highwire.pm

一个论文管理系统

use strict;
use Bibliotech::CitationSource;
use Bibliotech::CitationSource::NPG;


package Bibliotech::CitationSource::Highwire;
use base 'Bibliotech::CitationSource';
use URI;
use URI::QueryParam;
use Data::Dumper;

# Necessary for integrating 'Science' plug-in into the Highwire plug-in
use LWP;
use HTTP::Request::Common;

sub api_version {
  1;
}

sub name {
  'Highwire';
}

sub version {
  #'$Revision: 1.7 $';
  '$Revision: 1.7 $';
}

sub understands {
  my ($self, $uri, $ris_host) = @_;

  return 0 unless $uri->scheme eq 'http';

  #check the host
  return 0 unless Bibliotech::CitationSource::Highwire::HostTable::defined($uri->host);
  
  return 1 if ($uri->path =~ m!^/cgi/((content/(summary|short|extract|abstract|full))|reprint)/(${ris_host};)?.+!i);

  return 0;
}

sub citations {
  my ($self, $article_uri) = @_;

  my $ris;
  eval {
    my $ris_host = Bibliotech::CitationSource::Highwire::HostTable::getRISPrefix($article_uri->host);
    #
    # Some entries will require a login/password; structure in hash is a reference not scalar
    #
    my($ris_login);
    if(ref($ris_host)) {
      $ris_login = $self->highwire_account($ris_host->{ACCT_TYPE});
      $ris_host = $ris_host->{RIS_PREFIX};	# re-assign this as a scalar
    }

    die "do not understand URI\n" unless $self->understands($article_uri, $ris_host);

    my $file;
	$file = $article_uri->path;
	#strip fragments or queries
        $file =~ s/(?:#|\?).*//;

    die "no file name seen in URI\n" unless $file;

    #find the ID
    my $id;
    if($file =~ m!^/cgi/(?:(?:content/(?:summary|short|extract|abstract|full))|reprint)/(?:${ris_host};)?(.+)$!i)
    {
      $id = $1;
    }
    
    die "Couldn't extract Highwire ID\n" unless $id;


    my $query_uri = new URI("http://" . $article_uri->host . "/cgi/citmgr_refman?gca=" . $ris_host . ";" . $id);

    #
    # use query w/ authorization if needed
    #
    my $ris_raw;
    my $ua;
    my $response;
    if($ris_login) {
      $ua = $self->ua;
      $response = $ua->request(POST $query_uri, [ 'username' => $ris_login->{USER}, 'code' => $ris_login->{PW}]);
      if($response->is_success) {
        $ris_raw = $response->content;
      } else {
        die $response->status_line;
      }
    } else {
      $ris_raw = $self->get($query_uri);
    }

    $ris = new Bibliotech::CitationSource::NPG::RIS ($ris_raw);

    if (!$ris->has_data) {
        # give it one more try 
        sleep 2;

        if($ris_login) {
          $response = $ua->request(POST $query_uri, [ 'username' => $ris_login->{USER}, 'code' => $ris_login->{PW}]);
          if($response->is_success) {
            $ris_raw = $response->content;
          } else {
            die $response->status_line;
          }
        } else {
          $ris_raw = $self->get($query_uri);
        }

        $ris = new Bibliotech::CitationSource::NPG::RIS ($ris_raw);
    }
    die "RIS obj false\n" unless $ris;
    die "RIS file contained no data\n" unless $ris->has_data;
  };    
  die $@ if $@ =~ /at .* line \d+/;

  $self->errstr($@), return undef if $@;
  return bless [bless $ris, 'Bibliotech::CitationSource::Highwire::Result'], 'Bibliotech::CitationSource::ResultList';
}

