Merge pull request #226 from ncbo/csv_format_change_fix

CSV format change fix
ncbo · Nov 18, 2024 · 073fae6 · 073fae6
2 parents db05a3b + d380a88
commit 073fae6
Show file tree

Hide file tree

Showing 4 changed files with 270 additions and 20 deletions.
diff --git a/Gemfile.lock b/Gemfile.lock
@@ -55,23 +55,18 @@ GEM
       launchy (>= 2.1, < 4.0)
       mail (~> 2.7)
     eventmachine (1.2.7)
-    faraday (2.12.0)
-      faraday-net_http (>= 2.0, < 3.4)
-      json
-      logger
-    faraday-net_http (3.3.0)
-      net-http
-    ffi (1.17.0-aarch64-linux-gnu)
-    ffi (1.17.0-arm64-darwin)
-    ffi (1.17.0-x86_64-linux-gnu)
+    faraday (1.2.0)
+      multipart-post (>= 1.2, < 3)
+      ruby2_keywords
+    ffi (1.17.0)
     hashie (5.0.0)
     htmlentities (4.3.4)
     http-accept (1.7.0)
     http-cookie (1.0.7)
       domain_name (~> 0.5)
     i18n (0.9.5)
       concurrent-ruby (~> 1.0)
-    json (2.8.1)
+    json (2.8.2)
     json_pure (2.8.1)
     language_server-protocol (3.17.0.3)
     launchy (3.0.1)
@@ -99,8 +94,7 @@ GEM
       minitest (>= 2.12, < 5.0)
       powerbar
     multi_json (1.15.0)
-    net-http (0.5.0)
-      uri
+    multipart-post (2.4.1)
     net-http-persistent (2.9.4)
     net-imap (0.4.18)
       date
@@ -126,10 +120,10 @@ GEM
       mail (>= 2.0)
     powerbar (2.0.1)
       hashie (>= 1.1.0)
-    pry (0.14.2)
+    pry (0.15.0)
       coderay (~> 1.1)
       method_source (~> 1.0)
-    public_suffix (6.0.1)
+    public_suffix (5.1.1)
     racc (1.8.1)
     rack (2.2.10)
     rack-test (0.8.3)
@@ -167,6 +161,7 @@ GEM
     rubocop-ast (1.36.1)
       parser (>= 3.3.1.0)
     ruby-progressbar (1.13.0)
+    ruby2_keywords (0.0.5)
     rubyzip (1.3.0)
     simplecov (0.22.0)
       docile (~> 1.1)
@@ -186,7 +181,6 @@ GEM
     timeout (0.4.2)
     tzinfo (0.3.62)
     unicode-display_width (2.6.0)
-    uri (1.0.2)
     uuid (2.3.9)
       macaddr (~> 1.0)
 

diff --git a/lib/ontologies_linked_data/utils/ontology_csv_writer.rb b/lib/ontologies_linked_data/utils/ontology_csv_writer.rb
@@ -39,7 +39,7 @@ def write_class(ont_class)
         row[CLASS_ID] = ont_class.id
 
         # Preferred label
-        row[PREF_LABEL] = ont_class.prefLabel
+        row[PREF_LABEL] = Array(ont_class.prefLabel).first
 
         # Synonyms
         synonyms = ont_class.synonym
@@ -50,7 +50,7 @@ def write_class(ont_class)
         row[DEFINITIONS] = definitions.join('|') unless definitions.empty?
 
         # Obsolete
-        row[OBSOLETE] = ont_class.obsolete
+        row[OBSOLETE] = Array(ont_class.obsolete).first.to_s.upcase
 
