Use StrainCode to parse strain numbers

Author: lmrodriguezr

Gemfile

@@ -37,6 +37,7 @@ gem 'serrano', '~> 1.0'
 # loaded (see https://github.com/heartcombo/simple_form/issues/1824)
 gem 'simple_form', '~> 5.2.0'
 gem 'sqlite3'
+gem 'strain-code', '~> 0.3'
 # Turbolinks makes navigating your web application faster.
 # Read more: https://github.com/turbolinks/turbolinks
 gem 'turbolinks', '~> 5'

Gemfile.lock

@@ -350,6 +350,7 @@ GEM
       mini_portile2 (~> 2.8.0)
     sqlite3 (1.7.3-x86_64-darwin)
     sqlite3 (1.7.3-x86_64-linux)
+    strain-code (0.3.3)
     thor (1.3.2)
     tilt (2.4.0)
     time (0.4.1)
@@ -439,6 +440,7 @@ DEPENDENCIES
   spring
   spring-watcher-listen (~> 2.0.0)
   sqlite3
+  strain-code (~> 0.3)
   turbolinks (~> 5)
   tzinfo-data
   uglifier (>= 1.3.0)

app/assets/stylesheets/custom.scss

@@ -83,6 +83,10 @@ trix-editor {
   }
 }
 
+.tooltip-text {
+  text-decoration: underline dotted theme-color(secondary);
+}
+
 /* Header Bar */
 .bg-dark-ce, .bg-vdark-ce {
   a {

app/helpers/strains_helper.rb

@@ -30,9 +30,13 @@ def strain_numbers_html(numbers, opts = {})
     ext = opts[:ext] ? '<sup class="fas fa-external-link-alt "> </sup>' : ''
 
     numbers.map do |str|
-      if str.is_a? Hash
-        '<a href="%s" target="_blank">%s %s</a>' % [
-          str[:url], sanitize(str[:accession]), ext
+      if str.is_a?(StrainCode::Number) && str.url.present?
+        '<a href="%s" title="%s" target="_blank">%s %s</a>' % [
+          str.url, sanitize(str.catalogue&.name), sanitize(str.number), ext
+        ]
+      elsif str.is_a?(StrainCode::Number) && str.catalogue&.name.present?
+        '<span class="tooltip-text" title="%s">%s</span>' % [
+          sanitize(str.catalogue.name), sanitize(str.number)
         ]
       else
         str

app/models/strain.rb

@@ -12,139 +12,16 @@ class Strain < ApplicationRecord
   include HasExternalResources
   include Strain::ExternalResources
 
