annotations2triannon 0.1.0

data/bin/revs_annotations2csv.sh

@@ -0,0 +1,66 @@
+#!/bin/bash
+#===============================================================================
+#
+#          FILE:  revs_annotations2csv.sh
+# 
+#         USAGE:  ./revs_annotations2csv.sh 
+# 
+#   DESCRIPTION:  
+# 
+#       OPTIONS:  ---
+#  REQUIREMENTS:  ---
+#          BUGS:  ---
+#         NOTES:  ---
+#        AUTHOR:  Darren L. Weber, Ph.D. (), darren.weber@stanford.edu
+#       COMPANY:  Stanford University
+#       VERSION:  1.0
+#       CREATED:  02/04/2015 02:24:13 PM PST
+#      REVISION:  ---
+#===============================================================================
+
+source .env
+# .env should export: 
+#export REVS_SRC_USER - user with login privilege on REVS_SRC_HOST
+#export REVS_SRC_HOST - REVS_SRC_HOST should have access to REVS mysql db
+# There should be a script in $REVS_SRC_USER@$REVS_SRC_HOST:~/revs_annotations_dump.sh
+# That script should dump mysql tables into $REVS_SRC_USER@$REVS_SRC_HOST:~/revs_annotations.sql
+# These ssh/scp connections may rely on Kerberos authentication.
+ssh ${REVS_SRC_USER}@${REVS_SRC_HOST} '~/revs_annotations_dump.sh'
+scp ${REVS_SRC_USER}@${REVS_SRC_HOST}:~/revs_annotations.sql .
+
+sleep 1
+if [ ! -s revs_annotations.sql ]; then 
+    echo 'failed to dump and retrieve revs_annotations.sql'
+    exit 1
+fi
+
+# assumes the 'revs' db has been created and the current
+# shell $USER has required privileges to run SQL in revs_annotations.sql,
+# such as DROP/CREATE/LOCK table, INSERT etc.  Also note that the
+# $USER must have a MySQL password stored in ~/.mysql_pass
+pass4mysql=$(cat ~/.mysql_pass)
+mysql --user=$USER --password=${pass4mysql} revs < revs_annotations.sql
+if [ $? -ne 0 ]; then echo 'failed to import revs_annotations.sql'; exit 1; fi
+echo "Imported revs annotations into local MySQL db."
+
+out_path='/tmp/revs'
+
+mkdir -p $out_path
+if [ $? -ne 0 ]; then echo "failed to create ${out_path}"; exit 1; fi
+
+chmod a+rwx /tmp/revs
+if [ $? -ne 0 ]; then echo "failed to chmod  ${out_path}"; exit 1; fi
+
+# Assumes the $USER has FILE permission, i.e.
+# GRANT FILE ON *.* TO 'USER'@'localhost';
+mysqldump --tab="${out_path}" \
+    --fields-enclosed-by="'" \
+    --fields-terminated-by=';' \
+    --fields-escaped-by='"' \
+    --lines-terminated-by='\n' \
+    --user=$USER --password=${pass4mysql} revs annotations users
+
+if [ $? -ne 0 ]; then echo 'failed to export revs to CSV'; exit 1; fi
+echo "Exported revs annotations to CSV files at ${out_path}: "
+ls -lh /tmp/revs/
+

data/lib/annotations2triannon/annotation_list.rb

@@ -0,0 +1,37 @@
+
+module Annotations2triannon
+
+  class AnnotationList < Resource
+
+    include OpenAnnotationHarvest
+
+    attr_reader :open_annotations
+
+    def annotation_list?
+      sc_annotation_list? || iiif_annotation_list?
+    end
+
+    def iiif_annotation_list?
+      iri_type? RDF::IIIFPresentation.AnnotationList
+    end
+
+    def sc_annotation_list?
+      iri_type? RDF::SC.AnnotationList
+    end
+
+    def open_annotations
+      return @open_annotations unless @open_annotations.nil?
+      begin
+        oa_graphs = collect_open_annotations(rdf)
+        oa_graphs = oa_graphs.sample(@@config.limit_openannos) if @@config.limit_openannos > 0
+        oa_graphs.collect {|oa| Annotations2triannon::OpenAnnotation.new(oa)}
+      rescue => e
+        binding.pry if @@config.debug
+        @@config.logger.error(e.message)
+      end
+    end
+
+  end
+
+end
+