#
# Necessary for integrating 'Science' plug-in into the Highwire plug-in
#	'Science' requires login/password to get citation data
#   Add other login/pw as needed; follow science model in bibliotech.conf
#   Then add conditional for new "acct_type"
#
sub user 
{
	my($self, $var) = @_;
    shift->cfg($var);
}
sub password 
{
	my($self, $var) = @_;
    shift->cfg($var);
}

sub highwire_account {
    my($self, $acct_type) = @_;
	
	my($login);
	if($acct_type eq 'SCIENCE') {
		$login->{USER} = $self->user('SCI_USER');
		$login->{PW} = $self->password('SCI_PASSWORD');
	}

	($login->{USER} && $login->{PW}) ? return $login : return undef;
}


package Bibliotech::CitationSource::Highwire::Result;
use base ('Bibliotech::CitationSource::NPG::RIS', 'Bibliotech::CitationSource::Result');

sub type {
  'Highwire';
}

sub source {
  'Highwire RIS file';
}

sub identifiers {
  {doi => shift->doi};
}

sub justone {
  my ($self, $field) = @_;
  my $super = 'SUPER::'.$field;
  my $stored = $self->$super or return undef;
  return ref $stored ? $stored->[0] : $stored;
}

sub authors {
  my ($self) = @_;
  my $authors = $self->SUPER::authors;
  my @authors = map(Bibliotech::CitationSource::Highwire::Result::Author->new($_), ref $authors ? @{$authors} : $authors);
  bless \@authors, 'Bibliotech::CitationSource::Result::AuthorList';
}

# override - from Nature the abbreviated name arrives in JO
sub periodical_name  { shift->collect(qw/JF/); }
sub periodical_abbr  { shift->collect(qw/JO JA J1 J2/); }

sub journal {
  my ($self) = @_;
  return Bibliotech::CitationSource::Highwire::Result::Journal->new($self->justone('journal'),
							 $self->justone('journal_abbr'),
							 $self->justone('issn'));
}

sub pubmed  { undef; }
sub doi     { shift->justone('misc3'); }
sub title   { shift->justone('title'); }
sub volume  { shift->justone('volume'); }
sub issue   { shift->justone('issue'); }
sub page    { shift->page_range; }
sub url     { shift->collect(qw/UR L3/); }

sub date {
  my $date = shift->justone('date');
  $date =~ s|^(\d+/\d+/\d+)/.*$|$1|;
  return $date;
}

sub last_modified_date {
  shift->date(@_);
}

package Bibliotech::CitationSource::Highwire::Result::Author;
use base 'Class::Accessor::Fast';

__PACKAGE__->mk_accessors(qw/firstname forename initials lastname/);

sub new {
  my ($class, $author) = @_;
  my $self = {};
  bless $self, ref $class || $class;
  my ($lastname, $firstname);
  if ($author =~ /^(.+?),\s*(.*)$/) {
    ($lastname, $firstname) = ($1, $2);
  }
  elsif ($author =~ /^(.*)\s+(.+)$/) {
    ($firstname, $lastname) = ($1, $2);
  }
  else {
    $lastname = $author;
  }
  $self->forename($firstname);
  my $initials = join(' ', map { s/^(.).*$/$1/; $_; } split(/\s+/, $firstname)) || undef;
  $firstname =~ s/(\s\w\.?)+$//;
  $self->firstname($firstname);
  $self->lastname($lastname);
  $self->initials($initials);
  return $self;
}

package Bibliotech::CitationSource::Highwire::Result::Journal;
use base 'Class::Accessor::Fast';

__PACKAGE__->mk_accessors(qw/name medline_ta issn/);

sub new {
  my ($class, $name, $medline_ta, $issn) = @_;
  my $self = {};
  bless $self, ref $class || $class;
  $self->name($name);
  $self->medline_ta($medline_ta);
  $self->issn($issn);
  return $self;
}

package Bibliotech::CitationSource::Highwire::HostTable;

%Bibliotech::CitationSoure::Highwire::HostTable::Hosts = (

  #Archives of General Psychiatry
  'archpsyc.ama-assn.org'               =>          'archpsyc',

