Downgraded to version 0.9.0 code

Changed attribute name from 'bioproject_accession' to 'bioproject_id' Deleted rule of 'at least one required'(rule:36)
ddbj · Aug 16, 2017 · 2ba68f9 · 2ba68f9
1 parent 42470c8
commit 2ba68f9
Show file tree

Hide file tree

Showing 41 changed files with 84 additions and 217 deletions.
diff --git a/src/lib/validator/biosample_validator.rb b/src/lib/validator/biosample_validator.rb
@@ -174,10 +174,8 @@ def validate (data_xml, submitter_id=nil)
       ### 属性名や必須項目に関する検証
       # パッケージから属性情報(必須項目やグループ)を取得
       attr_list = get_attributes_of_package(biosample_data["package"])
-      attr_group_list = get_attribute_groups_of_package(biosample_data["package"])
       not_predefined_attribute_name("14", sample_name, biosample_data["attributes"], attr_list , line_num)
       missing_mandatory_attribute("27", sample_name, biosample_data["attributes"], attr_list , line_num)
-      missing_group_of_at_least_one_required_attributes("36", sample_name, biosample_data["attributes"], attr_group_list, line_num)
       missing_required_attribute_name("92", sample_name, biosample_data["attributes"], attr_list , line_num)
 
       ### 全属性値を対象とした検証
@@ -193,9 +191,9 @@ def validate (data_xml, submitter_id=nil)
           biosample_data["attributes"][attr_name] = value = CommonUtils::get_auto_annotation(@error_list.last)
         end
         attribute_value_is_not_integer("93", sample_name, attr_name.to_s, value, @conf[:int_attr], line_num)
-        ret = bioproject_submission_id_replacement("95", sample_name, biosample_data["attributes"]["bioproject_accession"], line_num)
+        ret = bioproject_submission_id_replacement("95", sample_name, biosample_data["attributes"]["bioproject_id"], line_num)
         if ret == false && !CommonUtils::get_auto_annotation(@error_list.last).nil? #save auto annotation value
-          biosample_data["attributes"]["bioproject_accession"] = value = CommonUtils::get_auto_annotation(@error_list.last)
+          biosample_data["attributes"]["bioproject_id"] = value = CommonUtils::get_auto_annotation(@error_list.last)
         end
       end
 
@@ -224,9 +222,9 @@ def validate (data_xml, submitter_id=nil)
       end
 
       ### 特定の属性値に対する検証
-      invalid_bioproject_accession("5", sample_name, biosample_data["attributes"]["bioproject_accession"], line_num)
-      bioproject_not_found("6", sample_name,  biosample_data["attributes"]["bioproject_accession"], @submitter_id, line_num)
-      invalid_bioproject_type("70", sample_name, biosample_data["attributes"]["bioproject_accession"], line_num)
+      invalid_bioproject_accession("5", sample_name, biosample_data["attributes"]["bioproject_id"], line_num)
+      bioproject_not_found("6", sample_name,  biosample_data["attributes"]["bioproject_id"], @submitter_id, line_num)
+      invalid_bioproject_type("70", sample_name, biosample_data["attributes"]["bioproject_id"], line_num)
       duplicated_locus_tag_prefix("91", sample_name, biosample_data["attributes"]["locus_tag_prefix"], @biosample_list, @submission_id, line_num)
       ret = format_of_geo_loc_name_is_invalid("94", sample_name, biosample_data["attributes"]["geo_loc_name"], line_num)
       if ret == false && !CommonUtils::get_auto_annotation(@error_list.last).nil? #save auto annotation value
@@ -294,49 +292,6 @@ def get_attributes_of_package (package_name)
     end
   end
 