data/lib/annotations2triannon/configuration.rb

@@ -0,0 +1,52 @@
+
+module Annotations2triannon
+
+  class Configuration
+
+    attr_reader :log_file
+    attr_reader :log_path
+    attr_reader :logger
+
+    attr_accessor :debug
+    attr_accessor :limit_manifests
+    attr_accessor :limit_annolists
+    attr_accessor :limit_openannos
+
+    def initialize
+      @debug = env_boolean('DEBUG')
+
+      # In development, enable options for random sampling the data
+      @limit_manifests = ENV['ANNO_LIMIT_MANIFESTS'].to_i  # 0 disables sampling
+      @limit_annolists = ENV['ANNO_LIMIT_ANNOLISTS'].to_i  # 0 disables sampling
+      @limit_openannos = ENV['ANNO_LIMIT_OPENANNOS'].to_i  # 0 disables sampling
+
+      # logger
+      log_file = ENV['ANNO_LOG_FILE'] || 'annotations2triannon.log'
+      log_file = File.absolute_path log_file
+      @log_file = log_file
+      @log_path = File.dirname log_file
+      unless File.directory? @log_path
+        # try to create the log directory
+        Dir.mkdir @log_path rescue nil
+      end
+      begin
+        log_dev = File.new(@log_file, 'w+')
+      rescue
+        log_dev = $stderr
+        @log_file = 'STDERR'
+      end
+      log_dev.sync = true if @debug # skip IO buffering in debug mode
+      @logger = Logger.new(log_dev, 'monthly')
+      @logger.level = @debug ? Logger::DEBUG : Logger::INFO
+
+    end
+
+    def env_boolean(var)
+      # check if an ENV variable is true, use false as default
+      ENV[var].to_s.upcase == 'TRUE' rescue false
+    end
+
+  end
+
+end
+

data/lib/annotations2triannon/iiif_annotation_list.rb

@@ -0,0 +1,17 @@
+
+module Annotations2triannon
+
+  # A filter to exclude any Shared Canvas namespace content
+  class IIIFAnnotationList < AnnotationList
+
+    def annotation_list?
+      iiif_annotation_list?
+    end
+
+    def sc_annotation_list?
+      false
+    end
+
+  end
+
+end

data/lib/annotations2triannon/iiif_collection.rb

@@ -0,0 +1,56 @@
+
+module Annotations2triannon
+
+  class IIIFCollection < Resource
+
+    attr_reader :manifests
+    attr_reader :iiif_manifests
+    attr_reader :sc_manifests
+
+    def collection?
+      iri_type? RDF::IIIFPresentation.Collection
+    end
+
+    def manifests
+      return @manifests unless @manifests.nil?
+      manifests = []
+      manifests.push(* manifest_uris(query_iiif_manifests))
+      manifests.push(* manifest_uris(query_sc_manifests))
+      @manifests = manifests.collect {|m| Annotations2triannon::Manifest.new(m)}
+    end
+
+    def sc_manifests
+      return @sc_manifests unless @sc_manifests.nil?
+      @sc_manifests = manifest_uris(query_sc_manifests).collect do |s|
+        Annotations2triannon::SharedCanvasManifest.new(s.subject)
+      end
+    end
+
+    def iiif_manifests
+      return @iiif_manifests unless @iiif_manifests.nil?
+      @iiif_manifests = manifest_uris(query_iiif_manifests).collect do |s|
+        Annotations2triannon::IIIFManifest.new(s.subject)
+      end
+    end
+
+
+    private
+
+    def manifest_uris(q)
+      uris = rdf.query(q).collect {|s| s.subject }
+      uris = uris.sample(@@config.limit_manifests) if @@config.limit_manifests > 0
+      uris
+    end
+
+    def query_iiif_manifests
+      [nil, RDF.type, RDF::IIIFPresentation.Manifest]
+    end
+
+    def query_sc_manifests
+      [nil, RDF.type, RDF::SC.Manifest]
+    end
+
+  end
+
+end
+

data/lib/annotations2triannon/iiif_manifest.rb

@@ -0,0 +1,32 @@
+
+module Annotations2triannon
+
+  # A filter to exclude any Shared Canvas namespace content
+  class IIIFManifest < Manifest
+
+    def manifest?
+       iiif_manifest?
+    end
+
+    def sc_manifest?
+      false
+    end
+
+    def annotation_lists
+      return @annotation_lists unless @annotation_lists.nil?
+      uris = collect_annotation_list_uris(query_iiif_annotation_list)
+      @annotation_lists = uris.collect do |uri|
+        Annotations2triannon::AnnotationList.new(uri)
+      end
+      @annotation_lists
+    end
+
+    def sc_annotation_lists
+      return @sc_annotation_lists unless @sc_annotation_lists.nil?
+      @sc_annotation_lists = []
+    end
+
+  end
+
+end
+

