move to wrapper 1.6.2, remove report-dir-check, initial commit of reach reporting stuff

6 files changed, 521 insertions, 0 deletions

diff --git a/reach_reports/reach_application.rb b/reach_reports/reach_application.rb
new file mode 100644
index 0000000..96b6b37
--- /dev/null
+++ b/reach_reports/reach_application.rb
@@ -0,0 +1,69 @@
+
+[ 'rubygems', 'sinatra', 'sinatra/url_for', 'active_record', 'ar-extensions', 'opentox-ruby-api-wrapper' ].each do |lib|
+  require lib
+end
+
+require 'reach_reports/reach_properties.rb'
+require 'reach_reports/reach_persistance.rb'
+require 'reach_reports/reach_service.rb'
+require 'reach_reports/reach_to_xml.rb'
+
+
+require "lib/format_util.rb"
+
+def extract_type(params)
+  halt 400, "illegal type, neither QMRF nor QPRF: "+params[:type] unless params[:type] && params[:type] =~ /(?i)Q(M|P)RF/
+  params.delete("type")
+end
+
+get '/reach_report/:type' do
+  content_type "text/uri-list"
+  type = extract_type(params)
+  LOGGER.info "list all "+type+" reports"
+  ReachReports.list_reports(type)
+end
+
+post '/reach_report/:type' do
+  content_type "text/uri-list"
+  type = extract_type(params)
+  LOGGER.info "creating "+type+" report "+params.inspect
+  ReachReports.create_report(type,params)
+end
+
+get '/reach_report/:type/:id' do
+  
+  type = extract_type(params)
+  LOGGER.info "get "+type+" report with id "+params[:id].to_s+"' "+request.env['HTTP_ACCEPT'].to_s+"'"
+  rep = ReachReports.get_report(type, params[:id])
+
+  case request.env['HTTP_ACCEPT'].to_s
+  when "application/rdf+xml"
+    owl = OpenTox::Owl.create(rep.type+"Report",rep.report_uri)
+    owl.set_data( rep.get_content.keys_to_rdf_format )
+    result = owl.rdf
+  when "application/qmrf-xml"
+    content_type "application/qmrf-xml"
+    result = ReachReports.reach_report_to_xml(rep)
+  when /application\/x-yaml|\*\/\*|^$/ # matches 'application/x-yaml', '*/*', ''
+    content_type "application/x-yaml"
+    result = rep.get_content.to_yaml
+  else
+    halt 400, "MIME type '"+request.env['HTTP_ACCEPT'].to_s+"' not supported, valid Accept-Headers are \"application/rdf+xml\", \"application/x-yaml\", \"application/qmrf-xml\"."
+  end
+  result
+end
+
+get '/reach_report/:type/:id/:section' do
+  
+  type = extract_type(params)
+  LOGGER.info "get "+type+" report section '"+params[:section].to_s+"', with id "+params[:id].to_s+"' "+request.env['HTTP_ACCEPT'].to_s+"'"
+  ReachReports.get_report(type, params[:id], params[:section]).to_yaml
+end
+
+get '/reach_report/:type/:id/:section/:subsection' do
+  
+  type = extract_type(params)
+  LOGGER.info "get "+type+" report subsection '"+params[:subsection].to_s+"', section '"+params[:section].to_s+"', with id "+params[:id].to_s+"' "+request.env['HTTP_ACCEPT'].to_s+"'"
+  ReachReports.get_report(type, params[:id], params[:section], params[:subsection]).to_yaml
+end
+
diff --git a/reach_reports/reach_persistance.rb b/reach_reports/reach_persistance.rb
new file mode 100644
index 0000000..786c48f
--- /dev/null
+++ b/reach_reports/reach_persistance.rb
@@ -0,0 +1,47 @@
+
+module ReachReports
+  
+  def self.get_uri( report )
+    raise "internal error, id not set "+to_yaml if report.id==nil
+    return $sinatra.url_for("/"+File.join(report.type,report.id.to_s), :full).to_s
+  end
+  
+  class QmrfReport < ActiveRecord::Base
+  
+    alias_attribute :date, :created_at
+    
+    QmrfProperties.serialized_properties.each do |p|
+      serialize p
+    end
+    
+    def type
+      "QMRF"
+    end
+    
+    def report_uri
+      ReachReports.get_uri(self)
+    end
+    
+    def get_content
+      hash = {}
+      [ :model_uri, :date ].each{ |p| hash[p] = self.send(p) }
+      QmrfProperties.properties.each{ |p| hash[p] = self.send(p) }
+      return hash
+    end
+  end
+  
+
+  class QprfReport < ActiveRecord::Base
+    
+    alias_attribute :date, :created_at
+
+    def report_uri
+      ReachReports.get_uri(self)
+    end
+
+    def type
+      "QPRF"
+    end
+  end
+  
+end  
\ No newline at end of file
diff --git a/reach_reports/reach_properties.rb b/reach_reports/reach_properties.rb
new file mode 100644
index 0000000..5397206
--- /dev/null
+++ b/reach_reports/reach_properties.rb
@@ -0,0 +1,158 @@
+
+module ReachReports
+  