-  #
-  # 指定されたpackageの属性グループのリストを取得して返す
-  # 属性グループがないpackageの場合には空のリストを返す
-  #
-  # ==== Args
-  # package name ex."Microbe"
-  #
-  # ==== Return
-  # An hash of the attributes.
-  # [
-  #   {
-  #     :group_name => "Source group attribute in Microbe",
-  #     :attribute_set => ["host", "isolation_source"]
-  #   },
-  #   {...}, ...
-  # ]
-  def get_attribute_groups_of_package (package_name)
-
-    #あればキャッシュを使用
-    if @cache.nil? || @cache.check(ValidatorCache::PACKAGE_ATTRIBUTE_GROUPS, package_name).nil?
-      sparql = SPARQLBase.new(@conf[:sparql_config]["master_endpoint"], @conf[:sparql_config]["slave_endpoint"])
-      params = {package_name: package_name}
-      template_dir = File.absolute_path(File.dirname(__FILE__) + "/sparql")
-      params[:version] = @conf[:version]["biosample_graph"]
-      sparql_query = CommonUtils::binding_template_with_hash("#{template_dir}/attribute_groups_of_package.rq", params)
-      result = sparql.query(sparql_query)
-      attr_group_list = []
-      result.group_by {|row| row[:group_name] }.each do |group, item|
-        attribute_set = []
-        item.each do |row|
-          attribute_set.push(row[:attribute_name])
-        end
-        attr_group_list.push({group_name: group, attribute_set: attribute_set})
-      end
-      @cache.save(ValidatorCache::PACKAGE_ATTRIBUTE_GROUPS, package_name, attr_group_list) unless @cache.nil?
-      attr_group_list
-    else
-      puts "use cache in get_attributes_of_package" if $DEBUG
-      attr_group_list = @cache.check(ValidatorCache::PACKAGE_ATTRIBUTE_GROUPS, package_name)
-      attr_group_list
-    end
-  end
-
 ### validate method ###
 
 ### validate method ###
@@ -625,45 +580,6 @@ def missing_mandatory_attribute (rule_code, sample_name, sample_attr, package_at
     end
   end
 
-  #
-  # 複数のうちいずれかの入力が必須である属性の記載がないものを検証
-  #
-  # ==== Args
-  # rule_code
-  # sample_attr ユーザ入力の属性リスト
-  # package_attr_group_list パッケージに対する属性グループリスト
-  # line_num
-  # ==== Return
-  # true/false
-  #
-  def missing_group_of_at_least_one_required_attributes (rule_code, sample_name, sample_attr, package_attr_group_list , line_num)
-    return nil if sample_attr.nil? || package_attr_group_list.nil?
-
-    ng_group_list = []
-    package_attr_group_list.each do |attr_group|
-      count = 0
-      attr_group[:attribute_set].each do |require_attribute|
-        if !sample_attr[require_attribute].nil? && !CommonUtils::blank?(sample_attr[require_attribute])
-          count = count + 1
-        end
-      end
-      if count == 0 # グループの属性のうち一つも値がない
-        ng_group_list.push("[ #{attr_group[:attribute_set].join(", ")} ]") #配列風の文字列を生成 "[ attr1, attr2 ]"
-      end
-    end
-
-    if ng_group_list.size <= 0
-      true
-    else #ルールに反した属性グループがひとつでもあればNG
-      annotation = [
-        {key: "Sample name", value: sample_name},
-        {key: "Attribute groups", value: ng_group_list.join(", ")}
-      ]
-      error_hash = CommonUtils::error_obj(@validation_config["rule" + rule_code], @data_file, annotation)
-      @error_list.push(error_hash)
-      false
-    end
-  end
   #
   # 必須属性名の記載がないものを検証
   #
@@ -849,7 +765,7 @@ def invalid_bioproject_accession (rule_code, sample_name, bioproject_accession,
     if result == false
       annotation = [
           {key: "Sample name", value: sample_name},
-          {key: "Attribute", value: "bioproject_accession"},
+          {key: "Attribute", value: "bioproject_id"},
           {key: "Attribute value", value: bioproject_accession}
       ]
       error_hash = CommonUtils::error_obj(@validation_config["rule" + rule_code], @data_file, annotation)
@@ -1799,7 +1715,7 @@ def bioproject_not_found (rule_code, sample_name, bioproject_accession, submitte
       annotation = [
         {key: "Sample name", value: sample_name},
         {key: "Submitter ID", value: submitter_id},
-        {key: "bioproject_acession", value: bioproject_accession}
+        {key: "bioproject_id", value: bioproject_accession}
       ]
       error_hash = CommonUtils::error_obj(@validation_config["rule" + rule_code], @data_file, annotation)
       @error_list.push(error_hash)
@@ -1808,7 +1724,7 @@ def bioproject_not_found (rule_code, sample_name, bioproject_accession, submitte
   end
 
