first stab at an ex-libris holdings xml parser

puts header elsewhere

using ruby-marc instead of plain nokogiri

Gemfile

@@ -20,6 +20,7 @@ gem "zinzout"
 gem "puma"
 gem "sidekiq", "~> 6.0"
 gem "sidekiq-batch"
+gem "marc"
 
 group :development, :test do
   gem "pry"

bin/ex_libris_holdings_xml_parser.rb

@@ -0,0 +1,97 @@
+require "marc"
+
+class ExLibrisHoldingsXmlParser
+  def main(files)
+    files.each do |file|
+      marc_reader = MARC::XMLReader.new(file)
+      marc_reader.each do |marc_record|
+        ht_record = HTRecord.new(marc_record)
+        # next if ...
+        yield ht_record
+      end
+    end
+  end
+end
+
+class HTRecord
+  def initialize(marc_record)
+    @marc_record = marc_record # Marc::Record
+  end
+
+  def self.header_tsv
+    %w[item_type oclc local_id status condition enum_chron issn govdoc].join("\t")
+  end
+
+  # Should use the same order as HTRecord.header_tsv
+  def to_tsv
+    [item_type, oclc, local_id, status, condition, enum_chron, issn, govdoc]
+      .join("\t")
+  end
+
+  def oclc
+    @oclc ||= @marc_record["035"]["a"]
+  end
+
+  def local_id
+    @local_id ||= @marc_record["001"]
+  end
+
+  def condition
+    @condition ||= @marc_record["ITM"]["c"]
+  end
+
+  def item_type
+    @item_type ||= map_item_type(@marc_record["ITM"]["m"])
+  end
+
+  # Todo: Figure out the structure of enum_chrons.
+  def enum_chron
+    @enum_chron ||= [
+      @marc_record["ITM"]["a"],
+      @marc_record["ITM"]["b"],
+      @marc_record["ITM"]["i"],
+      @marc_record["ITM"]["j"]
+    ].reject{ |x| x.nil? || x.empty? }.join(",")
+  end
+
+  def map_item_type(item_type)
+    # Todo: figure out mapping for item_type.
+    # see https://wiki.harvard.edu/confluence/display/LibraryStaffDoc/Item+Material+Types
+    {
+      "ISSUE" => "ser",
+      "DVD"   => "skip",
+      "BOOK"  => "mon",
+    }[item_type] || "mix"
+  end
+
+  # Todo: figure out status.
+  def status
+    @status ||= map_status(@marc_record["ITM"]["k"])
+  end
+
+  # Todo: figure out mapping for status.
+  def map_status(status)
+    {
+      "MISSING" => "LM",
+      "LOST_LOAN"  => "LM",
+    }[status] || "CH"
+  end
+
+  # Todo: figure out issn.
+  def issn
+    "dunno"
+  end
+
+  # Todo: figure out govdocness.
+  def govdoc
+    "dunno"
+  end
+end
+
+# Parse any incoming files, output to stdout.
+if $0 == __FILE__
+  puts HTRecord.header_tsv
+  ExLibrisHoldingsXmlParser.new.main(ARGV) do |ht_record|
+    puts ht_record.to_tsv
+  end
+end