+  class QmrfProperties
+
+    private
+    PROPS = {
+      "QSAR_title" => "1.1", "QSAR_models" => "1.2", "QSAR_software" => "1.3",
+      "QMRF_date" => "2.1", "QMRF_authors" => "2.2", "QMRF_date_revision" => "2.3", "QMRF_revision" => "2.4", 
+      "model_authors" => "2.5", "model_date" => "2.6", "references" => "2.7", "info_availability" => "2.8", 
+      "related_models" => "2.9",
+      "model_species" => "3.1", "model_endpoint" => "3.2", "endpoint_comments" => "3.3", "endpoint_units" => "3.4",
+      "endpoint_variable" => "3.5", "endpoint_protocol" => "3.6", "endpoint_data_quality" => "3.7",
+      "algorithm_type" => "4.1", "algorithm_explicit" => "4.2", "algorithms_descriptors" => "4.3", 
+      "descriptors_selection" => "4.4", "descriptors_generation" => "4.5", "descriptors_generation_software" => "4.6",
+      "descriptors_chemicals_ratio" => "4.7", 
+      "app_domain_description" => "5.1", "app_domain_method" => "5.2", "app_domain_software" => "5.3",
+      "applicability_limits" => "5.4",
+      "training_set_availability" => "6.1", "training_set_data" => "6.2","training_set_descriptors" => "6.3", 
+      "dependent_var_availability" => "6.4", "other_info" => "6.5", "preprocessing" => "6.6", "goodness_of_fit" => "6.7", 
+      "loo" => "6.8", "lmo" => "6.9", "yscrambling" => "6.10", "bootstrap" => "6.11", "other_statistics" => "6.12", 
+      "validation_set_availability" => "7.1", "validation_set_data" => "7.2", "validation_set_descriptors" => "7.3", 
+      "validation_dependent_var_availability" => "7.4", "validation_other_info" => "7.5", "experimental_design" => "7.6", 
+      "validation_predictivity" => "7.7", "validation_assessment" => "7.8", "validation_comments" => "7.9", 
+      "mechanistic_basis" => "8.1", "mechanistic_basis_comments" => "8.2", "mechanistic_basis_info" => "8.3",
+      "comments" => "9.1", "bibliography" => "9.2", "attachments" => "9.3",
+      "QMRF_number" => "10.1", "date_publication" => "10.2", "keywords" => "10.3", "summary_comments" => "10.4", 
+      }
+      
+    CHAPTER = { 
+      "QSAR_identifier" => 1, 
+      "QSAR_General_information" => 2, 
+      "QSAR_Endpoint" => 3,
+      "QSAR_Algorithm" => 4,
+      "QSAR_Applicability_domain" => 5,
+      "QSAR_Robustness" => 6,
+      "QSAR_Predictivity" => 7,
+      "QSAR_Interpretation" => 8,
+      "QSAR_Miscelaneous" => 9,
+      "QMRF_Summary" => 10 }
+    
+    CATALOGS = {
+      "software_catalog" =>
+        { :element => "software",
+          :fields => ["contact", "description", "name", "number", "url"] },
+      "authors_catalog" => {
+        :element => "author",
+        :fields => ["affiliation", "contact", "email", "name", "number", "url"]},
+      "algorithms_catalog" => {
+        :element => "algorithm",
+        :fields => ["definition", "description", "publication_ref"] },
+      "descriptors_catalog" => {
+        :element => "descriptor",
+        :fields => [ "description", "name", "publication_ref", "units"] },
+      "endpoints_catalog" => {
+        :element => "endpoint",
+        :fields => [ "group", "name", "subgroup" ]},
+      "publications_catalog" => {
+        :element => "publication",
+        :fields => [ "title", "url"]},  
+      }
+    
+    CATALOG_REFERENCES = {
+      "QSAR_software" => "software_catalog",
+      "references" => "publications_catalog",
+      "QMRF_authors" => "authors_catalog", 
+      "algorithm_explicit" => "algorithms_catalog",
+      "descriptors_generation_software" => "descriptors_catalog",
+      "model_endpoint" => "endpoints_catalog",
+      "bibliography" => "publications_catalog",
+      }
+      
+    SUBSUBSECTIONS = {
+      "algorithm_explicit" => [ "equation" ]
+    }
+        
+    #"attachment_training_data", "attachment_validation_data, "attachment_documents", 
+
+    public
+    def self.serialized_properties
+      return CATALOG_REFERENCES.keys | SUBSUBSECTIONS.keys
+    end
+
+    def self.properties
+      return PROPS.keys
+    end
+    
+    def self.chapter_number( chapter )
+      CHAPTER[chapter]
+    end
+  
+    def self.chapter_properties( chapter )
+      chapter_number = CHAPTER[chapter]
+      unsorted_chapter_props = []
+      PROPS.each do |k,v|
