Add ISBN file and job

Co-authored-by: Beck Davis <[email protected]>
Co-authored-by: Jane Sandberg <[email protected]>
Co-authored-by: Mark Zelesky <[email protected]>

app/models/gobi/isbn_file.rb

@@ -0,0 +1,106 @@
+# frozen_string_literal: false
+
+module Gobi
+  class IsbnFile
+    attr_reader :received_items_file, :bib_hash
+    def initialize(temp_file:)
+      @received_items_file = temp_file
+      @bib_hash = {}
+    end
+
+    def process
+      CSV.foreach(received_items_file, headers: true, encoding: 'bom|utf-8') do |row|
+        next unless published_within_five_years?(row:)
+        next unless relevant_library_code?(row:)
+        isbns = isbns_for_report(row:)
+        next if isbns.blank?
+        build_bib_hash(row:)
+      end
+      write_bib_hash_to_csv
+      Gobi::IsbnReportJob.working_file_name
+    end
+
+    def build_bib_hash(row:)
+      bib_id = row["MMS Id"]
+      if @bib_hash[bib_id]
+        @bib_hash[bib_id][:loc_combos] << loc_combo(row:)
+      else
+        @bib_hash[bib_id] = {
+          isbns: isbns_for_report(row:),
+          loc_combos: [loc_combo(row:)]
+        }
+      end
+    end
+
+    def write_bib_hash_to_csv
+      CSV.open(Gobi::IsbnReportJob.working_file_path, 'a', headers: true, encoding: 'bom|utf-8', col_sep: "|") do |csv|
+        @bib_hash.each do |_bib, data|
+          code_string = code_string(loc_combos: data[:loc_combos])
+          data[:isbns].each do |isbn|
+            csv << [isbn, code_string, Rails.application.config.gobi_sftp.gobi_account_code]
+          end
+        end
+      end
+    end
+
+    def code_string(loc_combos:)
+      limited_locations = Rails.application.config.gobi_locations.limited_locations
+      shared_locations = Rails.application.config.gobi_locations.shared_locations
+      has_limited_locations = (loc_combos & limited_locations)
+      has_shared_locations = (loc_combos & shared_locations)
+      has_circ_locations = (loc_combos - has_limited_locations - has_shared_locations).present?
+      code_string = ''
+      code_string << 'Cir' if has_circ_locations
+      code_string << 'NC' if has_limited_locations.present?
+      code_string << 'RCP' if has_shared_locations.present?
+      code_string
+    end
+
+    def loc_combo(row:)
+      library = row['Library Code']
+      location = row['Location Code']
+      [library, location].join('$')
+    end
+
+    # with items published in the last 5 years
+    #   Edge cases: 202u (sometime in the 2020s), 9999, publication dates in the future
+    def published_within_five_years?(row:)
+      pub_year = row["Begin Publication Date"]
+      return false if pub_year == '9999' || pub_year.nil?
+
+      pub_year.sub!('u', '0')
+      return false if pub_year.to_i < Time.current.year - 5
+
+      true
+    end
+
+    # that are not Library Code zobsolete, online, resshare
+    def relevant_library_code?(row:)
+      return true unless %w[zobsolete online resshare].include?(row["Library Code"])
+
+      false
+    end
+
+    # Need to have a row for each valid ISBN
+    # Could convert to 13 digit ISBN and de-dup
+    def isbns_for_report(row:)
+      isbns = row["ISBN Valid"].split("\; ").map { |isbn| isbn.delete(':') }
+      isbns.select! { |isbn| isbn.size == 13 || isbn.size == 10 }
+      isbns.map { |isbn| convert_to_isbn_thirteen(isbn:) }.uniq
+    end
+
+    def convert_to_isbn_thirteen(isbn:)
+      return isbn if isbn.length == 13
+
+      isbn.chop!
+      isbn.prepend('978')
+      check_digit_arr = isbn.split("").map(&:to_i)
+      modulo = check_digit_arr.each_with_index.map do |num, index|
+        index.odd? ? num * 3 : num * 1
+      end.sum % 10
+      check_digit = modulo.zero? ? 0 : 10 - modulo
+      isbn << check_digit.to_s
+      isbn
+    end
+  end
+end