  #Mutagenesis
  'mutage.oupjournals.org'               =>          'mutage',

  #Visual Communication
  'vcj.sagepub.com'               =>          'spvcj',

  #Social Science Computer Review
  'ssc.sagepub.com'               =>          'spssc',

  #Studies in Christian Ethics
  'sce.sagepub.com'               =>          'spsce',

  #Stem Cells
  'stemcells.alphamedpress.org'               =>          'stemcells',

  #QJM
  'qjmed.oupjournals.org'               =>          'qjmed',

  #Journal of Virology
  'jvi.asm.org'               =>          'jvi',

  #International Journal for Quality in Health Care
  'intqhc.oupjournals.org'               =>          'intqhc',

  #Environment and Behavior
  'eab.sagepub.com'               =>          'speab',

  #Gender & Society
  'gas.sagepub.com'               =>          'spgas',

  #Educational Policy
  'epx.sagepub.com'               =>          'spepx',

  #Critique of Anthropology
  'coa.sagepub.com'               =>          'spcoa',

  #Journal of Epidemiology & Community Health
  'jech.bmjjournals.com'               =>          'jech',

  #Journal of Conflict Resolution
  'jcr.sagepub.com'               =>          'spjcr',

  #The World Bank Research Observer
  'wbro.oupjournals.org'               =>          'wbro',

  #Family and Consumer Sciences Research Journal
  'fcs.sagepub.com'               =>          'spfcs',

  #Endocrine Reviews
  'edrv.endojournals.org'               =>          'edrv',

  #The EMBO Journal
  'embojournal.npgjournals.com'               =>          'emboj',

  #Journal of Macromarketing
  'jmk.sagepub.com'               =>          'spjmk',

  #Oxford Review of Economic Policy
  'oxrep.oupjournals.org'               =>          'oxrep',

  #Journal of European Social Policy
  'esp.sagepub.com'               =>          'spesp',

  #Cancer Epidemiology Biomarkers & Prevention
  'cebp.aacrjournals.org'               =>          'cebp',

  #American Journal of Clinical Nutrition
  'www.ajcn.org'               =>          'ajcn',

  #The Journal of Foraminiferal Research
  'jfr.geoscienceworld.org'               =>          'gsjfr',

  #Industrial and Corporate Change
  'icc.oupjournals.org'               =>          'indcor',

  #Genes & Development
  'www.genesdev.org'               =>          'genesdev',

  #International Journal of Lexicography
  'ijl.oupjournals.org'               =>          'lexico',

  #The International Journal of Robotics Research
  'ijr.sagepub.com'               =>          'spijr',

  #American Journal of Respiratory Cell and Molecular Biology
  'www.ajrcmb.org'               =>          'ajrcmb',

  #Journal of Holistic Nursing
  'jhn.sagepub.com'               =>          'spjhn',

  #Graft
  'gft.sagepub.com'               =>          'spgft',

  #Childhood
  'chd.sagepub.com'               =>          'spchd',

  #Journal of Consumer Culture
  'joc.sagepub.com'               =>          'spjoc',

  #The Plant Cell Online
  'www.plantcell.org'               =>          'plantcell',

  #Research on Social Work Practice
  'rsw.sagepub.com'               =>          'sprsw',

  #International Political Science Review/ Revue internationale de science politique
  'ips.sagepub.com'               =>          'spips',

  #Youth & Society
  'yas.sagepub.com'               =>          'spyas',

  #Journal of European Studies
  'jes.sagepub.com'               =>          'spjes',

  #French Studies
  'fs.oupjournals.org'               =>          'frestu',

  #Journal of Intelligent Material Systems and Structures
  'jim.sagepub.com'               =>          'spjim',

  #Qualitative Inquiry
  'qix.sagepub.com'               =>          'spqix',

  #European Journal of Endocrinology
  'www.eje-online.org'               =>          'eje',

