Ensure YAML is free of duplicate keys. (#15)

* Ensure YAML is free of duplicate keys.
* Refines one validation
* Fixes shebangs so Ruby version works correctly despite system ruby being installed
* Add Handbook glossary, and de-dupe entries. Fixes #10 and #12

bin/build

@@ -1,4 +1,4 @@
-#! /usr/bin/ruby
+#!/usr/bin/env ruby
 require_relative '../lib/markdownify'
 
 Markdownify.new(*ARGV).perform

bin/validate

@@ -1,4 +1,4 @@
-#! /usr/bin/ruby
+#!/usr/bin/env ruby
 require_relative '../lib/validate'
 
 GlossaryValidator.new(path: ARGV[0]).perform

lib/validate.rb

@@ -10,7 +10,7 @@ class GlossaryValidator
 
   include Common
 
-  attr_reader :data, :key
+  attr_reader :data, :key, :path
 
   # @param path [String | nil] path to glossary file
   def initialize(path: nil)
@@ -25,10 +25,30 @@ def initialize(path: nil)
   # Validate all entries
   def perform
     validator = EntryValidator.new(data[key])
+    report_duplicate_keys
     data[key].each do |entry|
       validator.validate([entry].to_h)
     end
   ensure
+    display_reminders(validator)
+  end
+
+  def report_duplicate_keys
+    file = File.read(path)
+    keys = file.scan(/^\s{2}(\S{1}.*):$/).flatten
+    keys.tally.detect do |k, count|
+      case count
+      when 0 then
+        # NO OP
+      when 1
+        # NO OP
+      else
+        raise DuplicateKeyError.new(k, count)
+      end
+    end
+  end
+
+  def display_reminders(validator)
     # Regardless of errors, show the reminders
     if validator.reminders.any?
       puts "\nReminders:"
@@ -37,10 +57,11 @@ def perform
       puts "\n"
     end
   end
-
 end
 
 
+
+
 # Validates an entry. Needs the entire glossary file to work,
 # as it needs to determine whether linked terms are present.
 class EntryValidator
@@ -139,8 +160,13 @@ def validate_acronym(acronym_data)
         raise TermNotFoundError.new(key, term)
       end
 
-      unless context[term.to_sym][:type] == "term"
+      case context[term.to_sym][:type]
+      when "acronym" then
         raise AcronymReferenceError.new(key, term)
+      when "term" then
+        # NO OP
+      else
+        raise EntryTypeError.new(term)
       end
     end
 
@@ -168,6 +194,29 @@ def assert_no_extra_keys(key, value_set, only_allow=[])
 
 end
 
+
+class DuplicateKeyError < StandardError
+  attr_reader :key, :count
+
+  def initialize(key, count)
+    @key = key
+    @count = count
+    super(message)
+  end
+
+  def message
+    <<~ERR
+
+
+      I found #{count} entries for "#{key}".
+
+      How can I know which one to use?
+
+      Please delete or combine until there is only one entry for "#{key}".
+    ERR
+  end
+end
+
 class MissingTermError < StandardError
   attr_reader :key, :term
 

test/test_glossary_dupe.yml

@@ -0,0 +1,13 @@
+entries:
+
+  FEMA:
+    type: acronym
+    term: Federal Emergency Management Agency
+
+  Federal Emergency Management Agency:
+    type: term
+    description: null
+
+  FEMA:
+    type: acronym
+    term: Federal Emergency Management Agency

test/validate_test.rb

@@ -4,9 +4,25 @@
 
 include Fixtures
 
+describe GlossaryValidator do
+  describe "with a valid glossary" do
+    it "raises no errors" do
+      subject = GlossaryValidator.new(path: './test/test_glossary.yml')
+      assert subject.perform
+    end
+  end
+
+  describe "with a glossary containing duplicate entries" do
+    it "raises an error" do
+      subject = GlossaryValidator.new(path: './test/test_glossary_dupe.yml')
+      assert_raises (DuplicateKeyError) { subject.perform }
+    end
+  end
+end
+
 describe EntryValidator do
   describe "with a malformed entry" do
-    it "is errors" do
+    it "errors" do
       subject = EntryValidator.new({})
       assert_raises (ArgumentError) { subject.validate({a: 1, b: 2}) }
     end