enju_biblio 0.3.18 → 0.4.0.beta.1

data/lib/enju_biblio/porta_cql.rb

@@ -0,0 +1,282 @@
+require 'strscan'
+
+class QueryError < RuntimeError; end
+class QuerySyntaxError < QueryError; end
+
+class Cql
+  def initialize(line)
+    @logic = nil
+    @query = split_clause_text(line).collect{|txt| Clause.new(txt)}
+
+    from_day, @query = extract_with_index(@query, /from/io)
+    @from = comp_date(from_day)
+    until_day, @query = extract_with_index(@query, /until/io)
+    @until = comp_date(until_day)
+    sort_by, @query = extract_with_index(@query, /sortBy/io)
+    @sort_by = sort_by ? sort_by.terms.first : ''
+  end
+
+  attr_reader :query, :from, :until, :sort_by, :logic
+
+  def ==(other)
+    instance_variables.all? do |val|
+      instance_variable_get(val) == other.instance_variable_get(val)
+    end
+  end
+
+  def to_sunspot
+    (@query.collect{|c| c.to_sunspot} + date_range(@from, @until)).join(" #{@logic} ")
+  end
+
+  def split_clause_text(line)
+    clauses = []
+
+    s = StringScanner.new(line)
+    text = ''
+    while s.rest?
+      case
+      when s.scan(/\s+/)
+        text << s.matched
+      when s.scan(/"(?:[^"\\]|\\.)*"/)
+        text << s.matched
+      when s.scan(/(AND|OR)/i)
+        logic = s.matched.upcase
+        if @logic
+          raise QuerySyntaxError unless @logic == logic
+        else
+          @logic = logic
+        end
+        clauses << text.strip
+        text = ''
+      when s.scan(/\S*/)
+        text << s.matched
+      end
+    end
+    clauses << text.strip
+    clauses.collect{|txt| txt.gsub(/(\A\(|\)\Z)/, '')}
+  end
+
+  private
+  def extract_with_index(arr, reg)
+    element, rest = arr.partition{|c| reg =~ c.index }
+    [element.last, rest]
+  end
+
+  def date_range(from_date, until_date)
+    unless from_date == '*' and until_date == '*'
+      ["pub_date_dm:[#{from_date} TO #{until_date}]"]
+    else
+      []
+    end
+  end
+
+  def comp_date(date)
+    if date
+      text = date.terms[0]
+      date_text = case text
+      when /\A\d{4}-\d{2}-\d{2}\Z/
+        text
+      when /\A\d{4}-\d{2}\Z/
+        (text + '-01')
+      when /\A\d{4}\Z/
+        (text + '-01-01')
+      else
+        raise QuerySyntaxError, text.to_s
+      end
+      begin
+        Time.zone.parse(date_text).utc.iso8601.to_s
+      rescue
+        raise QuerySyntaxError, date.to_s
+      end
+    else
+      '*'
+    end
+  end
+end
+
+class ScannerError < QuerySyntaxError; end
+class AdapterError < QuerySyntaxError; end
+
+class Clause
+  INDEX = /(dpid|dpgroupid|title|creator|publisher|ndc|description|subject|isbn|issn|jpno|from|until|anywhere|porta_type|digitalize_type|webget_type|payment_type|ndl_agent_type|ndlc|itemno)/io
+  SORT_BY = /sortBy/io
+  RELATION = /(=|exact|\^|any|all)/io
+
+  MATCH_ALL = %w[title creator publisher]
+  MATCH_EXACT = %w[dpid dpgroupid isbn issn jpno porta_type digitalize_type webget_type payment_type ndl_agent_type itemno]
+  MATCH_PART = %w[description subject]
+  MATCH_AHEAD = %w[ndc ndlc]
+  MATCH_DATE = %w[from until]
+  MATCH_ANYWHERE = %w[anywhere]
+  LOGIC_ALL = %w[title creator publisher description subject anywhere]
+  LOGIC_ANY = %w[dpid ndl_agent_type]
+  LOGIC_EQUAL = %w[dpgroupid ndc isbn issn jpno from until porta_type digitalize_type webget_type payment_type ndlc itemno]
+  MULTIPLE = %w[dpid title creator publisher description subject anywhere ndl_agent_type]
+
+  def initialize(text)
+    unless text.empty?
+      @index, @relation, @terms = scan(text)
+      porta_adapter
+      @field = @index
+    else
+      @index = ''
+    end
+  end
+
+  attr_reader :index, :relation, :terms
+
+  def ==(other)
+    instance_variables.all? do |val|
+      instance_variable_get(val) == other.instance_variable_get(val)