  #Journal of Bioactive and Compatible Polymers
  'jbc.sagepub.com'               =>          'spjbc',

  #American Journal of Public Health
  'www.ajph.org'               =>          'ajph',

  #Structural Health Monitoring
  'shm.sagepub.com'               =>          'spshm',

  #International Studies
  'isq.sagepub.com'               =>          'spisq',

  #Geology
  'geology.geoscienceworld.org'               =>          'gsgeology',

  #EMBO Reports
  'emboreports.npgjournals.com'               =>          'emborep',

  #Journal of Pharmacy Practice
  'jpp.sagepub.com'               =>          'spjpp',

  #Journal of Nuclear Medicine
  'jnm.snmjournals.org'               =>          'jnumed',

  #Journal of Psychotherapy Practice and Research
  'jppr.psychiatryonline.org'               =>          'jppr',

  #Work and Occupations
  'wox.sagepub.com'               =>          'spwox',

  #International Review for the Sociology of Sport
  'irs.sagepub.com'               =>          'spirs',

  #IEICE Transactions on Communications
  'ietcom.oupjournals.org'               =>          'ietcom',

  #Forum for Modern Language Studies
  'fmls.oupjournals.org'               =>          'formod',

  #The Medieval History Journal
  'mhj.sagepub.com'               =>          'spmhj',

  #British Journal of Criminology
  'bjc.oupjournals.org'               =>          'crimin',

  #The European Journal of Orthodontics
  'ejo.oupjournals.org'               =>          'eortho',

  #Journal of Anglican Studies
  'ast.sagepub.com'               =>          'spast',

  #Homicide Studies
  'hsx.sagepub.com'               =>          'sphsx',

  #Qualitative Health Research
  'qhr.sagepub.com'               =>          'spqhr',

  #The Journal of Physiology
  'jp.physoc.org'               =>          'jphysiol',

  #Infection and Immunity
  'iai.asm.org'               =>          'iai',

  #Mineralogical Magazine
  'minmag.geoscienceworld.org'               =>          'gsminmag',

  #Advances in Developing Human Resources
  'adh.sagepub.com'               =>          'spadh',

  #Bulletin of Science, Technology & Society
  'bst.sagepub.com'               =>          'spbst',

  #Journal of the American Podiatric Medical Association
  'www.japmaonline.org'               =>          'jpodma',

  #Annals of Oncology
  'annonc.oupjournals.org'               =>          'annonc',

  #Journal of Health Psychology
  'hpq.sagepub.com'               =>          'sphpq',

  #Journal of Marketing Education
  'jmd.sagepub.com'               =>          'spjmd',

  #SIMULATION
  'sim.sagepub.com'               =>          'spsim',

  #Chest
  'www.chestjournal.org'               =>          'chest',

  #Antimicrobial Agents and Chemotherapy
  'aac.asm.org'               =>          'aac',

  #Molecular Cancer Research
  'mcr.aacrjournals.org'               =>          'molcanres',

  #Molecular and Cellular Biology
  'mcb.asm.org'               =>          'mcb',

  #International Journal of Cultural Studies
  'ics.sagepub.com'               =>          'spics',

  #Nursing Science Quarterly
  'nsq.sagepub.com'               =>          'spnsq',

  #The Journal of Applied Behavioral Science
  'jab.sagepub.com'               =>          'spjab',

  #Recent Progress in Hormone Research
  'rphr.endojournals.org'               =>          'rphr',

  #The Diabetes Educator
  'tde.sagepub.com'               =>          'sptde',

  #Palynology
  'palynology.geoscienceworld.org'               =>          'gspalynol',

  #Archives of Internal Medicine
  'archinte.ama-assn.org'               =>          'archinte',

  #Teaching Mathematics and its Applications
  'teamat.oupjournals.org'               =>          'teamat',

  #Journal of Planning History
  'jph.sagepub.com'               =>          'spjph',