umlaut 3.0.5 → 3.1.0.pre1

data/app/models/sfx4/global/kb_objects.rb

@@ -0,0 +1,34 @@
+# +--------------------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------+------+-----+------------------------+-----------------------------+
+# | Field                    | Type                                                                                                                                                      | Null | Key | Default                | Extra                       |
+# +--------------------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------+------+-----+------------------------+-----------------------------+
+# | OBJECT_ID                | bigint(20) unsigned                                                                                                                                       | NO   | PRI | NULL                   | auto_increment              |
+# | OBJECT_TYPE              | enum('JOURNAL','BOOK','DISSERTATION','PROCEEDING','CONFERENCE','REPORT','DOCUMENT','SERIES','NEWSPAPER','TRANSCRIPT','DATABASE','WIRE','CD','MANUSCRIPT') | YES  | MUL | NULL                   |                             |
+# | LANGUAGE                 | char(3)                                                                                                                                                   | NO   | MUL | eng                    |                             |
+# | PEER_REVIEWED            | enum('YES','NO')                                                                                                                                          | YES  |     | NO                     |                             |
+# | OWNER                    | varchar(100)                                                                                                                                              | NO   |     |                        |                             |
+# | AVAILABLE_FOR            | varchar(100)                                                                                                                                              | NO   |     |                        |                             |
+# | ATTR_XML                 | mediumtext                                                                                                                                                | YES  |     | NULL                   |                             |
+# | MMS_ID                   | bigint(20) unsigned                                                                                                                                       | NO   |     | NULL                   |                             |
+# | STATUS                   | enum('ACTIVE','WITHDRAWN')                                                                                                                                | YES  |     | ACTIVE                 |                             |
+# | STATUS_DATE              | timestamp                                                                                                                                                 | NO   |     | 0000-00-00 00:00:00    |                             |
+# | DISTRIBUTION_STATUS      | enum('READY FOR DISTRIBUTION','NOT FOR DISTRIBUTION')                                                                                                     | YES  |     | READY FOR DISTRIBUTION |                             |
+# | DISTRIBUTION_STATUS_DATE | timestamp                                                                                                                                                 | NO   |     | 0000-00-00 00:00:00    |                             |
+# | VERSION_NUMBER           | int(10)                                                                                                                                                   | NO   |     | 1                      |                             |
+# | RELEASE_NUMBER           | int(10)                                                                                                                                                   | NO   |     | 1                      |                             |
+# | CRUD_TYPE                | enum('CREATE','UPDATE','DELETE')                                                                                                                          | NO   |     | CREATE                 |                             |
+# | CREATION_DATE            | timestamp                                                                                                                                                 | NO   |     | 0000-00-00 00:00:00    |                             |
+# | CREATED_BY               | varchar(255)                                                                                                                                              | NO   |     |                        |                             |
+# | LAST_UPDATE_DATE         | timestamp                                                                                                                                                 | NO   |     | CURRENT_TIMESTAMP      | on update CURRENT_TIMESTAMP |
+# | LAST_UPDATED_BY          | varchar(255)                                                                                                                                              | NO   |     |                        |                             |
+# +--------------------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------+------+-----+------------------------+-----------------------------+
+module Sfx4
+  module Global
+    class KbObject < Sfx4::Global::Base
+      self.table_name = 'KB_OBJECTS'
+      self.primary_key = 'OBJECT_ID'
+
+      has_many  :az_titles,
+                :foreign_key=>'OBJECT_ID'
+    end
+  end
+end

data/app/models/sfx4/local/az_extra_info.rb

@@ -0,0 +1,7 @@
+module Sfx4
+  module Local
+    class AzExtraInfo < Sfx4::Local::Base
+      include Sfx4::Abstract::AzExtraInfo
+    end
+  end
+end

data/app/models/sfx4/local/az_letter_group.rb

@@ -0,0 +1,7 @@
+module Sfx4
+  module Local
+    class AzLetterGroup < Sfx4::Local::Base
+      include Sfx4::Abstract::AzLetterGroup
+    end
+  end
+end

data/app/models/sfx4/local/az_title.rb

@@ -0,0 +1,7 @@
+module Sfx4
+  module Local
+    class AzTitle < Sfx4::Local::Base
+      include Sfx4::Abstract::AzTitle
+    end
+  end
+end

data/app/models/sfx4/local/az_title_search.rb

@@ -0,0 +1,7 @@
+module Sfx4
+  module Local
+    class AzTitleSearch < Sfx4::Local::Base
+      include Sfx4::Abstract::AzTitleSearch
+    end
+  end
+end