app/models/gobi/isbn_report_job.rb

@@ -0,0 +1,55 @@
+# frozen_string_literal: true
+
+module Gobi
+  class IsbnReportJob < LibJob
+    attr_reader :report_downloader, :report_uploader
+
+    def initialize
+      super(category: 'Gobi:IsbnReports')
+      @report_downloader = ReportDownloader.new(
+        sftp: AlmaSftp.new,
+        file_pattern: 'received_items_published_last_5_years_\d{12}.csv',
+        input_sftp_base_dir: '/alma/isbns',
+        process_class: Gobi::IsbnFile
+      )
+      @report_uploader = ReportUploader.new(
+        sftp: GobiSftp.new,
+        working_file_names: [IsbnReportJob.working_file_name],
+        working_file_directory: Rails.application.config.gobi_sftp.working_directory,
+        output_sftp_base_dir: Rails.application.config.gobi_sftp.output_sftp_base_dir
+      )
+    end
+
+    def self.working_file_path
+      working_file_directory = Rails.application.config.gobi_sftp.working_directory
+      FileUtils.mkdir_p(working_file_directory) unless File.exist?(working_file_directory)
+      File.join(working_file_directory, IsbnReportJob.working_file_name)
+    end
+
+    def self.working_file_name
+      date = Time.now.utc.strftime('%Y-%m-%d')
+      "#{date}-gobi-isbn-updates.txt"
+    end
+
+    def handle(data_set:)
+      create_csv
+      report_downloader.run
+      report_uploader.run
+      rename_files_on_sftp
+      data_set.data = "Number of ISBNs sent: #{CSV.read(IsbnReportJob.working_file_path).length}"
+      data_set
+    end
+
+    def rename_files_on_sftp
+      AlmaSftp.new.start do |sftp|
+        report_downloader.remote_filenames.each do |file_name|
+          sftp.rename(file_name, "#{file_name}.processed")
+        end
+      end
+    end
+
+    def create_csv
+      CSV.open(IsbnReportJob.working_file_path, 'w', encoding: 'bom|utf-8', col_sep: "|")
+    end
+  end
+end

config/application.rb

@@ -40,6 +40,7 @@ def config_for(*args)
 
     config.alma_sftp = config_for(:alma_sftp)
     config.gobi_sftp = config_for(:gobi_sftp)
+    config.gobi_locations = config_for(:gobi_locations)
     config.oclc_sftp = config_for(:oclc_sftp)
     config.lc_call_slips = config_for(:lc_call_slips)
     config.peoplesoft = config_for(:peoplesoft)

config/gobi_locations.yml

@@ -0,0 +1,85 @@
+default: &default
+  limited_locations:
+    - arch$ref
+    - arch$la
+    - eastasian$cjkref
+    - eastasian$ref
+    - eastasian$hyref
+    - engineer$ref
+    - firestone$aas
+    - firestone$dra
+    - firestone$dr
+    - firestone$drrr
+    - firestone$flm
+    - firestone$sc
+    - firestone$se
+    - firestone$slav
+    - firestone$sh
+    - firestone$shs
+    - firestone$sne
+    - firestone$spc
+    - firestone$sss
+    - firestone$ssa
+    - firestone$srel
+    - firestone$gestf
+    - firestone$dss
+    - firestone$fis
+    - firestone$sps
+    - firestone$seref
+    - firestone$egsr
+    - firestone$gss
+    - firestone$noncirc
+    - firestone$necnc
+    - firestone$clasnc
+    - firestone$xlnc
+    - firestone$secw
+    - firestone$ssrcfo
+    - firestone$ssrcdc
+    - plasma$ref
+    - plasma$rr
+    - plasma$la
+    - plasma$li
+    - plasma$theses
+    - plasma$nb
+    - lewis$lal
+    - lewis$laf
+    - lewis$refid
+    - lewis$ref
+    - marquand$fesrf
+    - marquand$ph
+    - marquand$phref
+    - marquand$ref
+    - mendel$rg
+    - mendel$facs
+    - mendel$ref
+    - mendel$pe
+    - mudd$scamudd
+    - rare$gestrare
+    - rare$scactsn
+    - rare$scathx
+    - rare$scahsvm
+    - rare$scagax
+    - rare$scaex
+    - rare$scamss
+    - rare$scawa
+    - rare$scahsvc
+    - rare$scawhs
+    - stokes$ref
+  shared_locations:
+    - recap$pa
+    - recap$gp
+    - mendel$qk
+    - firestone$pf
+    - marquand$pv
+
+development:
+  <<: *default
+
+test:
+  <<: *default
+
+staging: &staging
+  <<: *default
+
+production:
+  <<: *staging