Add New Glossary Page And Developer Search Feature

This commit adds the code necessary to generate a new
en/developre-glossary page with entry pages in the en/glossary/
directory, e.g. en/glossary/51-percent-attack.

The glossary page and the individual term pages feature a JavaScript
search engine (no CGI) for just glossary terms.  This search box has
also been added to the following pages:

    * Developer Documentation (the index page)
    * Developer Guide
    * Developer Reference
    * Developer Examples

The search box requires the following MIT-licensed libraries:

    * JQuery
    * JQuery UI
    * JQuery UI CSS stylesheet

These allow our JS code to be almost trivially simple in js/devsearch.js

This commit adds only code.  Actual glossary entry data will be added in
a subsequent commit.

_plugins/autocrossref.rb

@@ -35,12 +35,37 @@ require 'yaml'
 
       ## Workaround for inconsistent relative directory
       path = File.expand_path(File.dirname(__FILE__)) + "/.."
-      ## Load terms from file
+      ## Load terms from file only if we haven't loaded them before
       site = context.registers[:site].config
-      if !site.has_key?("crossref")
-        site['crossref'] = YAML.load_file(path + "/_autocrossref.yaml")
+      if !site.has_key?("crossref_loaded")
+
+        ## Load refs from file and then downcase them all so we can
+        ## easily detect when we define xrefs more than once
+        mixed_case_refs = YAML.load_file(path + "/_autocrossref.yaml")
+        unvalidated_refs = Hash.new
+        mixed_case_refs.each { |key, value|
+          unvalidated_refs[key.to_s.downcase] = value.to_s.downcase
+        }
+
+        if site.has_key?("crossref")
+          ## We already have refs loaded, so merge
+          site['crossref'].merge!(unvalidated_refs) {
+            |key, old_value, new_value|
+
+            if old_value != new_value
+              abort("Error: autocrossref key '#{key}' wants to point to both '#{old_value}' and '#{new_value}'")
+            end
+
+            new_value
+          }
+        else
+          ## We don't have refs loaded yet, so copy
+          site['crossref'] = unvalidated_refs
+        end
+        site['crossref_loaded'] = true
       end
 
+
       ## Sort terms by reverse length, so longest matches get linked
       ## first (e.g. "block chain" before "block"). Otherwise short
       ## terms would get linked first and there'd be nothing for long
@@ -76,7 +101,7 @@ require 'yaml'
             (?!\w)  ## Don't match inside words
             (?!`)   ## Don't match strings ending with a tic, unless the xref itself ends with a tic
           /xmi) {|s|
-              if term[1] == "DO NOT AUTOCROSSREF"
+              if term[1] == "do not autocrossref"
                   s.gsub(/( |$)/, "<!--noref-->\\&")
               else
                   "[#{s}][#{term[1]}]{:.auto-link}"