data/app/models/sfx4/local/base.rb

@@ -0,0 +1,26 @@
+module Sfx4
+  module Local
+    class Base < ActiveRecord::Base
+      attr_accessible
+
+      # Was a SFX DB connection set in database.yml to connect directly to sfx?
+      def self.connection_configured?
+        config = ActiveRecord::Base.configurations["sfx_db"]
+        (not (config.nil? or config.blank? or config["adapter"].blank?))
+      end
+
+      self.establish_connection :sfx_db if self.connection_configured?
+
+      # ActiveRecord likes it when we tell it this is an abstract
+      # class only. 
+      self.abstract_class = true 
+
+      extend Sfx4::Abstract::Base
+
+      # All SFX things are read-only!
+      def readonly?() 
+        return true
+      end
+    end
+  end
+end

data/{lib → app/presentation}/section_renderer.rb

@@ -313,18 +313,6 @@ class SectionRenderer
     (! show_partial_only?) && @options[:show_heading]
   end
   
-  def render_heading
-    content_tag(:div, :class=>"section_heading") 
-    
-    output = ''
-
-      output <<= '<div class="section_heading">' 
-      (output <<= '<h3>' << CGI::escapeHTML(section_title) << '</h3>') if section_title      
-      (output <<= '<p class="section_prompt">' << CGI::escapeHTML(section_prompt) << '</p>') if section_prompt      
-      output <<= '</div>'
-      
-    output.html_safe
-  end
 
   def show_spinner?
     (! show_partial_only?) && @options[:show_spinner] &&

data/app/referent_filters/dissertation_catch.rb

@@ -0,0 +1,67 @@
+class DissertationCatch < ReferentFilter
+   include MetadataHelper
+
+   # ISSNs taken from http://spotdocs.scholarsportal.info/display/sfxdocs/SFX+and+Dissertations
+   @@da_issns = []
+   @@da_issns << "00994375" # Microfilm abstracts
+   @@da_issns << "00993123" # Dissertation abstracts
+   @@da_issns << "0420073X" # Dissertation abstracts. B, The sciences and engineering
+   @@da_issns << "00959154" # Dissertation abstracts. A, The humanities and social sciences
+   @@da_issns << "04194209" # Dissertation abstracts. A, The humanities and social sciences
+   @@da_issns << "04194217" # Dissertation abstracts. A, The humanities and social sciences
+   @@da_issns << "03076075" # Dissertation abstracts international. C, European abstracts
+   @@da_issns << "10427279" # Dissertation abstracts international. C, Worldwide
+   @@da_issns << "00255106" # Masters abstracts
+   @@da_issns << "08989095" # Masters abstracts international
+   @@da_issns << "1086962X" # Dissertation summaries in mathematics
+   @@da_issns << "10644687" # Dissertation abstracts ondisc
+   @@da_issns << "03616657" # Comprehensive dissertation index. Supplement
+   @@da_issns << "10869700" # Dissertation summaries in chemical engineering
+   @@da_issns << "10869689" # Dissertation summaries in computer sciences
+   @@da_issns << "10869697" # Dissertation summaries in electrical engineering
+   @@da_issns << "10869670" # Dissertation summaries in mechanical engineering
+   # plus one more we've been using
+   @@da_issns << "00993123" # Dissertation Abstracts
+
+
+   
+  # input: ropenurl ContextObject
+  # Is this a citation to a Dissertation Abstracts
+  # issn, or do we otherwise think it's a dissertation citation? Then change
+  # it to a dissertation citation. 
+  def filter(referent)
+    issn = get_identifier(:urn, "issn", referent)
+
+    return unless issn
+    
+    # normalize removing hyphen
+    issn.gsub!('-', '')
+    
+    if ( @@da_issns.find { |i| i == issn } )
+      # || lc($jtitle) =~ /dissertation/i || lc($jtitle2) =~ /dissertation/i)
+
+      referent.enhance_referent("genre", "dissertation")
+  
+      metadata = referent.metadata
+      # Reset it's title to the dissertation title
+      title = metadata['atitle'] || metadata['title']
+      referent.enhance_referent("btitle", title)
+      referent.enhance_referent("title", title, true, false, :overwrite => true)
+      # Now erase titles that do not apply 
+      referent.remove_value("atitle")
+      referent.remove_value("jtitle")
+      referent.remove_value("stitle")
+      # issn or isbn are wrong, probably point to Dissertation Abstracts
+      referent.remove_value("issn")
+      referent.remove_value("isbn")
+      # Same with all article level metadata
+      referent.remove_value("volume")
+      referent.remove_value("issue")
+      referent.remove_value("issue_start")
+      referent.remove_value("spage")
+      referent.remove_value("epage")
+    end
+
+  end
+  
+end

