From 16d851768b7014bb286ab74b9c4069e40d625d2b Mon Sep 17 00:00:00 2001
From: Miki <mehranb@amazon.com>
Date: Fri, 19 Nov 2021 16:22:49 -0800
Subject: [PATCH] Add indexing of documentation for custom-search

Signed-off-by: Miki <mehranb@amazon.com>
---
 _plugins/search-indexer.rb | 118 +++++++++++++++++++++++++++++++++++++
 1 file changed, 118 insertions(+)
 create mode 100644 _plugins/search-indexer.rb

diff --git a/_plugins/search-indexer.rb b/_plugins/search-indexer.rb
new file mode 100644
index 00000000..8b938166
--- /dev/null
+++ b/_plugins/search-indexer.rb
@@ -0,0 +1,118 @@
+# frozen_string_literal: true
+
+require "jekyll/hooks"
+require "jekyll/document"
+require "json"
+
+##
+# This singleton facilitates production of an indexable JSON representation of the content to populate a data source
+# to provide search functionality.
+
+module Jekyll::ContentIndexer
+
+  ##
+  # The collection that will get stores as the output
+
+  @data = []
+
+  ##
+  # Pattern to identify documents that should be excluded based on their URL
+
+  @excluded_paths = /\.(css|js|json|map|xml|txt|yml)$/i.freeze
+
+  ##
+  # Pattern to identify block HTML tags (not comprehensive)
+
+  @html_block_tags = /\s*<[?\/]?(article|blockquote|d[dlt]|div|fieldset|form|h|li|main|nav|[ou]l|p|section|table|t[rd]).*?>\s*/im.freeze
+
+  ##
+  # Pattern to identify certain HTML tags whose content should be excluded from indexing
+
+  @html_excluded_tags = /\s*<(head|style|script|h1).*?>.*?<\/\1>/im.freeze
+
+  ##
+  # Initializes the singleton by recording the site
+
+  def self.init(site)
+    @site = site
+  end
+
+  ##
+  # Processes a Document or Page and adds it to the collection
+
+  def self.add(page)
+    return if @excluded_paths.match(page.url)
+
+    content = page.content
+                  .gsub(@html_excluded_tags, ' ')             # Strip certain HTML blocks
+                  .gsub(@html_block_tags, "\n")               # Strip some block HTML tags, replacing with newline
+                  .gsub(/\s*<[?\/!]?[a-z]+.*?>\s*/im, ' ')    # Strip all remaining HTML tags
+                  .gsub(/\s*[\r\n]+\s*/, "\n")                # Clean line-breaks
+                  .gsub(/\s{2,}/, ' ')                        # Trim long spaces
+                  .gsub(/\s+([.:;,)!\]?])/, '\1')             # Remove spaces before some punctuations
+                  .strip                                      # Trim leading and tailing whitespaces
+
+    return if content.empty?
+
+    url = @site.config["baseurl"] + page.url
+
+    # Produce a breadcrumb
+    ancestors = []
+    if page.instance_of?(Jekyll::Document)
+      ancestors.push(@site.config.dig("just_the_docs", "collections", page.collection&.label, "name"))
+    end
+
+    ancestors.push(page.data["grand_parent"]) unless
+      page.data["grand_parent"].nil? ||
+      page.data["grand_parent"]&.empty? ||
+      ancestors.include?(page.data["grand_parent"])     # Make sure collection name is not added
+
+    ancestors.push(page.data["parent"]) unless
+      page.data["parent"].nil? ||
+        page.data["parent"]&.empty? ||
+        ancestors.include?(page.data["parent"])         # Make sure collection name is not added
+
+    data = {
+      url: url,
+      title: page.data["title"],
+      content: content,
+      ancestors: ancestors,
+      type: "DOCS"
+    }
+
+    @data.push(data)
+  end
+
+  ##
+  # Saves the collection as a JSON file
+
+  def self.save
+    File.open(File.join(@site.config["destination"], "search-index.json"), 'w') do |f|
+      f.puts JSON.pretty_generate(@data)
+    end
+  end
+end
+
+# Before any Document or Page is processed, initialize the ContentIndexer
+
+Jekyll::Hooks.register :site, :pre_render do |site|
+  Jekyll::ContentIndexer.init(site)
+end
+
+# Process a Page as soon as its content is ready
+
+Jekyll::Hooks.register :pages, :post_convert do |page|
+  Jekyll::ContentIndexer.add(page)
+end
+
+# Process a Document as soon as its content is ready
+
+Jekyll::Hooks.register :documents, :post_convert do |document|
+  Jekyll::ContentIndexer.add(document)
+end
+
+# Save the produced collection after Jekyll is done writing all its stuff
+
+Jekyll::Hooks.register :site, :post_write do |_|
+  Jekyll::ContentIndexer.save()
+end
\ No newline at end of file