_plugins/glossary.rb

@@ -0,0 +1,131 @@
+# This file is licensed under the MIT License (MIT) available on
+# http://opensource.org/licenses/MIT.
+
+require 'yaml'
+
+module Jekyll
+
+  class GlossaryPage < Page
+
+    def initialize(site, base, lang, srcdir, src, output_directory)
+      @site = site
+      @base = base
+      @dir = '/' + output_directory
+
+      ## Output file is the source file base name plus .md for Markdown
+      ## (converted later to HTML)
+      output_file = src.split('.')[0] + ".md"
+
+      ## Read in the file's YAML header
+      self.read_yaml(File.join(base, srcdir), src)
+
+      ## Pass in the full path to enable edit-on-github links
+      self.data["filename"] = srcdir + '/' + src
+
+      ## Page Title: <title> - Bitcoin Glossary
+      self.data["title"] = self.data["required"]["title_max_40_characters_no_formatting"] + " - Bitcoin Glossary"
+
+      ## Output file is v<version>.md (converted later to HTML)
+      @name = output_file
+      self.process(output_file)
+
+      ## Output full path
+      output_full_path = dir + "/" + output_file.gsub('.md', '')
+
+      self.data['layout'] = 'glossary_entry'
+      self.data['category'] = 'glossary_entry'
+
+      ## Combine required (displayed) and optional (non-displayed)
+      ## synonyms into an array
+      if self.data["optional"]["synonyms_and_pluralizations_not_shown_in_glossary"].nil?
+        mixed_case_terms = self.data["required"]["synonyms_shown_in_glossary_capitalize_first_letter"]
+      else
+        mixed_case_terms = self.data["required"]["synonyms_shown_in_glossary_capitalize_first_letter"] +\
+        self.data["optional"]["synonyms_and_pluralizations_not_shown_in_glossary"]
+      end
+
+      ## Downcase all terms so we can easily detect when we create
+      ## duplicated autocrossreference links
+      terms = Array.new
+      mixed_case_terms.each { |term|
+        terms.push(term.downcase)
+      }
+
+      ## Add all synonyms to the autocrossref hash table for automatic linking
+      site.config["crossref"] = site.config["crossref"] ? site.config["crossref"] : {}
+      for term in terms do
+        site.config["crossref"].merge!({ term => output_full_path }) {
+            |key, old_value, new_value|
+
+            if old_value != new_value
+              abort("Error: autocrossref key '#{key}' wants to point to both '#{old_value}' and '#{new_value}'")
+            end
+
+            new_value
+        }
+      end
+
+      ## Create a newline-containing Markdown-formatted string that
+      ## includes links for all of our autocrossref-created synonyms.
+      ## TODO: this should probably be done as an `include`-able
+      ## template for proper division between the logic and
+      ## presentation layers
+      site.config["glossary_links"] = site.config["glossary_links"] ? site.config["glossary_links"] : ''
+      site.config["glossary_links"] \
+        += "[" \
+        + output_full_path \
+        + "]: " \
+        + output_full_path \
+        + ' "' \
+        + self.data["required"]["summary_max_255_characters_no_formatting"].chomp() \
+        + '"' \
+        + "\n"
+
+      ## Add only shown synonyms to the glossary hash-tables-inside-sorted-array
+      ## for use in the search box and on the master listing page
+      site.config["devsearches"]["Glossary"] = site.config["devsearches"]["Glossary"] ? site.config["devsearches"]["Glossary"] : []
+      for term in self.data["required"]["synonyms_shown_in_glossary_capitalize_first_letter"] do
+        site.config["devsearches"]["Glossary"].unshift({ term => output_full_path })
+      end
+
+      ## Sort the shown synonyms array alphabetically (case
+      ## insensitive). We have to do this here because the version of
+      ## Jekyll/Liquid we use does not provide the ability to sort
+      ## arrays when the template is compiled. Higher version of Jekyll
+      ## do support this feature, so if we upgrade to Jekyll 2.2 or
+      ## higher, look at doing this at template time to save CPU cycles
+      ## and increase flexibility
+      site.config["devsearches"]["Glossary"].sort_by!{|hash|
+          hash.to_s.downcase.gsub(/"=>.*/,'')
+      }
+
+    end
+  end
+
+  class GlossaryPageGenerator < Generator
+    def generate(site)
+
+      #Do nothing if plugin is disabled
+      if !ENV['ENABLED_PLUGINS'].nil? and ENV['ENABLED_PLUGINS'].index('glossary').nil?
+        print 'Glossary disabled' + "\n"
+        return
+      end
+
+      glossary_dir='_data/glossary/en'
+
+      #generate each release based on templates
+      Dir.foreach(glossary_dir) do |file|
+        next if file == '.' or file == '..'
+        lang = 'en'
+        src = file
+        srcdir = '_data/glossary/en'
+        output_directory = lang + '/glossary/'
+        site.pages << GlossaryPage.new(site, site.source, lang, glossary_dir, src, output_directory)
+      end
+      #TODO releases are only generated for English language,
+      #but they could also be translated at some point. They would however
+      #need to fallback to English when no translation is available.
+    end
+  end
+
+end