data/{lib → app}/service_adaptors/amazon.rb

@@ -1,3 +1,5 @@
+require 'aws_product_sign'
+
 #
 #   As of Aug 15 2009, Amazon API calls will require a secret key from Amazon.
 #   If you are unable to get such an account/key, this service can still be used

data/{lib → app}/service_adaptors/blacklight.rb

@@ -70,7 +70,7 @@ class Blacklight < Service
   def handle(request)
     ids_processed = []
     holdings_added = 0
-
+    
     if (@identifier_search && url = blacklight_precise_search_url(request) )
       doc = Nokogiri::XML( http_fetch(url).body )
       
@@ -98,7 +98,8 @@ class Blacklight < Service
         # filter out matches whose titles don't really match at all, or
         # which have already been seen in identifier search. 
         entries = filter_keyword_entries( doc.xpath("atom:feed/atom:entry", xml_ns) , :exclude_ids => ids_processed, :remove_subtitle => (! title_is_serial?(request.referent)) )
-
+        
+        
         marc_by_atom_id = {}
         
         # Grab the marc from our entries. Important not to do a // xpath
@@ -196,7 +197,13 @@ class Blacklight < Service
     # We need both title and author to search keyword style, or
     # we get too many false positives. Except serials we'll do
     # title only. sigh, logic tree. 
-    title = get_search_title(request.referent)
+    
+    # Also need to use appropriate 'container' title if avail, not
+    # article title. 
+    title = request.referent['jtitle']     
+    title = request.referent['btitle'] if title.blank?
+    title = request.referent['title'] if title.blank?
+
     author = get_top_level_creator(request.referent)
     return nil unless title && (author || (@bl_fields["serials_limit_clause"] && title_is_serial?(request.referent)))
     # phrase search for title, just raw dismax for author
@@ -273,24 +280,29 @@ class Blacklight < Service
   def filter_keyword_entries(atom_entries, options = {})
     options[:exclude_ids] ||= []
     options[:remove_subtitle] ||= true
+    
+    title = request.referent['jtitle']     
+    title = request.referent['btitle'] if title.blank?
+    title = request.referent['title'] if title.blank?
+    
     request_title_forms = [
-        raw_search_title(request.referent).downcase,        
-        normalize_title( raw_search_title(request.referent) )
+        title.downcase,        
+        normalize_title( title )
     ]
-    request_title_forms << normalize_title( raw_search_title(request.referent), :remove_subtitle => true) if options[:remove_subtitle]
-    request_title_forms.compact
+    request_title_forms << normalize_title( title, :remove_subtitle => true) if options[:remove_subtitle]
+    request_title_forms = request_title_forms.compact.uniq
 
     # Only keep entries with title match, and that aren't in the
     # exclude_ids list. 
     good_entries = atom_entries.find_all do |atom_entry|
-      title = atom_entry.xpath("atom:title/text()", xml_ns).to_s  
+      title = atom_entry.xpath("atom:title/text()", xml_ns).text  
    
       entry_title_forms = [
         title.downcase,
         normalize_title(title)
       ]
       entry_title_forms << normalize_title(title, :remove_subtitle=>true) if options[:remove_subtitle]
-      entry_title_forms.compact
+      entry_title_forms = entry_title_forms.compact.uniq
       
       ((entry_title_forms & request_title_forms).length > 0 &&
        (bib_ids_from_atom_entries(atom_entry) & options[:exclude_ids]).length == 0)
@@ -310,7 +322,7 @@ class Blacklight < Service
   
     return base_url + "?search_field=#{@cql_search_field}&content_format=#{format}&q=" + CGI.escape("#{@bl_fields["id"]} any \"#{ids.join(" ")}\"")
   end
-    
+
 
