Extend support of miga ncbi_get

Include genomes without reported FTP path in summary tables

Author: lmrodriguezr

.github/workflows/build.yml

@@ -63,10 +63,10 @@ jobs:
         uses: actions-ecosystem/action-regex-match@v2
         with:
           text: ${{ steps.publish.outputs.gem_version }}
-          regex: '^([0-9]*\.[0-9]*)\.'
+          regex: '^[0-9]*\.[0-9]*'
       - name: Update major.minor tag
         if: ${{ steps.major-minor.outputs.match != '' }}
         uses: actions-ecosystem/action-push-tag@v1
         with:
-          tag: v${{ steps.regex-match.outputs.group1 }}
+          tag: v${{ steps.regex-match.outputs.match }}
 

lib/miga/cli/action/download/ncbi.rb

@@ -71,16 +71,13 @@ def parse_csv_as_datasets(doc)
     CSV.parse(doc, headers: true).each do |r|
       asm = r['assembly']
       next if asm.nil? || asm.empty? || asm == '-'
-      next unless r['ftp_path_genbank']
 
       rep = remote_row_replicons(r)
       n = remote_row_name(r, rep, asm)
 
       # Register for download
-      fna_url = '%s/%s_genomic.fna.gz' %
-                [r['ftp_path_genbank'], File.basename(r['ftp_path_genbank'])]
       ds[n] = {
-        ids: [fna_url], db: :assembly_gz, universe: :web,
+        ids: [asm], db: :assembly, universe: :ncbi,
         md: {
           type: :genome, ncbi_asm: asm, strain: r['strain']
         }
@@ -123,8 +120,7 @@ def remote_list_url
         'matching(tab==["Prokaryotes"] and q=="' \
           "#{cli[:taxon]&.tr('"', "'")}\"",
       fields: 'organism|organism,assembly|assembly,replicons|replicons,' \
-        'level|level,ftp_path_genbank|ftp_path_genbank,' \
-        'release_date|release_date,strain|strain',
+        'level|level,release_date|release_date,strain|strain',
       nolimit: 'on'
     }
     if cli[:reference]

lib/miga/remote_dataset.rb

@@ -216,7 +216,7 @@ def ncbi_asm_json_doc
     end
     return nil unless metadata[:ncbi_asm]
 
-    ncbi_asm_id = self.class.ncbi_asm_acc2id metadata[:ncbi_asm]
+    ncbi_asm_id = self.class.ncbi_asm_acc2id(metadata[:ncbi_asm])
     txt = nil
     3.times do
       txt = self.class.download(:ncbi_summary, :assembly, ncbi_asm_id, :json)
@@ -226,6 +226,12 @@ def ncbi_asm_json_doc
     return if doc.nil? || doc['result'].nil? || doc['result'].empty?
 
     @_ncbi_asm_json_doc = doc['result'][ doc['result']['uids'].first ]
+    url_dir = @_ncbi_asm_json_doc['ftppath_genbank']
+    if url_dir
+      metadata[:web_assembly_gz] ||=
+        '%s/%s_genomic.fna.gz' % [url_dir, File.basename(url_dir)]
+    end
+    @_ncbi_asm_json_doc
   end
 
   private

lib/miga/remote_dataset/download.rb

@@ -49,7 +49,7 @@ def ncbi_asm_rest(opts)
         )
       end
 
-      url = "#{url_dir}/#{File.basename url_dir}_genomic.fna.gz"
+      url = '%s/%s_genomic.fna.gz' % [url_dir, File.basename(url_dir)]
       download(
         :web, :assembly_gz, url,
         opts[:format], opts[:file], opts[:extra], opts[:obj]
@@ -134,9 +134,8 @@ module MiGA::RemoteDataset::Download
   ##
   # Download data into +file+
   def download(file)
-    self.class.download(
-      universe, db, ids, self.class.UNIVERSE[universe][:dbs][db][:format],
-      file, [], self
-    )
+    format = self.class.UNIVERSE[universe][:dbs][db][:format]
+    # MiGA::MiGA.DEBUG "download: #{universe}, #{db}, #{ids}, #{format}"
+    self.class.download(universe, db, ids, format, file, [], self)
   end
 end

lib/miga/version.rb

@@ -12,15 +12,15 @@ module MiGA
   # - String indicating release status:
   #   - rc* release candidate, not released as gem
   #   - [0-9]+ stable release, released as gem
-  VERSION = [1.3, 1, 0].freeze
+  VERSION = [1.3, 2, 0].freeze
 
   ##
   # Nickname for the current major.minor version.
-  VERSION_NAME = 'prima'
+  VERSION_NAME = 'mezzotint'
 
   ##
   # Date of the current gem relese.
-  VERSION_DATE = Date.new(2023, 3, 21)
+  VERSION_DATE = Date.new(2023, 3, 25)
 
   ##
   # References of MiGA