+        chapter_nr = v.split(".")[0].to_i
+        next unless chapter_number == chapter_nr
+        unsorted_chapter_props << [ k, v.split(".")[1].to_i ]
+      end
+      unsorted_chapter_props.sort{|a,b| a[1]<=>b[1]}.collect{ |a| a[0] } 
+    end
+    
+    def self.chapters()
+      #h = { "a" => 20, "b" => 30, "c" => 10  }
+      # h.sort {|a,b| a[1]<=>b[1]}   #=> [["c", 10], ["a", 20], ["b", 30]]
+      CHAPTER.sort{|a,b| a[1]<=>b[1]}.collect{ |a| a[0] } 
+    end
+    
+    def self.to_xml(property, content, catalogs_node)
+      node = REXML::Element.new(property)
+      node.add_attribute "chapter",PROPS[property]
+      
+      return node unless content
+      
+      unless serialized_properties.include?(property)
+        node.text = content
+      else
+        if content[:text]
+          node.text = content.text
+        end
+        if SUBSUBSECTIONS.has_key?(property)
+          SUBSUBSECTIONS[property].each do |p|
+            subsubsection_node = REXML::Element.new(p)
+            subsubsection_node.text = content[p]
+            node << subsubsection_node
+          end
+        end
+        if CATALOG_REFERENCES.has_key?(property) and content[CATALOG_REFERENCES[property]]
+          catalog_prop = CATALOG_REFERENCES[property]
+          catalog_content = content[catalog_prop]
+          catalog_attributes = CATALOGS[catalog_prop]
+          catalog_node = REXML::Element.new(catalog_prop)
+          
+          catalog_content.each do |element|
+            ref_node = REXML::Element.new(catalog_attributes[:element]+"_ref")
+            id = catalog_attributes[:element]+"_"+(catalog_content.index(element)+1).to_s
+            ref_node.add_attribute "idref",id
+            ref_node.add_attribute "catalog",catalog_prop
+            node << ref_node
+            element_node = REXML::Element.new(catalog_attributes[:element])
+            element_node.add_attribute "id",id
+            catalog_attributes[:fields].each do |k|
+              element_node.add_attribute k,element[k.to_sym]
+            end
+            catalog_node << element_node
+          end
+          catalogs_node << catalog_node
+        end
+      end
+      
+      node
+    end
+  end
+
+end
+
+
+
diff --git a/reach_reports/reach_service.rb b/reach_reports/reach_service.rb
new file mode 100644
index 0000000..cdc0f9c
--- /dev/null
+++ b/reach_reports/reach_service.rb
@@ -0,0 +1,125 @@
+module ReachReports
+  
+  def self.list_reports(type)
+    case type
+    when /(?i)QMRF/
+      ReachReports::QmrfReport.all.collect{ |r| r.report_uri }.join("\n")
+    when /(?i)QPRF/
+      ReachReports::QprfReport.all.collect{ |r| r.report_uri }.join("\n")
+    end
+  end 
+  
+  def self.create_report( type, params )
+    
+    #content_type "text/uri-list"
+    #task_uri = OpenTox::Task.as_task do |task|
+    
+    case type
+    when /(?i)QMRF/
+      if params[:model_uri]
+        report = ReachReports::QmrfReport.new :model_uri => params[:model_uri]
+        build_qmrf_report(report)
+      else
+        $sinatra.halt 400, "illegal parameters for qmrf-report, use either\n"+
+          "* model_uri\n"+ 
+          "params given: "+params.inspect
+      end
+    when /(?i)QPRF/
+      if params[:compound_uri]
+        report = ReachReports::QprfReport.new :compound_uri => params[:compound_uri]
+      else
+        $sinatra.halt 400, "illegal parameters for qprf-report, use either\n"+
+          "* compound-uri\n"+ 
+          "params given: "+params.inspect
+      end
+    end
+    
+    report.report_uri
+    
+    #end
+    #halt 202,task_uri
+  end
+  
+  def self.build_qmrf_report(r)
+
+    model = OpenTox::Model::Generic.find(r.model_uri)
+     
+    # chapter 1
+    r.QSAR_title = model.title
+    # TODO
+    # QSAR_models -> sparql same endpoint     
+    software = []
+    software << { :url => model.uri, :name => model.title, :contact => model.creator }
+    algorithm = OpenTox::Algorithm::Generic.find(model.algorithm) if model.algorithm
+    software << { :url => algorithm.uri, :name => algorithm.title }
+    r.QSAR_software = {"software_catalog" => software}
+    #chpater 2
+    r.QMRF_date = DateTime.now.to_s
+    # EMPTY: QMRF_authors, QMRF_date_revision, QMRF_revision
+    # TODO: model_authors ?
