Script to show content items with missing topics

bin/missing_topics_and_browse

@@ -0,0 +1,24 @@
+#!/usr/bin/env ruby
+
+require 'optparse'
+require './lib/database'
+
+options = {format: :text}
+
+parser = OptionParser.new do |opts|
+  opts.banner = "Usage: example.rb [options]"
+
+  opts.on("-fFORMAT", [:csv, :text], "--format=FORMAT") do |form|
+    options[:format] = form
+
+  end
+end
+
+begin
+  parser.parse!
+rescue OptionParser::InvalidOption => error
+  die error.message, :usage => true
+end
+
+database = Database.new
+database.find_missing_topics_and_browse!(output: options[:format])

lib/database.rb

@@ -1,3 +1,4 @@
+require 'csv'
 require 'pg'
 
 class Database
@@ -52,7 +53,8 @@ def compare!
   end
 
   # Identify rummager content that is not in the publishing api.
-  # This content is ignored in other queries.
+  # If the target of a link is not in the publishing api it gets
+  # ignored by the other queries.
   def find_unmatched_base_paths!
     query = <<-SQL
       SELECT DISTINCT base_path FROM rummager
@@ -108,6 +110,52 @@ def find_missing_publishing_api_link_types!(publishing_app:)
     puts "#{results.ntuples} missing #{publishing_app} links found"
   end
 
+  # All items with missing topics or browse pages
+  def find_missing_topics_and_browse!(output:)
+    query = <<-SQL
+      WITH missing_links as (
+        SELECT
+          base_path, link_type
+        FROM rummager
+        WHERE link_type in ('topics', 'mainstream_browse_pages')
+
+        EXCEPT
+
+        SELECT
+          base_path,link_type
+        FROM publishing_api
+        WHERE link_type in ('topics', 'mainstream_browse_pages')
+      )
+
+      SELECT base_path, link_type, format, publishing_app
+      FROM
+        missing_links
+      JOIN
+        api_content USING(base_path)
+    SQL
+
+    results = @connection.exec(query)
+
+    if (output == :csv)
+      CSV do |csv|
+        csv << %w(base_path link_type format publishing_app)
+
+        results.each_row do |row|
+          csv << row
+        end
+      end
+    elsif (output == :text)
+      results.each_row do |row|
+        %w(base_path link_type format publishing_app).zip(row).each do |name, value|
+          puts ("%-20s " % name) + value
+        end
+        puts '------------------------------'
+      end
+
+      puts "#{results.ntuples} rows found"
+    end
+  end
+
   def summarise_missing_publishing_api_link_types!
     query = <<-SQL
       WITH missing_links as (
@@ -138,5 +186,4 @@ def summarise_missing_publishing_api_link_types!
     end
 
   end
-
 end