-  class << self
-    ##
-    # Hash of culture collection prefixes and URL rule.
-    # See also: https://ftp.ncbi.nih.gov/pub/taxonomy/Ccode_dump.txt
-    # See also: https://jcm.brc.riken.jp/en/abbr_e
-    def culture_collections
-      {
-        ATCC: 'https://www.atcc.org/products/%s',
-        DSM:  'https://www.dsmz.de/collection/catalogue/details/culture/DSM-%s',
-        IFO:  'https://www.nite.go.jp/nbrc/catalogue/' \
-              'NBRCCatalogueDetailServlet?ID=IFO&CAT=%s',
-        JCM:  'https://www.jcm.riken.jp/cgi-bin/jcm/jcm_number?JCM=%s',
-        KCTC: 'https://kctc.kribb.re.kr/collection/view?sn=%s',
-        NBRC: 'https://www.nite.go.jp/nbrc/catalogue/' \
-              'NBRCCatalogueDetailServlet?ID=IFO&CAT=%s',
-        NCTC: 'https://www.culturecollections.org.uk/products/bacteria/' \
-              'detail.jsp?collection=nctc&refId=NCTC+%s',
-
-        # Bioresource Collection and Research Center
-        # [TW: Food Industry Research and Development Insitute]
-        BCRC: 'https://catalog.bcrc.firdi.org.tw/BcrcContent?bid=%s',
-        # Collection of Aquatic Important Microorganisms
-        # [MX: CIAD, Centro de Investigación en Alimentación y Desarrollo]
-        CAIM: 'https://www.ciad.mx/caim/busqueda.php?' \
-              'searchtype=caim&criterio=%s',
-        # Culture Collection of Autotrophic Organisms
-        # [CZ: Czech Academy of Sciences]
-        CCALA: 'https://ccala.butbn.cas.cz/strain/%s',
-        # Czech Collection of Microorganisms
-        # [CZ: Masaryk University]
-        CCM: 'https://www.sci.muni.cz/ccm/bakterie/camb/%s',
-        # China General Microbiological Culture Collection Center
-        # [CN: National Science and Technology Infrastructure]
-        # -- From NCBI Taxonomy (not working):
-        # -- 'http://www.cgmcc.net/english/cata.php?stn=CGMCC%%20%s'
-        CGMCC: 'https://cgmcc.net/english/search?stn=%s',
-        # NCMA, National Center for Marine Algae and Microbiota,
-        # formerly Culture Collection for Marine Phytoplankton
-        # [US: Bigelow Laboratory for Ocean Sciences]
-        CCMP: 'https://ncma.bigelow.org/CCMP%s',
-        # Culture Collection University of Gothenburg
-        # [SE: University of Gothenburg]
-        CCUG: 'https://www.ccug.se/strain?id=%s',
-        # China Center of Industrial Culture Collection
-        # [CN: China National Research Institute of Food & Fermentation
-        # Industries]
-        CICC: 'http://www.china-cicc.org/search/?classtype=1&keyword=%s',
-        # Collection de l'Institut Pasteur
-        # (Collection of the Institut Pasteur)
-        # [FR: Institut Pasteur]
-        CIP:  'https://catalogue-crbip.pasteur.fr/' \
-              'fiche_catalogue.xhtml?crbip=CIP%%20%s',
-        # BCCM, Belgian Coordinated Collections of Microorganisms /
-        # Institute of Tropical Medicine Antwerp Mycobacteria Collection
-        # [BE: Belgian Science Policy (BELSPO)]
-        ITM:  'https://bccm.belspo.be/catalogues/bm-details?' \
-              'accession_number=ITM%%20%s',
-        # Jena Microbial Resource Collection
-        # [DE: Friedrich Schiller University Jena]
-        JMRC: 'http://www.jmrc.uni-jena.de/data.php?fsu=%s',
-        # Korean Agricultural Culture Collection
-        # [KR: National Academy of Agricultural Science]
-        KACC: 'https://genebank.rda.go.kr/eng/mic/cat/MicrobeSearch.do' \
-              '?sSearchWith=no&sTxt1=%s',
-        # BCCM, Belgian Coordinated Collections of Microorganisms /
-        # Bacteria Collection Laboratorium voor Microbiologie Universiteit Gent
-        # (Bacteria Collection Laboratory of Microbiology, University of Ghent)
-        # [BE: Belgian Science Policy (BELSPO)]
-        LMG:  'https://bccm.belspo.be/catalogues/lmg-strain-details?NUM=%s',
-        # Marine Culture Collection of China
-        # [CN: Third Institute of Oceanography]
-        MCCC: 'https://mccc.org.cn/detailRecord3.asp?bcbh=%s',
-        # North East Pacific Culture Collection
-        # [CA: The University of British Columbia]
-        NEPCC: 'https://db.botany.ubc.ca/cccm/mfa/%s',
-        # National Institute for Environmental Studies Collection
-        # [JP: National Institute for Environmental Studies]
-        NIES: 'https://mcc.nies.go.jp/numberSearch.do?strainNumber=%s',
-        # Agricultural Research Service Culture Collection,
-        # formerly Northern Regional Research Laboratory
-        # [US: National Center for Agricultural Utilization Research]
-        NRRL: 'https://nrrl.ncaur.usda.gov/cgi-bin/usda/prokaryote/' \
-              'report.html?nrrlcodes=%s',
-        # Pasteur Cultures of Cyanobacteria
-        # [FR: Institut Pasteur]
-        PCC:  'https://catalogue-crbip.pasteur.fr/' \
-              'fiche_catalogue.xhtml?crbip=PCC%%20%s',
-        # Sammlung von Algenkulturen
-        # (Culture Collection of Algae)
-        # [DE: University of Göttingen]
-        SAG: 'https://sagdb.uni-goettingen.de/detailedList.php?str_number=%s',
-        # Thailand Bioresource Research Center
-        # [TH: Thailand Bioresource Research Center]
-        TBRC: 'https://tbrcnetwork.org/microb_detail.php?code=TBRC_%s',
-        # BCCM, Belgian Coordinated Collections of Microorganisms /
-        # University of of Liège Cyanobacteria Collection
-        # [BE: Belgian Science Policy (BELSPO)]
-        ULC:  'https://bccm.belspo.be/catalogues/bm-details?' \
-              'accession_number=ULC%%20%s',
-        # ВСЕРОССИЙСКАЯ КОЛЛЕКЦИЯ МИКРООРГАНИЗМОВ
-        # (All-Russian Collection of Microorganisms)
-        # [RU: Pushchino Scientific Center for Biological Research of the
-        # Russian Academy of Sciences]
-        VKM: 'http://www.vkm.ru/strains.php?vkm=%s'
-      }
-    end
-  end
-
   def numbers
     numbers_string.split(' = ')
   end
 
   def numbers_parsed
-    numbers.map do |str|
-      parts = str.split(/[ :-]+/, 2)
-      coll = parts.count == 2 ? parts[0].upcase.to_sym : nil
-
-      if url_base = self.class.culture_collections[coll]
-        {
-          collection: coll,
-          accession: parts.join(' '),
-          url: url_base % parts[1]
-        }
-      else
-        str
-      end
-    end
+    @numbers_parsed ||= StrainCode.parse(numbers_string)
   end
 
   def collections
-    numbers_parsed
-      .map { |i| i.is_a?(Hash) ? i[:collection] : nil }
-      .compact.uniq
+    numbers_parsed.map(&:catalogue).compact.uniq
   end
 
   def type_of_type