   #
-  # Submisison に含まれる複数の BioSample 間で sample name, title, bioproject accession, description 以外で
+  # Submisison に含まれる複数の BioSample 間で sample name, title, bioproject id, description 以外で
   # ユニークな属性を持っている他のサンプルがないかの検証
   # 全サンプルを対象に検証し、同値であるサンプルには同じグループIDを降り、どのサンプルのセットが重複しているかを示す
   #
@@ -1824,7 +1740,7 @@ def identical_attributes (rule_code, biosample_list)
 
     result = true
     # 同値比較しない基本属性
-    keys_excluding = ["sample_name", "sample_title", "bioproject_accession", "description"]
+    keys_excluding = ["sample_name", "sample_title", "bioproject_id", "description"]
 
     duplicate_sample_error_list = [] #エラー出力用データ
     duplicate_groups = {} #同値データのグループ情報
@@ -1901,7 +1817,7 @@ def invalid_bioproject_type (rule_code, sample_name, bioproject_accession, line_
     if is_umbrella == true #NG
       annotation = [
           {key: "Sample name", value: sample_name},
-          {key: "bioproject_accession", value: bioproject_accession}
+          {key: "bioproject_id", value: bioproject_accession}
       ]
       error_hash = CommonUtils::error_obj(@validation_config["rule" + rule_code], @data_file, annotation)
       @error_list.push(error_hash)
@@ -1993,7 +1909,7 @@ def warning_about_bioproject_increment (rule_code, biosample_list)
     result = true
     bioproject_accession_list = []
     biosample_list.each do |biosample_data|
-      bioproject_accession_list.push(biosample_data["attributes"]["bioproject_accession"])
+      bioproject_accession_list.push(biosample_data["attributes"]["bioproject_id"])
     end
     compact_list = bioproject_accession_list.compact
     if bioproject_accession_list.size != compact_list.size #nilが含まれていた場合には連続値ではないものとする
@@ -2017,7 +1933,7 @@ def warning_about_bioproject_increment (rule_code, biosample_list)
         biosample_list.each do |biosample_data|
           annotation = [
             {key: "Sample name", value: biosample_data["attributes"]["sample_name"]},
-            {key: "Attribute", value: biosample_data["attributes"]["bioproject_accession"]}
+            {key: "Attribute", value: biosample_data["attributes"]["bioproject_id"]}
           ]
           error_hash = CommonUtils::error_obj(@validation_config["rule" + rule_code], @data_file, annotation)
           @error_list.push(error_hash)
@@ -2105,13 +2021,13 @@ def bioproject_submission_id_replacement (rule_code, sample_name, psub_id, line_
 
       annotation = [
           {key: "Sample name", value: sample_name},
-          {key: "Attribute", value: "bioproject_accession"},
+          {key: "Attribute", value: "bioproject_id"},
           {key: "Attribute value", value: psub_id}
       ]
 
       # biosample_accessionにAuto-annotationできる
       if !biosample_accession.nil?
-        location = @xml_convertor.xpath_from_attrname("bioproject_accession", line_num)
+        location = @xml_convertor.xpath_from_attrname("bioproject_id", line_num)
         annotation.push(CommonUtils::create_suggested_annotation([biosample_accession], "Attribute value", location, true))
         error_hash = CommonUtils::error_obj(@validation_config["rule" + rule_code], @data_file, annotation, true)
         @error_list.push(error_hash)