+    r.model_date = model.date.to_s
+    # TODO: references?
+    # EMPTY: info_availablity
+    # TODO: related_models = find qmrf reports for QSAR_models 
+     
+    # chapter 3
+    # TODO "model_species" ?
+    endpoints = []
+    model.predictedVariables.each do |p|
+      endpoints << { :name => p } # TODO :group, :subgroup ? 
+    end
+    r.model_endpoint = { "endpoints_catalog" => endpoints }
+    # TODO "endpoint_comments" => "3.3", "endpoint_units" => "3.4",
+    r.endpoint_variable = model.dependentVariables
+    # TODO "endpoint_protocol" => "3.6", "endpoint_data_quality" => "3.7",
+
+    # chapter 4
+    # TODO algorithm_type (='type of model')
+    # TODO algorithm_explicit.equation
+    # TODO algorithm_explicit.algorithms_catalog
+    # TODO algorithms_descriptors, descriptors_selection, descriptors_generation, descriptors_generation_software, descriptors_chemicals_ratio
+
+    # chapter 5
+    # TODO app_domain_description, app_domain_method, app_domain_software, applicability_limits
+
+    # chapter 6
+    # "training_set_availability" => "6.1", "training_set_data" => "6.2","training_set_descriptors" => "6.3", 
+    # "dependent_var_availability" => "6.4", "other_info" => "6.5", "preprocessing" => "6.6", "goodness_of_fit" => "6.7", 
+    # "loo" => "6.8", "lmo" => "6.9", "yscrambling" => "6.10", "bootstrap" => "6.11", "other_statistics" => "6.12",
+
+    # chapter 7 
+    # "validation_set_availability" => "7.1", "validation_set_data" => "7.2", "validation_set_descriptors" => "7.3", 
+    # "validation_dependent_var_availability" => "7.4", "validation_other_info" => "7.5", "experimental_design" => "7.6", 
+    # "validation_predictivity" => "7.7", "validation_assessment" => "7.8", "validation_comments" => "7.9", 
+
+    # chapter 8
+    # "mechanistic_basis" => "8.1", "mechanistic_basis_comments" => "8.2", "mechanistic_basis_info" => "8.3",
+
+    # chapter 9
+    # "comments" => "9.1", "bibliography" => "9.2", "attachments" => "9.3",
+     
+    r.save!
