Checkpoint for #3725, #3634 & #2151

benwbrum · Aug 28, 2023 · 96b5842 · 96b5842
1 parent f9408ae
commit 96b5842
Show file tree

Hide file tree

Showing 2 changed files with 103 additions and 12 deletions.
diff --git a/app/helpers/export_helper.rb b/app/helpers/export_helper.rb
@@ -417,20 +417,24 @@ def xml_to_export_tei(xml_text, context, page_id = "", add_corrsp=false)
     # xml_text = titles_to_divs(xml_text, context)
     doc = REXML::Document.new(xml_text)
     #paras_string = ""
-
+    binding.pry
     my_display_html = ""
-    doc.elements.each_with_index("//p") do |e,i|
-      transform_links(e)
-      transform_expansions(e)
-      transform_regularizations(e)
-      transform_marginalia_and_catchwords(e)
-      transform_footnotes(e)
-      transform_lb(e)
-      e.add_attribute("xml:id", "#{page_id_to_xml_id(page_id, context.translation_mode)}P#{i}")
-      if add_corrsp
-        e.add_attribute("corresp", "#{page_id_to_xml_id(page_id, !context.translation_mode)}P#{i}")
+    tags = ['table', 'p']
+    tags.each do |tag|
+      doc.elements.each_with_index("//#{tag}") do |e,i|
+        transform_links(e)
+        transform_expansions(e)
+        transform_regularizations(e)
+        transform_marginalia_and_catchwords(e)
+        transform_footnotes(e)
+        transform_lb(e)
+        transform_tables(e)
+        e.add_attribute("xml:id", "#{page_id_to_xml_id(page_id, context.translation_mode)}P#{i}")
+        if add_corrsp
+          e.add_attribute("corresp", "#{page_id_to_xml_id(page_id, !context.translation_mode)}P#{i}")
+        end
+        my_display_html << e.to_s
       end
-      my_display_html << e.to_s
     end
 
     return my_display_html.gsub('<lb/>', "<lb/>\n").gsub('</p>', "\n</p>\n\n").gsub('<p>', "<p>\n").encode('utf-8')
@@ -501,6 +505,56 @@ def transform_marginalia_and_catchwords(p_element)
     end
   end
 
+  def transform_tables(p_element)
+    # convert HTML tables to TEI tables
+    p_element_string = p_element.to_s
+    p_element.elements.each("//table") do |e|
+      row_count = 0
+      max_column_count = 0
+      table = REXML::Element.new("table")
+      # does the table have a header?
+      if e.elements["thead"]
+        # convert the header into a row element with role="label"
+        header = REXML::Element.new("row")
+        header.add_attribute("role", "label")
+        e.elements.each("thead/tr/th") do |th|
+          # convert the th into a cell element
+          cell = REXML::Element.new("cell")
+          cell.add_attribute("role", "data")
+          th.children.each { |child| cell.add(child) }
+          header.add(cell)
+        end
+        table.add(header)
+      end
+      # now convert the body of the table
+      e.elements.each("tbody/tr") do |tr|
+        row = REXML::Element.new("row")
+        tr.elements.each("td") do |td|
+          cell = REXML::Element.new("cell")
+          cell.add_attribute("role", "data")
+          td.children.each { |child| cell.add(child) }
+          if cell.children.count > max_column_count
+            max_column_count = cell.children.count
+          end
+          row.add(cell)
+        end
+        row_count += 1
+        table.add(row)
+      end # end of tbody
+      table.add_attribute("rows", row_count)
+      table.add_attribute("cols", max_column_count)
+      e.replace_with(table)
+
+    end # end of tables
+    # now delete any lb elements from tables elements in the document
+    p_element.elements.each("//table") do |table|
+      table.elements.each("//lb") do |lb|
+        lb.remove
+      end
+    end
+
+  end
+
   def transform_footnotes(p_element)
     p_element.elements.each('//footnote') do |e|
       marker = e.attributes['marker']

diff --git a/app/helpers/tei_helper.rb b/app/helpers/tei_helper.rb
@@ -35,6 +35,43 @@ def xml_to_tei(xml_text, preserve_lb=true)
         e.replace_with(lb)
       end
     end
+    binding.pry
+    # convert HTML tables to TEI tables
+    doc.elements.each("//table") do |e|
+      table = REXML::Element.new("table")
+      # does the table have a header?
+      if e.elements["thead"]
+        # convert the header into a row element with role="label"
+        header = REXML::Element.new("row")
+        header.add_attribute("role", "label")
+        e.elements.each("thead/tr/th") do |th|
+          # convert the th into a cell element
+          cell = REXML::Element.new("cell")
+          cell.add_attribute("role", "data")
+          cell.add_child(th.children)
+        end
+        table.add(header)
+      end
+      # now convert the body of the table
+      e.elements.each("tbody/tr") do |tr|
+        row = REXML::Element.new("row")
+        tr.elements.each("td") do |td|
+          cell = REXML::Element.new("cell")
+          cell.add_attribute("role", "data")
+          cell.add_child(td.children)
+          row.add(cell)
+        end
+        table.add(row)
+      end # end of tbody
+      e.replace_with(table)
+    end # end of table
+    # now delete any lb elements from tables elements in the document
+    doc.elements.each("//table") do |table|
+      table.elements.each("//lb") do |lb|
+        lb.remove
+      end
+    end
+
     unless user_signed_in?
       doc.elements.each("//sensitive") do |e|
         e.replace_with(REXML::Comment.new("sensitive information suppressed"))