diff --git a/src/test/data/biosample/14_not_predefined_attribute_name_SSUB000019_error.xml b/src/test/data/biosample/14_not_predefined_attribute_name_SSUB000019_error.xml
@@ -40,7 +40,7 @@
         <Attribute attribute_name="elev">88.8m</Attribute>
         <Attribute attribute_name="user_attr1">user_data1</Attribute>
         <Attribute attribute_name="depth">0m</Attribute>
-        <Attribute attribute_name="bioproject_accession">PRJDB1654</Attribute>
+        <Attribute attribute_name="bioproject_id">PRJDB1654</Attribute>
         <Attribute attribute_name="user_attr2">user_data2</Attribute>
     </Attributes>
 </BioSample>
diff --git a/src/test/data/biosample/14_not_predefined_attribute_name_SSUB000019_ok.xml b/src/test/data/biosample/14_not_predefined_attribute_name_SSUB000019_ok.xml
@@ -39,6 +39,6 @@
         <Attribute attribute_name="ref_biomaterial">16088826</Attribute>
         <Attribute attribute_name="elev">88.8m</Attribute>
         <Attribute attribute_name="depth">0m</Attribute>
-        <Attribute attribute_name="bioproject_accession">PRJDB1654</Attribute>
+        <Attribute attribute_name="bioproject_id">PRJDB1654</Attribute>
     </Attributes>
 </BioSample>
diff --git a/src/test/data/biosample/18_missing_sample_name_SSUB000019_error.xml b/src/test/data/biosample/18_missing_sample_name_SSUB000019_error.xml
@@ -39,6 +39,6 @@
         <Attribute attribute_name="ref_biomaterial">16088826</Attribute>
         <Attribute attribute_name="elev">88.8m</Attribute>
         <Attribute attribute_name="depth">0m</Attribute>
-        <Attribute attribute_name="bioproject_accession">PRJDB1654</Attribute>
+        <Attribute attribute_name="bioproject_id">PRJDB1654</Attribute>
     </Attributes>
 </BioSample>
diff --git a/src/test/data/biosample/18_missing_sample_name_SSUB000019_error2.xml b/src/test/data/biosample/18_missing_sample_name_SSUB000019_error2.xml
@@ -38,6 +38,6 @@
         <Attribute attribute_name="ref_biomaterial">16088826</Attribute>
         <Attribute attribute_name="elev">88.8m</Attribute>
         <Attribute attribute_name="depth">0m</Attribute>
-        <Attribute attribute_name="bioproject_accession">PRJDB1654</Attribute>
+        <Attribute attribute_name="bioproject_id">PRJDB1654</Attribute>
     </Attributes>
 </BioSample>
diff --git a/src/test/data/biosample/18_missing_sample_name_SSUB000019_ok.xml b/src/test/data/biosample/18_missing_sample_name_SSUB000019_ok.xml
@@ -39,6 +39,6 @@
         <Attribute attribute_name="ref_biomaterial">16088826</Attribute>
         <Attribute attribute_name="elev">88.8m</Attribute>
         <Attribute attribute_name="depth">0m</Attribute>
-        <Attribute attribute_name="bioproject_accession">PRJDB1654</Attribute>
+        <Attribute attribute_name="bioproject_id">PRJDB1654</Attribute>
     </Attributes>
 </BioSample>
diff --git a/src/test/data/biosample/20_missing_organism_SSUB000019_error.xml b/src/test/data/biosample/20_missing_organism_SSUB000019_error.xml
@@ -39,6 +39,6 @@
         <Attribute attribute_name="ref_biomaterial">16088826</Attribute>
         <Attribute attribute_name="elev">88.8m</Attribute>
         <Attribute attribute_name="depth">0m</Attribute>
-        <Attribute attribute_name="bioproject_accession">PRJDB1654</Attribute>
+        <Attribute attribute_name="bioproject_id">PRJDB1654</Attribute>
     </Attributes>
 </BioSample>
diff --git a/src/test/data/biosample/20_missing_organism_SSUB000019_error2.xml b/src/test/data/biosample/20_missing_organism_SSUB000019_error2.xml
@@ -38,6 +38,6 @@
         <Attribute attribute_name="ref_biomaterial">16088826</Attribute>
         <Attribute attribute_name="elev">88.8m</Attribute>
         <Attribute attribute_name="depth">0m</Attribute>
-        <Attribute attribute_name="bioproject_accession">PRJDB1654</Attribute>
+        <Attribute attribute_name="bioproject_id">PRJDB1654</Attribute>
     </Attributes>
 </BioSample>
diff --git a/src/test/data/biosample/20_missing_organism_SSUB000019_ok.xml b/src/test/data/biosample/20_missing_organism_SSUB000019_ok.xml
@@ -39,6 +39,6 @@
         <Attribute attribute_name="ref_biomaterial">16088826</Attribute>
         <Attribute attribute_name="elev">88.8m</Attribute>
         <Attribute attribute_name="depth">0m</Attribute>
-        <Attribute attribute_name="bioproject_accession">PRJDB1654</Attribute>
+        <Attribute attribute_name="bioproject_id">PRJDB1654</Attribute>
     </Attributes>
 </BioSample>
diff --git a/src/test/data/biosample/24_identical_attributes_SSUB004321_ng.xml b/src/test/data/biosample/24_identical_attributes_SSUB004321_ng.xml
@@ -39,7 +39,7 @@
     </Models>
     <Attributes>
       <Attribute attribute_name="sample_name">coffee-1</Attribute>
-      <Attribute attribute_name="bioproject_accession">PRJDB3929</Attribute>
+      <Attribute attribute_name="bioproject_id">PRJDB3929</Attribute>
       <Attribute attribute_name="collection_date">2015-05-08</Attribute>
       <Attribute attribute_name="env_biome">village biome</Attribute>
       <Attribute attribute_name="env_feature">farm</Attribute>
@@ -90,7 +90,7 @@
     </Models>
     <Attributes>
       <Attribute attribute_name="sample_name">coffee-2</Attribute>
-      <Attribute attribute_name="bioproject_accession">PRJDB3929</Attribute>
+      <Attribute attribute_name="bioproject_id">PRJDB3929</Attribute>
       <Attribute attribute_name="collection_date">2015-05-08</Attribute>
       <Attribute attribute_name="env_biome">village biome</Attribute>
       <Attribute attribute_name="env_feature">farm</Attribute>
@@ -141,7 +141,7 @@
     </Models>
     <Attributes>
       <Attribute attribute_name="sample_name">coffee-3</Attribute>
-      <Attribute attribute_name="bioproject_accession">PRJDB3929</Attribute>
+      <Attribute attribute_name="bioproject_id">PRJDB3929</Attribute>
       <Attribute attribute_name="collection_date">2015-05-08</Attribute>
       <Attribute attribute_name="env_biome">village biome</Attribute>
       <Attribute attribute_name="env_feature">farm</Attribute>
@@ -192,7 +192,7 @@
     </Models>
     <Attributes>
       <Attribute attribute_name="sample_name">coffee-4</Attribute>
-      <Attribute attribute_name="bioproject_accession">PRJDB3929</Attribute>
+      <Attribute attribute_name="bioproject_id">PRJDB3929</Attribute>
       <Attribute attribute_name="collection_date">2015-05-08</Attribute>
       <Attribute attribute_name="env_biome">village biome</Attribute>
       <Attribute attribute_name="env_feature">farm</Attribute>
@@ -243,7 +243,7 @@
     </Models>
     <Attributes>
       <Attribute attribute_name="sample_name">goma-1</Attribute>
-      <Attribute attribute_name="bioproject_accession">PRJDB3929</Attribute>
+      <Attribute attribute_name="bioproject_id">PRJDB3929</Attribute>
       <Attribute attribute_name="collection_date">2015-05-08</Attribute>
       <Attribute attribute_name="env_biome">village biome</Attribute>
       <Attribute attribute_name="env_feature">farm</Attribute>