+    end
+  end
+
+  def scan(text)
+    ss = StringScanner.new(text)
+    index = ''
+    relation = ''
+    terms = []
+
+    if ss.scan(INDEX) or ss.scan(SORT_BY)
+      index = ss[0]
+    end
+    # else
+    #  raise ScannerError, "index or the sortBy is requested in '#{text}'"
+    # end
+    ss.scan(/\s+/)
+    if ss.scan(RELATION)
+      relation = ss[0].upcase
+    end
+    # else
+    #  raise ScannerError, "relation is requested in '#{text}'"
+    # end
+    ss.scan(/\s+/)
+    if ss.scan(/.+/)
+      terms = ss[0].gsub(/(\A\"|\"\Z)/, '').split
+    else
+      raise ScannerError, "search term(s) is requested in '#{text}'"
+    end
+
+    [index, relation, terms]
+  end
+
+  def porta_adapter
+    logic_adapter
+    multiple_adapter
+  end
+
+  def logic_adapter
+    case
+    when LOGIC_ALL.include?(@index)
+      raise AdapterError unless %w[ALL ANY = EXACT ^].include?(@relation)
+    when LOGIC_ANY.include?(@index)
+      raise AdapterError unless %w[ANY =].include?(@relation)
+    when LOGIC_EQUAL.include?(@index)
+      raise AdapterError unless %w[=].include?(@relation)
+    end
+  end
+
+  def multiple_adapter
+    unless MULTIPLE.include?(@index)
+      raise AdapterError if @terms.size > 1
+    end
+  end
+
+  def to_sunspot
+    case
+    when MATCH_ALL.include?(@index)
+      to_sunspot_match_all
+    when MATCH_EXACT.include?(@index)
+      to_sunspot_match_exact
+    when MATCH_PART.include?(@index)
+      to_sunspot_match_part
+    when MATCH_AHEAD.include?(@index)
+      to_sunspot_match_ahead
+    when MATCH_ANYWHERE.include?(@index)
+      to_sunspot_match_anywhere
+    when @index.empty?
+      @terms.join(' ')
+    end
+  end
+
+  def to_sunspot_match_all
+    term = @terms.join(' ')
+    case @relation
+    when /\A=\Z/
+      unless /\A\^(.+)/ =~ term
+        "%s_%s:(%s)" % [@field, :text, term]
+      else
+        ahead_tarm = $1.gsub("\s", '').downcase
+        "connect_%s_%s:(%s*)" % [@field, :s, ahead_tarm]
+      end
+    when /\AEXACT\Z/
+      "%s_%s:(%s)" % [@field, :sm, term.gsub(' ', '')]
+    when /\AANY\Z/
+      "%s_%s:(%s)" % [@field, :text, multiple_to_sunspot(@terms, :any)]
+    when /\AALL\Z/
+      "%s_%s:(%s)" % [@field, :text, multiple_to_sunspot(@terms, :all)]
+    else
+      raise QuerySyntaxError
+    end
+  end
+
+  def to_sunspot_match_exact
+    case @relation
+    when /\A=\Z/
+      term = @terms.join(' ')
+      type = @field != 'issn' ? :sm : :s
+      "%s_%s:(%s)" % [@field, type, term]
+    when /\AANY\Z/
+      "%s_%s:(%s)" % [@field, :sm, multiple_to_sunspot(@terms, :any)]
+    else
+      raise QuerySyntaxError
+    end
+  end
+
+  def to_sunspot_match_part
+    case @relation
+    when /\A=\Z/
+      term = @terms.join(' ')
+      "%s_%s:(%s)" % [@field, :text, trim_ahead(term)]
+    when /\AANY\Z/
+      "%s_%s:(%s)" % [@field, :text, multiple_to_sunspot(@terms, :any)]
+    when /\AALL\Z/
+      "%s_%s:(%s)" % [@field, :text, multiple_to_sunspot(@terms, :all)]
+    else
+      raise QuerySyntaxError
+    end
+  end
+
+  def to_sunspot_match_ahead
+    "%s_%s:(%s*)" % [@field, :s, @terms.first]
+  end
+
+  def to_sunspot_match_anywhere
+    case @relation
+    when /\A=\Z/
+      term = @terms.join(' ')
+      "(%s)" % [trim_ahead(term)]
+    when /\AANY\Z/
+      "(%s)" % [multiple_to_sunspot(@terms, :any)]
+    when /\AALL\Z/
+      "(%s)" % [multiple_to_sunspot(@terms, :all)]
+    else
+      raise QuerySyntaxError
+    end
+  end
+
+
+  private
+  def multiple_to_sunspot(terms, relation)
+    boolean = relation == :any ? ' OR ' : ' AND '
+    (terms.map{|t| trim_ahead(t)}.join(boolean)).to_s
+  end
+
+  def trim_ahead(term)
+    term.sub(/\A\^+/,'')
+  end
+end
+
+if $PROGRAM_NAME == __FILE__
+  require 'porta_cql_test'
+end