   def get_solr_id(rft)
     rft.identifiers.each do |id|

data/app/service_adaptors/google_scholar_link.rb

@@ -0,0 +1,71 @@
+# Generates a LINK to search google scholar for the requested article.
+# G.Scholar has no API, all we can do is link out. 
+#
+# Will only generate link for citations that appear article-level
+# and have author and title. 
+#
+# You probably want to only execute if there is no found fulltext, by configuring
+# in umlaut_services.yml with:
+#
+# preempted_by:
+#   existing_type: fulltext
+#
+# It will only be preempted by fulltext created in earlier waves of service
+# execution, service order 'priority' matters. 
+#
+# optional config: service_type, defaults to 'highlighted_link', but maybe
+#                  you want it actually under 'fulltext' or something. 
+class GoogleScholarLink < Service
+  
+  
+  def initialize(config)
+    @service_type = "highlighted_link"
+    @display_name = "Google Scholar"
+    
+    super(config)
+  end
+  
+  def service_types_generated
+    [ServiceTypeValue[@service_type]]
+  end
+  
+  
+  def handle(request)
+    return request.dispatched(self, true) unless should_link_out_to?(request)
+
+    link = "http://scholar.google.com/scholar?q=#{CGI.escape construct_query(request)}"
+    
+    request.add_service_response(
+      :service      => self,      
+      :display_text => "Look for article on Google Scholar",
+      :url          => link,
+      :service_type_value => @service_type,
+      :notes        => "This article <b>may</b> be available on the public web, look for links labelled <span class='gscholar_example'>[html]</span> or <span class='gscholar_example'>[pdf]</span>".html_safe
+    )
+    
+    return request.dispatched(self, true)
+  end
+  
+  
+  def construct_query(request)
+    metadata = request.referent.metadata
+    
+    title_query   = "allintitle: \"#{metadata["atitle"]}\""
+      
+    author_query  = (metadata["aulast"] || metadata["au"]).strip.split(/\s+|[[:punct:]]+/).
+      reject {|term| term.length <= 3}.
+      collect { |term| "author:#{term}" }.join(" ")
+    
+    "#{title_query} #{author_query}"
+  end
+  
+  def should_link_out_to?(request)
+    return false unless (! request.title_level_citation?)
+    
+    metadata = request.referent.metadata
+    
+    metadata["atitle"] && (metadata["au"] || metadata["aulast"])
+    
+  end
+  
+end

data/{lib → app}/service_adaptors/isi.rb

@@ -18,6 +18,8 @@
 # http://scientific.thomson.com/scientific/techsupport/cpe/form.html.
 # to request a change.
 #
+# File support/outage tickets at http://ip-science.thomsonreuters.com/support/
+#
 # Note, as of 13 april 09, there's a bug in ISI where journal titles including
 # ampersands cause an error. We will catch those errors and output a 'warning'
 # instead of an 'error', since it's a known problem. 
@@ -61,7 +63,16 @@ class Isi < Service
     
     
     begin
-      add_responses( request, isi_response )
+      #raise if it's an error HTTP response
+      isi_response.value
+    
+      response_xml = Nokogiri::XML(isi_response.body)    
+      # Check for errors.
+      if (error = (response_xml.at('val[@name = "error"]') || response_xml.at("error") || response_xml.at('null[@name = "error"]')))
+        raise IsiResponseException.new("ISI service reported error: #{error.inner_text}")      
+      end
+
+      add_responses( request, response_xml )
     rescue IsiResponseException => e
       # Is this the known problem with ampersands?
       # if so, output a warning, but report success not exception,
@@ -73,7 +84,14 @@ class Isi < Service
         return request.dispatched(self, true)
       else    
         # Log the error, return exception condition. 
-        Rails.logger.error("#{e.message} ; OpenURL: ?#{request.to_context_object.kev}")
+
+        error_parts = []
+        error_parts << "ISI URI: #{@api_url}"
+        error_parts << "Request XML: #{xml}"
+        error_parts << "ISI Response body: #{response_xml}"
+
+
+        Rails.logger.error("#{e.message} ; " + error_parts.join("\n     ") )
         return request.dispatched(self, false, e)
       end
     end
@@ -193,17 +211,8 @@ class Isi < Service
     return http.post(uri.request_uri, xml, headers)
   end
 
-  def add_responses(request, isi_response)
-    #raise if it's an error HTTP response
-    isi_response.value
-    
-    response_xml = Nokogiri::XML(isi_response.body)
-    
-    # Check for errors.
-    if (error = (response_xml.at('val[@name = "error"]') || response_xml.at("error") || response_xml.at('null[@name = "error"]')))
-     raise IsiResponseException.new("ISI service reported error: #{error.inner_text}")      
-    end
-    
+  def add_responses(request, response_xml)
+
     results = response_xml.at('map[@name ="cite_id"] map[@name="WOS"]')
     unless (results)
       error_message = "#{self.id}: "