diff --git a/src/test/data/biosample/24_identical_attributes_SSUB004321_ok.xml b/src/test/data/biosample/24_identical_attributes_SSUB004321_ok.xml
@@ -39,7 +39,7 @@
     </Models>
     <Attributes>
       <Attribute attribute_name="sample_name">coffee-1</Attribute>
-      <Attribute attribute_name="bioproject_accession">PRJDB3929</Attribute>
+      <Attribute attribute_name="bioproject_id">PRJDB3929</Attribute>
       <Attribute attribute_name="collection_date">2015-05-08</Attribute>
       <Attribute attribute_name="env_biome">village biome</Attribute>
       <Attribute attribute_name="env_feature">farm</Attribute>
@@ -91,7 +91,7 @@
     </Models>
     <Attributes>
       <Attribute attribute_name="sample_name">coffee-2</Attribute>
-      <Attribute attribute_name="bioproject_accession">PRJDB3929</Attribute>
+      <Attribute attribute_name="bioproject_id">PRJDB3929</Attribute>
       <Attribute attribute_name="collection_date">2015-05-08</Attribute>
       <Attribute attribute_name="env_biome">village biome</Attribute>
       <Attribute attribute_name="env_feature">farm</Attribute>
@@ -143,7 +143,7 @@
     </Models>
     <Attributes>
       <Attribute attribute_name="sample_name">coffee-3</Attribute>
-      <Attribute attribute_name="bioproject_accession">PRJDB3929</Attribute>
+      <Attribute attribute_name="bioproject_id">PRJDB3929</Attribute>
       <Attribute attribute_name="collection_date">2015-05-08</Attribute>
       <Attribute attribute_name="env_biome">village biome</Attribute>
       <Attribute attribute_name="env_feature">farm</Attribute>
@@ -195,7 +195,7 @@
     </Models>
     <Attributes>
       <Attribute attribute_name="sample_name">coffee-4</Attribute>
-      <Attribute attribute_name="bioproject_accession">PRJDB3929</Attribute>
+      <Attribute attribute_name="bioproject_id">PRJDB3929</Attribute>
       <Attribute attribute_name="collection_date">2015-05-08</Attribute>
       <Attribute attribute_name="env_biome">village biome</Attribute>
       <Attribute attribute_name="env_feature">farm</Attribute>
@@ -247,7 +247,7 @@
     </Models>
     <Attributes>
       <Attribute attribute_name="sample_name">goma-1</Attribute>
-      <Attribute attribute_name="bioproject_accession">PRJDB3929</Attribute>
+      <Attribute attribute_name="bioproject_id">PRJDB3929</Attribute>
       <Attribute attribute_name="collection_date">2015-05-08</Attribute>
       <Attribute attribute_name="env_biome">village biome</Attribute>
       <Attribute attribute_name="env_feature">farm</Attribute>