         # CUI
         cuis = ont_class.cui

diff --git a/test/data/ontology_files/chebi_test.obo b/test/data/ontology_files/chebi_test.obo
@@ -0,0 +1,228 @@
+format-version: 1.2
+data-version: 237
+date: 30:10:2024 06:16
+saved-by: chebi
+subsetdef: 1_STAR "Preliminary entries"
+subsetdef: 2_STAR "Annotated by 3rd party"
+subsetdef: 3_STAR "Manually annotated by ChEBI Team"
+synonymtypedef: BRAND_NAME "BRAND NAME"
+synonymtypedef: INN "INN"
+synonymtypedef: IUPAC_NAME "IUPAC NAME"
+default-namespace: chebi_ontology
+remark: Author: ChEBI curation team
+remark: ChEBI Release version 237
+remark: ChEBI subsumes and replaces the Chemical Ontology first
+remark: developed by Michael Ashburner & Pankaj Jaiswal.
+remark: For any queries contact [email protected]
+ontology: chebi
+
+[Term]
+id: CHEBI:137366
+name: CHEBI:4042
+is_obsolete: true
+
+[Term]
+id: CHEBI:137377
+name: CHEBI:81850
+is_obsolete: true
+
+[Term]
+id: CHEBI:143109
+name: waterssdfsdfss
+is_obsolete: true
+
+[Term]
+id: CHEBI:177198
+name: CHEBI:50860
+is_obsolete: true
+
+[Term]
+id: CHEBI:189822
+name: testing532
+is_obsolete: true
+
+[Term]
+id: CHEBI:24431
+name: chemical entity
+def: "A chemical entity is a physical entity of interest in chemistry including molecular entities, parts thereof, and chemical substances." []
+subset: 3_STAR
+synonym: "chemical entity" EXACT [UniProt]
+
+[Term]
+id: CHEBI:27189
+name: unclassifieds
+is_obsolete: true
+
+[Term]
+id: CHEBI:30430
+name: indium atom
+def: "A metallic element first identified and named from the brilliant indigo (Latin indicum) blue line in its flame spectrum." []
+subset: 3_STAR
+synonym: "49In" RELATED [IUPAC]
+synonym: "In" RELATED [IUPAC]
+synonym: "indio" RELATED [ChEBI]
+synonym: "Indium" RELATED [ChEBI]
+synonym: "indium" EXACT IUPAC_NAME [IUPAC]
+synonym: "indium" RELATED [ChEBI]
+xref: CAS:7440-74-6 {source="ChemIDplus"}
+xref: CAS:7440-74-6 {source="NIST Chemistry WebBook"}
+xref: Gmelin:16297 {source="Gmelin"}
+xref: WebElements:In
+is_a: CHEBI:33317 ! boron group element atom
+property_value: http://purl.obolibrary.org/obo/chebi/charge "0" xsd:string
+property_value: http://purl.obolibrary.org/obo/chebi/formula "In" xsd:string
+property_value: http://purl.obolibrary.org/obo/chebi/inchi "InChI=1S/In" xsd:string
+property_value: http://purl.obolibrary.org/obo/chebi/inchikey "APFVFJFRJDLVQX-UHFFFAOYSA-N" xsd:string
+property_value: http://purl.obolibrary.org/obo/chebi/mass "114.81800" xsd:string
+property_value: http://purl.obolibrary.org/obo/chebi/monoisotopicmass "114.90388" xsd:string
+property_value: http://purl.obolibrary.org/obo/chebi/smiles "[In]" xsd:string
+
+[Term]
+id: CHEBI:33250
+name: atom
+alt_id: CHEBI:22671
+alt_id: CHEBI:23907
+def: "A chemical entity constituting the smallest component of an element having the chemical properties of the element." []
+subset: 3_STAR
+synonym: "atom" EXACT IUPAC_NAME [IUPAC]
+synonym: "atome" RELATED [IUPAC]
+synonym: "atomo" RELATED [IUPAC]
+synonym: "atoms" RELATED [ChEBI]
+synonym: "atomus" RELATED [ChEBI]
+synonym: "element" RELATED [ChEBI]
+synonym: "elements" RELATED [ChEBI]
+is_a: CHEBI:24431 ! chemical entity
+
+[Term]
+id: CHEBI:33317
+name: boron group element atom
+subset: 3_STAR
+synonym: "boron group element" RELATED [ChEBI]
+synonym: "boron group elements" RELATED [ChEBI]
+synonym: "Element der Borgruppe" RELATED [ChEBI]
+synonym: "group 13 elements" EXACT IUPAC_NAME [IUPAC]
+synonym: "group III elements" RELATED [ChEBI]
+is_a: CHEBI:33560 ! p-block element atom
+
+[Term]
+id: CHEBI:33318
+name: main group element atom
+def: "An atom belonging to one of the main groups (found in the s- and p- blocks) of the periodic table." []
+subset: 3_STAR
+synonym: "Hauptgruppenelement" RELATED [ChEBI]
+synonym: "Hauptgruppenelemente" RELATED [ChEBI]
+synonym: "main group element" RELATED [ChEBI]
+synonym: "main group elements" EXACT IUPAC_NAME [IUPAC]
+is_a: CHEBI:33250 ! atom
+
+[Term]
+id: CHEBI:33560
+name: p-block element atom
+def: "Any main group element atom belonging to the p-block of the periodic table." []
+subset: 3_STAR
+synonym: "p-block element" RELATED [ChEBI]
+synonym: "p-block elements" RELATED [ChEBI]
+is_a: CHEBI:33318 ! main group element atom
+
+[Term]
+id: CHEBI:49631
+name: gallium atom
+alt_id: CHEBI:33326
+alt_id: CHEBI:49630
+def: "A metallic element predicted as eka-aluminium by Mendeleev in 1870 and discovered by Paul-Emile Lecoq de Boisbaudran in 1875. Named in honour of France (Latin Gallia) and perhaps also from the Latin gallus cock, a translation of Lecoq." []
+subset: 3_STAR
+synonym: "31Ga" RELATED [IUPAC]
+synonym: "Ga" RELATED [IUPAC]
+synonym: "galio" RELATED [ChEBI]
+synonym: "gallium" EXACT IUPAC_NAME [IUPAC]
+synonym: "gallium" RELATED [ChEBI]
+xref: CAS:7440-55-3 {source="ChemIDplus"}
+xref: CAS:7440-55-3 {source="NIST Chemistry WebBook"}
+xref: WebElements:Ga
+is_a: CHEBI:33317 ! boron group element atom
+property_value: http://purl.obolibrary.org/obo/chebi/charge "0" xsd:string
+property_value: http://purl.obolibrary.org/obo/chebi/formula "Ga" xsd:string
+property_value: http://purl.obolibrary.org/obo/chebi/inchi "InChI=1S/Ga" xsd:string
+property_value: http://purl.obolibrary.org/obo/chebi/inchikey "GYHNNYVSQQEPJS-UHFFFAOYSA-N" xsd:string
+property_value: http://purl.obolibrary.org/obo/chebi/mass "69.72300" xsd:string
+property_value: http://purl.obolibrary.org/obo/chebi/monoisotopicmass "68.92557" xsd:string
+property_value: http://purl.obolibrary.org/obo/chebi/smiles "[Ga]" xsd:string
+
+[Term]
+id: CHEBI:64352
+name: UDP-N-acetyl-D-glucosamine(2-)
+is_obsolete: true
+
+[Term]
+id: CHEBI:64360
+name: tocilizumab
+is_obsolete: true
+
+[Term]
+id: CHEBI:64867
+name: PHS C26
+is_obsolete: true
+
+[Typedef]
+id: has_functional_parent
+name: has functional parent
+is_cyclic: false
+is_transitive: false
+
+[Typedef]
+id: has_major_microspecies_at_pH_7_3
+name: has major microspecies at pH 7.3
+is_cyclic: true
+is_transitive: false
+
+[Typedef]
+id: has_parent_hydride
+name: has parent hydride
+is_cyclic: false
+is_transitive: false
+
+[Typedef]
+id: has_part
+name: has part
+xref: BFO:0000051
+is_cyclic: false
+is_transitive: true
+
+[Typedef]
+id: has_role
+name: has role
+xref: RO:0000087
+is_cyclic: false
+is_transitive: false
+
+[Typedef]
+id: is_conjugate_acid_of
+name: is conjugate acid of
+is_cyclic: true
+is_transitive: false
+inverse_of: is_conjugate_base_of ! is conjugate base of
+
+[Typedef]
+id: is_conjugate_base_of
+name: is conjugate base of
+is_cyclic: true
+is_transitive: false
+
+[Typedef]
+id: is_enantiomer_of
+name: is enantiomer of
+is_cyclic: true
+is_transitive: false
+
+[Typedef]
+id: is_substituent_group_from
+name: is substituent group from
+is_cyclic: false
+is_transitive: false
+
+[Typedef]
+id: is_tautomer_of
+name: is tautomer of
+is_cyclic: true
+is_transitive: true
+
diff --git a/test/util/test_ontology_csv_writer.rb b/test/util/test_ontology_csv_writer.rb
@@ -49,8 +49,16 @@ def self.before_suite
   end
 