data/lib/enju_biblio/sru.rb

@@ -0,0 +1,83 @@
+require 'enju_biblio/porta_cql'
+
+class QueryArgumentError < QueryError; end
+
+class Sru
+  ASC_KEYS = %w(title creator) unless Object.const_defined?(:ASC_KEYS)
+  DESC_KEYS = %w(created_at updated_at date_of_publication) unless Object.const_defined?(:DESC_KEYS)
+  SORT_KEYS = ASC_KEYS + DESC_KEYS unless Object.const_defined?(:SORY_KEYS)
+  MULTI_KEY_MAP = {'title' => 'sort_title'} unless Object.const_defined?(:MULTI_KEY_MAP)
+  def initialize(params)
+    raise QueryArgumentError, 'sru :query is required item.' unless params.key?(:query)
+
+    @cql = Cql.new(params[:query])
+    @version = params.key?(:version) ? params[:version] : '1.2'
+    @start = params.key?(:startRecord) ? params[:startRecord].to_i : 1
+    @maximum = params.key?(:maximumRecords) ? params[:maximumRecords].to_i : 200
+    @maximum = 1000 if 1000 < @maximum
+    @packing = params.key?(:recordPacking) ? params[:recordPacking] : 'string'
+    @schema = params.key?(:recordSchema) ? params[:recordSchema] : 'dc'
+    @sort_key = params[:sortKeys]
+
+    @manifestations = []
+    @extra_response_data = {}
+  end
+
+  attr_reader :version, :cql, :start, :maximum, :packing, :schema, :path, :ascending
+  attr_reader :manifestations, :extra_response_data, :number_of_records, :next_record_position
+
+  def sort_by
+    sort = {sort_by: 'created_at', order: 'desc'}
+    unless '1.1' == @version
+      @path, @ascending = @cql.sort_by.split('/')
+    else
+      @path, @ascending = @sort_key.split(',') if @sort_key
+    end
+    # TODO ソート基準が入手しやすさの場合の処理
+    if SORT_KEYS.include?(@path)
+      if MULTI_KEY_MAP.keys.include?(@path)
+        sort[:sort_by] = MULTI_KEY_MAP[@path]
+      else
+        sort[:sort_by] = @path
+      end
+      sort[:order] = 'asc' if ASC_KEYS.include?(@path)
+      case @ascending
+      when /\A(1|ascending)\Z/
+        sort[:order] = 'asc'
+      when /\A(0|descending)\Z/
+        sort[:order] = 'desc'
+      end
+    end
+    sort
+  end
+
+  def search
+    sunspot_query = @cql.to_sunspot
+    search = Sunspot.new_search(Manifestation)
+    search.build do
+      fulltext sunspot_query
+      paginate page: 1, per_page: 10000
+    end
+    @manifestations = search.execute!.results
+    @extra_response_data = get_extra_response_data
+    @number_of_records, @next_record_position = get_number_of_records
+
+    @manifestations
+  end
+
+  def get_extra_response_data
+    # TODO: NDL で必要な項目が決定し、更に enju にそのフィールドが設けられた後で正式な実装を行なう。
+    if @search.respond_to?(:erd)
+      @schema == 'dc' ? @search.erd : {}
+    end
+  end
+
+  def get_number_of_records
+    # TODO: sunspot での取得方法が分かり次第、正式な実装を行なう。
+    @schema == 'dc' ? [1405, 1406] : [40,11]
+  end
+end
+
+if $PROGRAM_NAME == __FILE__
+  require 'sru_test'
+end