data/lib/annotations2triannon/iiif_navigator.rb

@@ -0,0 +1,172 @@
+
+module Annotations2triannon
+
+  class IIIFNavigator
+
+    @@config = nil
+
+    attr_accessor :collection
+    attr_accessor :manifests
+    attr_accessor :annotation_lists
+    attr_accessor :open_annotations
+
+    attr_accessor :iiif_collection
+    attr_accessor :iiif_manifests
+    attr_accessor :iiif_annotation_lists
+    attr_accessor :iiif_open_annotations
+
+    attr_accessor :sc_manifests
+    attr_accessor :sc_annotation_lists
+    attr_accessor :sc_open_annotations
+
+
+    # @param collection_uri [URI|String] an HTTP URI for a collection
+    def initialize(collection_uri)
+      @@config ||= Annotations2triannon.configuration
+      @uri = RDF::URI.parse(collection_uri)
+      @collection = nil
+      @manifests = nil
+      @annotation_lists = nil
+      @open_annotations = nil
+      @iiif_collection = nil
+      @iiif_manifests = nil
+      @iiif_annotation_lists = nil
+      @iiif_open_annotations = nil
+      @sc_manifests = nil
+      @sc_annotation_lists = nil
+      @sc_open_annotations = nil
+    end
+
+
+    # ----
+    # Collection utilities
+
+    # @return collection - a IIIF Presentation collection
+    def collection
+      # There may be no distinction between IIIF and SC at the collection level
+      iiif_collection
+    end
+
+    # @return iiif_collection - a IIIF Presentation collection
+    def iiif_collection
+      @iiif_collection ||= Annotations2triannon::IIIFCollection.new(@uri)
+    end
+
+    # There is no RDF::SC.Collection because SC uses alternate
+    # vocabularies for this level of discovery.
+    # def sc_collection
+    #   @sc_collection ||= Annotations2triannon::SCCollection.new(@uri)
+    # end
+
+
+    # ----
+    # Manifest utilities
+
+    # @return [boolean] are there any manifests in the collection?
+    def manifests?
+      ! manifests.empty?
+    end
+    # @return manifests [Array] generic manifests, either IIIF or SC manifests
+    def manifests
+      @manifests ||= collection.manifests
+    end
+
+    # IIIF manifests (excluding SC manifests)
+    # The RDF type of these manifests is declared in the parent collection.
+    # But, watch out, the manifest itself may declare a different RDF type!
+
+    # @return [boolean] are there any IIIF manifests in the collection?
+    def iiif_manifests?
+      ! iiif_manifests.empty?
+    end
+    # @return iiif_manifests [Array] IIIF presentation manifests
+    def iiif_manifests
+      @iiif_manifests ||= collection.iiif_manifests
+    end
+
+    # SC manifests (excluding IIIF manifests)
+    # The RDF type of these manifests is declared in the parent collection.
+    # But, watch out, the manifest itself may declare a different RDF type!
+
+    # @return [boolean] are there any Shared Canvas manifests in the collection?
+    def sc_manifests?
+      ! sc_manifests.empty?
+    end
+    # @return sc_manifests [Array] Shared Canvas manifests
+    def sc_manifests
+      @sc_manifests ||= collection.sc_manifests
+    end
+
+
+    # ----
+    # Annotation List utilities
+
+    # @return annotation_lists [Array] generic annotation lists
+    def annotation_lists
+      @annotation_lists ||= collect_annotation_lists(manifests)
+    end
+
+    # @return iiif_annotation_lists [Array] IIIF Presentation annotation lists
+    def iiif_annotation_lists
+      @iiif_annotation_lists ||= collect_annotation_lists(iiif_manifests)
+    end
+
+    # @return sc_annotation_lists [Array] Shared Canvas annotation lists
+    def sc_annotation_lists
+      @sc_annotation_lists ||= collect_annotation_lists(sc_manifests)
+    end
+
+
+    # ----
+    # Open Annotation utilities
+    # Note that these open annotations are from annotation lists, not
+    # directly from manifests.
+
+    # @return open_annotations [Array] Open Annotations from Annotation Lists
+    def open_annotations
+      @open_annotations ||= collect_open_annotations(annotation_lists)
+    end
+
+    # @return iiif_open_annotations [Array] Open Annotations from a IIIF Annotation Lists
+    def iiif_open_annotations
+      @iiif_open_annotations ||= collect_open_annotations(iiif_annotation_lists)
+    end
+
+    # @return sc_open_annotations [Array] Open Annotations from a Shared Canvas Annotation Lists
+    def sc_open_annotations
+      @sc_open_annotations ||= collect_open_annotations(sc_annotation_lists)
+    end
+
+
+    # def oa2triannon
+    #
+    #   # TODO:   - post the open_annotation to triannon-app
+    #   # TODO:     - check the http status on the post
+    #   # TODO:     - log.debug on success; log.error on errors
+    #
+    # end
+
+    private
+
+    def collect_annotation_lists(manifests)
+      anno_lists = {}
+      manifests.collect {|m| anno_lists[m.iri.to_s] = m.annotation_lists }
+      anno_lists
+    end
+
+    def collect_open_annotations(annotation_lists)
+      oa = {}
+      annotation_lists.each_pair do |manifest_uri, annotations_list|
+        oa[manifest_uri] = {} if oa[manifest_uri].nil?
+        annotations_list.each do |list|
+          list_uri = list.iri.to_s
+          oa[manifest_uri][list_uri] = list.open_annotations
+        end
+      end
+      oa
+    end
+
+  end
+
+end
+

