jekyll_search 0.0.2 → 0.0.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 266d467d948e0f06e9b67d80b4460e39127e57e1
-  data.tar.gz: 42098ce4e4628cf9a2ac55ea3dad5fe5f0871f11
+  metadata.gz: e903bb117234165daca55baee25cc4a352e5cd91
+  data.tar.gz: db9eb1cf7534ec9929057194af1e0248b52dc243
 SHA512:
-  metadata.gz: 7a74dead448ac960a3c32e6546b181fe7561d9c1a57ca6ddcdbca4104378f7dae2d2ef0fe7ff2482adb11d77b0b72709d57bf8a0947c2cce69777931780f8f55
-  data.tar.gz: 4a08fdc4d9474469f2e0a8f1c4cb1c30146ae8bf83dc499498b9bfd4b7e944e4607d1672c08a640880903fc7a1712a8b13517b87e958b25801f1876026cc8369
+  metadata.gz: 7e6f6263cb8391612c44c4f81b1cb24285f9bdf68b910ae3f2bbe6f2d71eccede16a95cf61648b3e576fc3368766caaf7fe70706a75277e2d2ca2b72486a200d
+  data.tar.gz: 84529eaef4b0b389bb13bfb1d11ed6ad3bd22e997bef30c59eef4ff18c4706d593ae04f28934939cc568d5c22e6716b6d9abe6244db9b2049f795560187b0d31

data/lib/jekyll_search/html_processor.rb

@@ -0,0 +1,42 @@
+require 'loofah'
+require 'loofah/helpers'
+
+module JekyllSearch
+  class HtmlProcessor
+    def self.strip_html(input)
+      strip_pre = Loofah::Scrubber.new do |node|
+        if node.name == 'pre'
+          node.remove
+          Loofah::Scrubber::STOP
+        end
+      end
+
+      Loofah.fragment(input).
+          scrub!(:prune).
+          scrub!(strip_pre).
+          to_text.
+          gsub(/([\r\n\t\s]+)/, ' ').strip
+    end
+
+    def self.detect_sections(input)
+      result = []
+
+      current = { :id => nil, :title => nil, :content => '' }
+      Loofah.fragment(input).children.each { |node|
+        if node.name =~ /^h\d$/
+          result << current
+          current = { :id => nil, :title => nil, :content => '' }
+          current[:id] = if node.has_attribute?('id') then node.attribute('id').value else nil end
+          current[:title] = node.text
+        else
+          current[:content] += node.to_html
+        end
+      }
+      if current[:title] != nil or current[:content] != ''
+        result << current
+      end
+
+      result
+    end
+  end
+end

data/lib/jekyll_search/version.rb

@@ -1,3 +1,3 @@
 module JekyllSearch
-  VERSION = "0.0.2"
+  VERSION = "0.0.3"
 end

data/lib/jekyll_search.rb

@@ -2,6 +2,7 @@ require 'jekyll'
 require 'elasticsearch'
 require 'loofah'
 require 'loofah/helpers'
+require 'jekyll_search/html_processor'
 
 module Jekyll
   module Commands
@@ -35,29 +36,24 @@ module Jekyll
               select { |p| p.data['searchable'].nil? or p.data['searchable'] != false }
 
           for page in pages
-            body = {
+            page_body = {
               url: site.baseurl + page.url,
               title: page.data['title'],
-              content: clean_content(page.content)
+              content: JekyllSearch::HtmlProcessor.strip_html(page.content)
             }
 
-            client.index index: settings['index']['name'], type: 'page', body: body
-          end
-        end
+            client.index index: settings['index']['name'], type: 'page', body: page_body
 
-        def clean_content(dirty)
-          strip_pre = Loofah::Scrubber.new do |node|
-            if node.name == 'pre'
-              node.remove
-              Loofah::Scrubber::STOP
+            for section in JekyllSearch::HtmlProcessor.detect_sections(page.content)
+              section_body = {
+                url: if section[:id] != nil then site.baseurl + page.url + '#' + section[:id] else site.baseurl + page.url end,
+                title: if section[:title] != nil then section[:title] else page.data['title'] end,
+                content: JekyllSearch::HtmlProcessor.strip_html(section[:content])
+              }
+
+              client.index index: settings['index']['name'], type: 'section', body: section_body
             end
           end
-
-          Loofah.fragment(dirty).
-              scrub!(:prune).
-              scrub!(strip_pre).
-              to_text.
-              gsub(/([\r\n\t\s]+)/, ' ').strip
         end
 
         def create_index(client, settings)

data/spec/lib/jekyll_search/html_processor_spec.rb

@@ -0,0 +1,23 @@
+require 'jekyll_search/html_processor'
+
+RSpec.describe JekyllSearch::HtmlProcessor do
+  it 'stripes html' do
+    expect(JekyllSearch::HtmlProcessor.strip_html('test')).to eq 'test'
+    expect(JekyllSearch::HtmlProcessor.strip_html("<h1>title</h1>\n<p>text</p>")).to eq 'title text'
+  end
+
+  it 'detects sections' do
+    expect(JekyllSearch::HtmlProcessor.detect_sections('test')).to eq [
+        { :id => nil, :title => nil, :content => 'test'}
+    ]
+  end
+
+  it 'detects sections' do
+    expect(JekyllSearch::HtmlProcessor.detect_sections("foo<h1>first</h1>bar <p>\npara\n</p>\n\n<H4 ID=\"sec\">sec<span>ond</span></H4>apple<h2>third</h2>")).to eq [
+        { :id => nil, :title => nil, :content => "foo"},
+        { :id => nil, :title => 'first', :content => "bar <p>\npara\n</p>\n\n"},
+        { :id => 'sec', :title => 'second', :content => "apple" },
+        { :id => nil, :title => 'third', :content => '' }
+    ]
+  end
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: jekyll_search
 version: !ruby/object:Gem::Version
-  version: 0.0.2
+  version: 0.0.3
 platform: ruby
 authors:
 - Christian Hoffmeister
@@ -110,8 +110,9 @@ files:
 - Rakefile
 - jekyll_search.gemspec
 - lib/jekyll_search.rb
+- lib/jekyll_search/html_processor.rb
 - lib/jekyll_search/version.rb
-- spec/lib/jekyll_search_spec.rb
+- spec/lib/jekyll_search/html_processor_spec.rb
 - spec/spec_helper.rb
 homepage: https://github.com/choffmeister/jekyll_search
 licenses:
@@ -138,5 +139,5 @@ signing_key:
 specification_version: 4
 summary: An Elasticsearch full text search index generator for Jekyll.
 test_files:
-- spec/lib/jekyll_search_spec.rb
+- spec/lib/jekyll_search/html_processor_spec.rb
 - spec/spec_helper.rb

data/spec/lib/jekyll_search_spec.rb

@@ -1,5 +0,0 @@
-require 'jekyll_search'
-
-RSpec.describe Jekyll::Commands::Index do
-  it 'works'
-end