data/lib/enju_biblio/version.rb

@@ -1,3 +1,3 @@
 module EnjuBiblio
-  VERSION = "0.3.18".freeze
+  VERSION = "0.4.0.beta.1"
 end

data/lib/enju_biblio.rb

@@ -1,5 +1,7 @@
 require "enju_biblio/engine"
 require "enju_biblio/openurl"
+require "enju_biblio/porta_cql"
+require "enju_biblio/sru"
 
 module EnjuBiblio
 end

data/lib/generators/enju_biblio/setup/setup_generator.rb

@@ -5,8 +5,6 @@ class EnjuBiblio::SetupGenerator < Rails::Generators::Base
   def copy_setup_files
     directory("db/fixtures", "db/fixtures/enju_biblio")
     return if file == 'fixture'
-    inject_into_class 'app/controllers/application_controller.rb', ApplicationController,
-      "  include EnjuBiblio::Controller\n"
     append_to_file("config/schedule.rb", File.open(File.expand_path('../templates', __FILE__) + '/config/schedule.rb').read)
   end
 end

data/lib/generators/enju_biblio/setup/templates/db/fixtures/agent_relationship_types.yml

@@ -3,18 +3,18 @@
 agent_relationship_type_00001:
   id: 1
   name: see_also
-  display_name: See also
+  display_name_translations: {"en": "See also"}
   note: 
   position: 1
 agent_relationship_type_00002:
   id: 2
   name: member
-  display_name: Member
+  display_name_translations: {"en": "Member"}
   note: 
   position: 2
 agent_relationship_type_00003:
   id: 3
   name: child
-  display_name: Child
+  display_name_translations: {"en": "Child"}
   note: 
   position: 3

data/lib/generators/enju_biblio/setup/templates/db/fixtures/agent_types.yml

@@ -2,15 +2,13 @@
 agent_type_00001:
   id: 1
   name: person  
-  display_name: "en: Person\r\n\
-    ja: 個人"
+  display_name_translations: {"en": "Person", "ja": "個人"}
   note: 
   position: 1
 
 agent_type_00002:
   id: 2
   name: corporate_body  
-  display_name: "en: CorporateBody\r\n\
-    ja: 団体"
+  display_name_translations: {"en": "CorporateBody", "ja": "団体"}
   note: 
   position: 2

data/lib/generators/enju_biblio/setup/templates/db/fixtures/carrier_types.yml

@@ -1,39 +1,21 @@
 --- 
 carrier_type_00001: 
-  name: volume
-  display_name: "en: Volume\r\n\
-    ja: 冊子体"
   id: 1
+  name: volume
+  display_name_translations: {"en": "Volume", "ja": "冊子体"}
   position: 1
 carrier_type_00002: 
-  name: audio_disc
-  display_name: "en: Audio disc\r\n\
-    ja: オーディオディスク"
   id: 2
+  name: audio_disc
+  display_name_translations: {"en": "Audio disc", "ja": "オーディオディスク"}
   position: 2
 carrier_type_00003: 
-  name: videodisc
-  display_name: "en: Videodisc\r\n\
-    ja: ビデオディスク"
   id: 3
+  name: videodisc
+  display_name_translations: {"en": "Videodisc", "ja": "ビデオディスク"}
   position: 3
 carrier_type_00004: 
-  name: online_resource
-  display_name: "en: Online resource\r\n\
-    ja: オンラインリソース"
   id: 4
+  name: online_resource
+  display_name_translations: {"en": "Online resource", "ja": "オンラインリソース"}
   position: 4
-
-# == Schema Information
-#
-# Table name: carrier_types
-#
-#  id           :integer          not null, primary key
-#  name         :string(255)      not null
-#  display_name :text
-#  note         :text
-#  position     :integer
-#  created_at   :datetime         not null
-#  updated_at   :datetime         not null
-#
-