data/lib/annotations2triannon/manifest.rb

@@ -0,0 +1,86 @@
+
+module Annotations2triannon
+
+  class Manifest < Resource
+
+    include OpenAnnotationHarvest
+
+    attr_reader :annotation_lists
+    attr_reader :open_annotations
+
+    def manifest?
+       iiif_manifest? || sc_manifest?
+    end
+
+    def iiif_manifest?
+      iri_type? RDF::IIIFPresentation.Manifest
+    end
+
+    def sc_manifest?
+      iri_type? RDF::SC.Manifest
+    end
+
+    def annotation_lists
+      return @annotation_lists unless @annotation_lists.nil?
+      uris = []
+      uris.push(* collect_annotation_list_uris(query_iiif_annotation_list))
+      uris.push(* collect_annotation_list_uris(query_sc_annotation_list))
+      @annotation_lists = uris.collect do |uri|
+        Annotations2triannon::AnnotationList.new(uri)
+      end
+      @annotation_lists
+    end
+
+    def iiif_annotation_lists
+      return @iiif_annotation_lists unless @iiif_annotation_lists.nil?
+      uris = collect_annotation_list_uris(query_iiif_annotation_list)
+      @iiif_annotation_lists = uris.collect do |uri|
+        Annotations2triannon::IIIFAnnotationList.new(uri)
+      end
+      @iiif_annotation_lists
+    end
+
+    def sc_annotation_lists
+      return @sc_annotation_lists unless @sc_annotation_lists.nil?
+      uris = collect_annotation_list_uris(query_sc_annotation_list)
+      @sc_annotation_lists = uris.collect do |uri|
+        Annotations2triannon::SharedCanvasAnnotationList.new(uri)
+      end
+      @sc_annotation_lists
+    end
+
+    def open_annotations
+      return @open_annotations unless @open_annotations.nil?
+      begin
+        oa_graphs = collect_open_annotations(rdf)
+        oa_graphs = oa_graphs.sample(@@config.limit_openannos) if @@config.limit_openannos > 0
+        oa_graphs
+      rescue => e
+        binding.pry if @@config.debug
+        @@config.logger.error(e.message)
+      end
+    end
+
+
+    protected
+
+    # @return a query triple to find RDF::SC.AnnotationList
+    def query_sc_annotation_list
+      [nil, RDF.type, RDF::SC.AnnotationList]
+    end
+
+    # @return a query triple to find RDF::IIIFPresentation.AnnotationList
+    def query_iiif_annotation_list
+      [nil, RDF.type, RDF::IIIFPresentation.AnnotationList]
+    end
+
+    def collect_annotation_list_uris(q)
+      uris = rdf.query(q).collect {|s| s.subject }
+      uris = uris.sample(@@config.limit_annolists) if @@config.limit_annolists > 0
+      uris
+    end
+
+  end
+
+end
+