Merge pull request #270 from learningtapestry/hotfix-tables-with-headers

fixed parsing tables with thead

Author: shlag3n

lib/doc_template/tables/activity.rb

@@ -13,7 +13,7 @@ def parse(fragment, *args)
         idx = 0
         [].tap do |result|
           fragment.xpath(xpath_meta_headers, XpathFunctions.new).each do |el|
-            table = el.ancestors('table').first
+            table = self.class.flatten_table(el.ancestors('table').first)
             data = fetch table
 
             data = process_title(data)

lib/doc_template/tables/base.rb

@@ -12,6 +12,20 @@ def self.parse(fragment, *)
         new.parse(fragment, *)
       end
 
+      def self.flatten_table(table)
+        return unless table.present?
+
+        # remove blank tbody
+        table.xpath('.//tbody').each { |tbody| tbody.remove if tbody.text.blank? }
+        # move data from thead to tbody (first children)
+        table.xpath('.//thead').each do |thead|
+          tbody = table.at_xpath('.//tbody').presence || table.add_child('<tbody></tbody>').first
+          thead.children.reverse.each { |child| tbody.prepend_child child }
+          thead.remove
+        end
+        table
+      end
+
       def initialize
         @data = {}
         @errors = []
@@ -65,7 +79,8 @@ def parse(fragment, *args)
 
         # get the table
         table_key_cell = fragment.at_xpath("table//tr[1]/td[1][contains(., '#{self.class::HEADER_LABEL}')]")
-        table = table_key_cell&.ancestors('table')&.first
+        # flatten table to simple structure with tbody only
+        table = self.class.flatten_table(table_key_cell&.ancestors('table')&.first)
         @table_exists = table.present?
         return self unless @table_exists
 

lib/doc_template/tables/section.rb

@@ -16,7 +16,7 @@ def parse(fragment, *args)
 
         [].tap do |result|
           section_tables.each do |el|
-            table = el.ancestors('table').first
+            table = self.class.flatten_table(el.ancestors('table').first)
             data = fetch table
 
             value = data['section-title'].parameterize

lib/doc_template/tables/target.rb

@@ -9,7 +9,7 @@ class Target < Base
       def parse(fragment, *_args)
         path = ".//table/*/tr[1]/td[1][case_insensitive_contains(.//*/text(), '#{HEADER_LABEL_PIECE}')]"
         return unless (element = fragment.at_xpath path, XpathFunctions.new)
-        return unless (table = element.ancestors('table').first)
+        return unless (table = self.class.flatten_table(element.ancestors('table').first))
 
         @target_data = fetch table
 

lib/doc_template/tags/table_tag.rb

@@ -4,7 +4,7 @@ module DocTemplate
   module Tags
     class TableTag < BaseTag
       def parse(node, opts = {})
-        unless (table = node.ancestors('table').first)
+        unless (table = ::DocTemplate::Tables::Base.flatten_table(node.ancestors('table').first))
           raise ::Lcms::Engine::DocumentError, "Tag #{self.class::TAG_NAME.upcase} placed outside table"
         end
 

sig/lib/doc_template/tables/base.rbs

@@ -14,6 +14,7 @@ module DocTemplate
       attr_reader data: untyped
 
       def self.parse: (untyped fragment, *untyped) -> untyped
+      def self.flatten_table: (untyped table) -> untyped
 
       def initialize: () -> void
       def collect_and_render_tags: (untyped data, untyped fields, ?::Hash[untyped, untyped] opts) -> untyped

spec/fixtures/tables/material-metadata-theader.html

@@ -0,0 +1,28 @@
+# frozen_string_literal: true
+
+require 'rails_helper'
+
+describe DocTemplate::Tables::MaterialMetadata do
+  let(:table) { described_class.new }
+
+  describe '#parse' do
+    let(:html) { Lcms::Engine::HtmlSanitizer.sanitize(data) }
+    let(:fragment) { Nokogiri::HTML.fragment html }
+
+    subject { table.parse fragment }
+
+    context 'table with headers' do
+      let(:data) { file_fixture('tables/material-metadata-theader.html').read }
+
+      include_examples 'removes metadata table'
+
+      it 'fetching all fields' do
+        expect(subject.data['type']).to eq 'AK'
+        expect(subject.data['title']).to eq 'Ashoka Reading (Teacher Version)'
+        expect(subject.data['grade']).to eq '7'
+        expect(subject.data['unit']).to eq '1'
+        expect(subject.data['cluster']).to eq '4'
+      end
+    end
+  end
+end

spec/lib/doc_template/tables/material_metadata_spec.rb

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+
+require 'rails_helper'
+
+describe DocTemplate::Tags::ImageTag do
+  let(:node) do
+    html = Nokogiri::HTML original_content
+    html.at_xpath('*//table/thead/tr[1]/td[1]')
+  end
+  let(:metadata) { { 'grade' => '5', 'unit' => '1', 'subject' => 'Math' } }
+  let(:opts) { { value: 'example_image', metadata: } }
+  let(:original_content) do
+    <<-HTML
+      <table>
+        <thead>
+          <tr><td>[#{described_class::TAG_NAME}]</td></tr>
+          <tr><td>Caption text</td></tr>
+        </thead>
+      </table>
+    HTML
+  end
+  let(:tag) { described_class.new }
+  subject { tag.parse(node, opts) }
+
+  describe '#parse_table' do
+    it 'removes original node' do
+      expect(subject.content).to_not include("[#{described_class::TAG_NAME}]")
+    end
+
+    it 'substitues tag with image with caption' do
+      expect(subject.content).to include('figcaption>Caption text</figcaption>')
+    end
+  end
+end