data/lib/generators/enju_biblio/setup/templates/db/fixtures/content_types.yml

@@ -2,84 +2,72 @@
 content_type_00001:
   id: 1
   name: text
-  display_name: "en: text\r\n\
-    ja: 文字"
+  display_name_translations: {"en": "text", "ja": "文字"}
   note: 
   position: 1
 content_type_00002:
   id: 2
   name: performed_music
-  display_name: "en: Performed music\r\n\
-    ja: 音楽"
+  display_name_translations: {"en": "Performed music", "ja": "音楽"}
   note: 曲の演奏など
   position: 2
 content_type_00003:
   id: 3
   name: two_dimensional_moving_image
-  display_name: "en: Two-dimensional moving image\r\n\
-    ja: 動画"
+  display_name_translations: {"en": "Two-dimensional moving image", "ja": "動画"}
   note: 
   position: 3
 content_type_00004:
   id: 4
   name: tactile_text
-  display_name: "en: Tactile text\r\n\
-    ja: 点字など"
+  display_name_translations: {"en": "Tactile text", "ja": "点字など"}
   note: 触ることのできる文字
   position: 4
 content_type_00005:
   id: 5
   name: still_image
-  display_name: "en: Still image\r\n\
-    ja: 画像"
+  display_name_translations: {"en": "Still image", "ja": "画像"}
   note: 静止画像
   position: 5
 content_type_00006:
   id: 6
   name: cartographic_image
-  display_name: "en: Cartographic image\r\n\
-    ja: 地図"
+  display_name_translations: {"en": "Cartographic image", "ja": "地図"}
   note: 
   position: 6
 content_type_00007:
   id: 7
   name: spoken_word
-  display_name: "en: Spoken word\r\n\
-    ja: 音声"
+  display_name_translations: {"en": "Spoken word", "ja": "音声"}
   note: 読み上げ音声など
   position: 7
 content_type_00008:
   id: 8
   name: sounds
-  display_name: "en: Sounds\r\n\
-    ja: 音"
+  display_name_translations: {"en": "Sounds", "ja": "音"}
   note: 波音など
   position: 8
 content_type_00009:
   id: 9
   name: notated_music
-  display_name: "en: Notated music\r\n\
-    ja: 楽譜"
+  display_name_translations: {"en": "Notated music", "ja": "楽譜"}
   note: 
   position: 9
 content_type_00010:
   id: 10
   name: three_dimensional_moving_image
-  display_name: "en: Three-dimensional moving image\r\n\
-    ja: 3D動画"
+  display_name_translations: {"en": "Three-dimensional moving image", "ja": "3D動画"}
   note: 
   position: 10
 content_type_00011:
   id: 11
   name: other
-  display_name: "en: Other\r\n\
-    ja: その他"
+  display_name_translations: {"en": "Other", "ja": "その他"}
   note: LCMARCに由来
   position: 11
 content_type_00012:
   id: 12
   name: unspecified
-  display_name: "en: Unspecified\r\n\
-    ja: 不明"
+  display_name_translations: {"en": "Unspecified", "ja": "不明"}
   note: LCMARCに由来
   position: 12

data/lib/generators/enju_biblio/setup/templates/db/fixtures/create_types.yml

@@ -2,46 +2,28 @@
 
 author:
   name: author
-  display_name: "en: Author\r\n\
-    ja: 著者"
+  display_name_translations: {"en: Author", "ja": "著者"}
   note: 
   position: 1
   id: 1
 
 editor:
   name: editor
-  display_name: "en: Editor\r\n\
-    ja: 編者"
+  display_name_translations: {"en": "Editor", "ja": "編者"}
   note: 
   position: 2
   id: 2
 
 translator:
   name: translator
-  display_name: "en: Translator\r\n\
-    ja: 訳者"
+  display_name_translations: {"en": "Translator", "ja": "訳者"}
   note: 
   position: 3
   id: 3
 
 illustrator:
   name: illustrator
-  display_name: "en: Illustrator\r\n\
-    ja: 作画者"
+  display_name_translations: {"en": "Illustrator", "ja": "作画者"}
   note: 
   position: 4
   id: 4
-
-# == Schema Information
-#
-# Table name: create_types
-#
-#  id           :integer         not null, primary key
-#  name         :string(255)
-#  display_name :text
-#  note         :text
-#  position     :integer
-#  created_at   :datetime
-#  updated_at   :datetime
-#
-

data/lib/generators/enju_biblio/setup/templates/db/fixtures/form_of_works.yml

@@ -1,7 +1,7 @@
 --- 
 form_of_work_00001: 
   name: unknown  
-  display_name: unknown
+  display_name_translations: {"en": "(not specified)", "ja": "不明"}
   id: 1
   position: 1
   note: ""