   def get_csv_string
-    gz = Zlib::GzipReader.open(@@csv_path)
-    return gz.read
+    get_csv_string_from_path(@@csv_path)
+  end
+
+  def enclosed_in_square_brackets_with_quotes?(string)
+    /\A\[\s*(["']).*\1\s*\]\z/ === string
+  end
+
+  def get_csv_string_from_path(csv_path)
+    gz = Zlib::GzipReader.open(csv_path)
+    gz.read
   end
 
   def test_csv_writer_valid
@@ -252,7 +260,7 @@ def test_csv_writer_content_non_obsolete
     classes = CSV.parse(get_csv_string, headers:true)
     classes.select do |row|
       if row[LinkedData::Utils::OntologyCSVWriter::PREF_LABEL] == preferred_label
-        assert_equal 'false', row[LinkedData::Utils::OntologyCSVWriter::OBSOLETE]
+        assert_equal 'false', row[LinkedData::Utils::OntologyCSVWriter::OBSOLETE].to_s.downcase
         class_exists = true
       end
     end
@@ -309,4 +317,24 @@ def test_csv_writer_content_props_other
 
     assert class_exists, %Q<Class not found: "#{preferred_label}">
   end
+
+  def test_for_non_array_values
+    acronym = 'CHEBITEST'
+    sub_id = 1
+    submission_parse(acronym, "CHEBI Ontology TEST",
+                     "./test/data/ontology_files/chebi_test.obo", sub_id,
+                     process_rdf: true, index_search: true, extract_metadata: false)
+    sub = LinkedData::Models::OntologySubmission.where(ontology: [acronym: acronym], submissionId: sub_id)
+                                                .include(:version, :submissionId, :ontology).first
+    sub.ontology.bring(:acronym)
+    classes = CSV.parse(get_csv_string_from_path(sub.csv_path), headers:true)
+    assert_equal 20, classes.count
+
+    classes.each do |row|
+      row.each do |_, val|
+        assert_equal false, enclosed_in_square_brackets_with_quotes?(val), "Expected a String, but received an Array: #{val}"
+      end
+    end
+  end
+
 end