+  end
+  
+  def self.get_report_content(type, id, *keys)
+    
+    report_content = get_report(type, id).get_content
+    keys.each do |k|
+      $sinatra.halt 400, type+" unknown report property '#{key}'" unless report_content.is_a?(Hash) and report_content.has_key?(k)
+      report_content = report_content[k]
+    end
+    report_content    
+  end
+  
+  def self.get_report(type, id)
+    
+    case type
+    when /(?i)QMRF/
+      report = ReachReports::QmrfReport.find(id)
+    when /(?i)QPRF/
+      report = ReachReports::QprfReport.find(id)
+    end
+    $sinatra.halt 404, type+" report with id '#{id}' not found." unless report
+    return report
+  end
+end 
\ No newline at end of file
diff --git a/reach_reports/reach_test.rb b/reach_reports/reach_test.rb
new file mode 100644
index 0000000..c96340a
--- /dev/null
+++ b/reach_reports/reach_test.rb
@@ -0,0 +1,62 @@
+require "uri"
+require "yaml"
+ENV['RACK_ENV'] = 'test'
+require 'application.rb'
+require 'test/unit'
+require 'rack/test'
+require 'lib/test_util.rb'
+require 'test/test_examples.rb'
+
+LOGGER = MyLogger.new(STDOUT)
+LOGGER.datetime_format = "%Y-%m-%d %H:%M:%S "
+LOGGER.formatter = Logger::Formatter.new
+
+
+class ReachTest < Test::Unit::TestCase
+  include Rack::Test::Methods
+  include Lib::TestUtil
+
+  
+  def app
+    Sinatra::Application
+  end
+  
+  def test_it
+    #$test_case = self
+
+    post '/reach_report/qmrf',:model_uri=>"http://localhost/model/16"
+    ##post '/reach_report/qprf',:compound_uri=>"http://localhost/compound/XYZ"
+    uri = last_response.body
+    id = uri.split("/")[-1]
+    puts uri
+
+    #get '/reach_report/qmrf'
+    #puts last_response.body
+    
+    get '/reach_report/qmrf/'+id.to_s,nil,'HTTP_ACCEPT'=>"application/x-yaml"
+    puts "YAML"
+    puts last_response.body
+    
+#    get '/reach_report/qmrf/'+id.to_s,nil,'HTTP_ACCEPT'=>"application/rdf+xml"
+#    puts "RDF"
+#    puts last_response.body
+
+    get '/reach_report/qmrf/'+id,nil,'HTTP_ACCEPT' => "application/qmrf-xml"
+    puts "XML"
+    puts last_response.body
+
+    
+  end
+end
+
+
+#    query = <<EOF
+#PREFIX ot:<http://www.opentox.org/api/1.1#>
+#PREFIX rdf:<http://www.w3.org/1999/02/22-rdf-syntax-ns#>
+#select ?model  
+#where {
+#?model rdf:type ot:Model
+#}
+#EOF
+#    puts OpenTox::RestClientWrapper.post("http://apps.ideaconsult.net:8080/ontology/",{:accept => "application/rdf+xml", :query => query}) 
+#    exit
\ No newline at end of file
diff --git a/reach_reports/reach_to_xml.rb b/reach_reports/reach_to_xml.rb
new file mode 100644
index 0000000..311f59f
--- /dev/null
+++ b/reach_reports/reach_to_xml.rb
@@ -0,0 +1,60 @@
+
+module ReachReports include REXML
+  
+  def self.reach_report_to_xml( report )
+    
+    @@info = {}
+    if report.type=="QMRF"
+      @@info[:type ] = "QMRF"
+      @@info[:dtd_full ] = "http://ambit.sourceforge.net/qmrf/jws/qmrf.dtd"
+      @@info[:dtd ] = "qmrf.dtd"
+      @@info[:version ] = 1.2
+      @@info[:schema_version] = 0.9
+      @@info[:name ] = "(Q)SAR Model Reporting Format"
+    elsif report.type=="QPRF"
+      @@info[:type ] = "QPRF"
+      @@info[:dtd_full ] = "not-yet-defined/qprf.dtd"
+      @@info[:dtd ] = "qprf.dtd"
+      @@info[:version ] = 0.1
+      @@info[:schema_version] = 0.1
+      @@info[:name ] = "(Q)SAR Prediction Reporting Format"
+    end
+    @@info[:author] ="Joint Research Centre, European Commission"
+    @@info[:contact] ="Joint Research Centre, European Commission"
+    @@info[:date ] = Time.new.to_s
+    @@info[:email] ="qsardb@jrc.it"
+    @@info[:schema_version] ="0.9"
+    @@info[:url ] = "http://ecb.jrc.ec.europa.eu/qsar/"
+    
+    doc = Document.new
+    decl = XMLDecl.new
+    doc << decl
+    type = DocType.new(@@info[:type]+' PUBLIC "'+@@info[:dtd_full]+'" "'+@@info[:dtd]+'"')
+    doc << type
+    
+    main = Element.new(@@info[:type])
+    main.add_attributes( Hash[*[:version, :schema_version, :name, :author, :contact,
+      :date, :email, :schema_version, :url].collect { |v| [v.to_s, @@info[v]]}.flatten] )
+    
+    catalogs = Element.new("Catalogs")
+    chapters = Element.new(@@info[:type]+"_chapters")
+    
+    QmrfProperties.chapters.each do |c|
+      chapter = Element.new(c)
+      chapter.add_attribute("chapter",QmrfProperties.chapter_number(c))
+      QmrfProperties.chapter_properties(c).each do |p|
+        chapter << QmrfProperties.to_xml(p, report.send(p), catalogs)
+      end
+      chapters << chapter
+    end
+    
+    main << chapters
+    main << catalogs
+    doc << main
+    
+    s = ""
+    doc.write(s, 2, false, false)
+    return s
+  end
+  
+end
\ No newline at end of file