diff --git a/src/test/data/biosample/25_missing_package_information_error.xml b/src/test/data/biosample/25_missing_package_information_error.xml
@@ -38,6 +38,6 @@
         <Attribute attribute_name="ref_biomaterial">16088826</Attribute>
         <Attribute attribute_name="elev">88.8m</Attribute>
         <Attribute attribute_name="depth">0m</Attribute>
-        <Attribute attribute_name="bioproject_accession">PRJDB1654</Attribute>
+        <Attribute attribute_name="bioproject_id">PRJDB1654</Attribute>
     </Attributes>
 </BioSample>
diff --git a/src/test/data/biosample/25_missing_package_information_ok.xml b/src/test/data/biosample/25_missing_package_information_ok.xml
@@ -39,6 +39,6 @@
         <Attribute attribute_name="ref_biomaterial">16088826</Attribute>
         <Attribute attribute_name="elev">88.8m</Attribute>
         <Attribute attribute_name="depth">0m</Attribute>
-        <Attribute attribute_name="bioproject_accession">PRJDB1654</Attribute>
+        <Attribute attribute_name="bioproject_id">PRJDB1654</Attribute>
     </Attributes>
 </BioSample>
diff --git a/src/test/data/biosample/27_missing_mandatory_attribute_SSUB000019_error.xml b/src/test/data/biosample/27_missing_mandatory_attribute_SSUB000019_error.xml
@@ -39,6 +39,6 @@
         <Attribute attribute_name="ref_biomaterial">16088826</Attribute>
         <Attribute attribute_name="elev">88.8m</Attribute>
         <Attribute attribute_name="depth">0m</Attribute>
-        <Attribute attribute_name="bioproject_accession">PRJDB1654</Attribute>
+        <Attribute attribute_name="bioproject_id">PRJDB1654</Attribute>
     </Attributes>
 </BioSample>