::BibTeX::Entry;
	$entry->set_metatype(BTE_REGULAR);
	$entry->set_type('book');
	$entry->set_key($isbn);
 
	# set title field
	$entry->set( 'title', $book->get_longtitle || $book->get_title );
 
	# set author or editor field
	my $authors = $book->get_authors_text;
	# some clean-up
	$authors =~ s/^by //;
	$authors =~ s/,$//;
	$authors =~ s/,s+/ and /g;
	$authors =~ s/;s+/ and /g;
	# authors or editors ?	
	if ( $authors =~ /^s*[?edited bys+]?(.*)$/i ) {
		(my $editors = $1) =~ s/with/and/;
		$entry->set('editor', $editors);
	}
	elsif ( $authors =~ /(Editor)/i ) {
		$authors =~ s/s*(Editor)//gi;
	}
	else {
		$entry->set('author', $authors);
	}
 
	# parse publisher and edition fields for publisher and year data
	if ( $book->get_publisher_text =~ m/^(.*?),s+c?(d{4}).?$/ ) {
		$entry->set( 'publisher', $1 ) ;
		$entry->set( 'year', $2 );
 
	}
	else {
		$entry->set( 'publisher', $book->get_publisher_text ) ;
		if ( $book->get_edition_info =~ m/(d{4})/ ) {
			$entry->set( 'year', $1 );
		}		
	}
 
	# miscellaneous fields
	my $notes = $book->get_notes;
	$entry->set( 'notes', $notes ) if $notes ;
	my $abstract = $book->get_summary;
	$entry->set( 'abstract', $abstract ) if $abstract ;
 
	$entry->set( 'local-url', $file);
 
	$entry->write($bib);
 
	# sleep 2;
}
 
__END__
 
 
=head1 NAME
 
isbn2bibtex.pl - Convert ISBN file names to BibTeX records 
 
=head1 SYNOPSIS
 
isbn2bibtex.pl [-? | --help] | [directory] [outfile.bib]
 
=head1 DESCRIPTION
 
Scans a directory for PDF files whose name are ISBN-10 identifiers,
fetches the corresponding book's data from isbndb.com, parses data
fields to get rid of inconsistencies, and finally, outputs a bibtex 
file with all fields set accordingly.
 
	-?             print usage
	-h --help      verbose help message
 
If no directory is given, scans the current directory. Outputs result
to STDOUT, unless a second argument is given.
 
An API key is required to access isbndb.com services. You can either 
paste it in the source code or set the environment variable ISBNDB_KEY.
 
=head1 LICENSE
 
Free to use and modifiy, same terms as Perl itself.
 
=head1 AUTHOR
 
i-blis, I<i-blis yandex ru>. 
 
=cut