fix(html table cell): better support for html lists

lib/md_to_pdf/elements/html.rb

@@ -107,14 +107,11 @@ def data_inlinehtml_tag(tag, node, opts)
         when 'img'
           result.push({ image: sub.attr('src') })
         when 'input'
-          if sub.attr('type') == 'checkbox'
-            checkbox = opt_task_list_point_sign(@styles.task_list_point, sub.attributes.key?('checked'))
-            result.push(text_hash(checkbox, current_opts))
-          else
-            data_array, current_opts = handle_unknown_inline_html_tag(sub, node, current_opts)
-            result.concat(data_array)
-          end
-        when 'ul', 'ol', 'li', 'label', 'p'
+          data_array, current_opts = handle_unknown_inline_html_tag(sub, node, current_opts)
+          result.concat(data_array)
+        when 'ul', 'ol'
+          result.concat(data_inlinehtml_list_tag(sub, node, opts))
+        when 'label', 'p', 'li'
           result.concat(data_inlinehtml_tag(sub, node, opts))
         when 'br'
           result.push(text_hash_raw("\n", current_opts))
@@ -126,6 +123,25 @@ def data_inlinehtml_tag(tag, node, opts)
       result
     end
 
+    def data_inlinehtml_list_tag(tag, node, opts)
+      result = []
+      points, level, _list_style, content_opts = data_html_list(tag, node, opts)
+      result.push(text_hash_raw("\n", content_opts).merge({ list_level: level, list_indent: 0 })) if level > 1
+      points.each do |point|
+        data = data_inlinehtml_tag(point[:tag], node, content_opts)
+        data.push(text_hash_raw("\n", content_opts).merge({ list_entry_type: 'end' }))
+        data[0][:list_entry_type] = 'first' unless data.empty?
+        data.unshift(text_hash(point[:bullet], point[:opts]).merge({ list_entry_type: 'bullet' }))
+        data.each do |item|
+          item[:list_level] = level if item[:list_level].nil?
+          item[:list_indent] = point[:width] if item[:list_indent].nil?
+          item[:list_indent_space] = point[:space_width] if item[:list_indent_space].nil?
+        end
+        result.concat(data)
+      end
+      result
+    end
+
     def collect_html_table_tag_rows(tag, table_font_opts, opts)
       rows = []
       tag.children.each do |sub|
@@ -257,16 +273,51 @@ def collect_html_table_tag_cell(tag, opts)
       cell_data
     end
 
+    def space_stuffing(width, space_width)
+      amount = (width / space_width).truncate
+      return '' if amount < 1
+
+      Prawn::Text::NBSP * amount
+    end
+
+    def indent_html_table_list_items(cell_data)
+      cell_data.each do |item|
+        next if item[:list_level].nil?
+
+        # Note: There is no settings for paddings of text fragments in Prawn::Table
+        # so as a workaround the lists are stuffed with spaces, which is of course not pixel perfect
+
+        # first indenting with spaces of multiline list items
+        # * item
+        #   multiline item
+        #   multiline item
+        if item[:list_entry_type].nil? && item[:text] != "\n"
+          item[:text] = "#{space_stuffing(item[:list_indent], item[:list_indent_space])}#{item[:text]}"
+        end
+
+        # second indenting of nested lists
+        # * item
+        #   multiline item
+        #   * sub list item
+        #   * sub list item
+        #     sub list multiline item
+        if item[:list_level] > 1 && (item[:list_entry_type].nil? || item[:list_entry_type] == 'bullet') && item[:text] != "\n"
+          item[:text] = "#{space_stuffing(item[:list_indent], item[:list_indent_space])}#{item[:text]}"
+        end
+      end
+      cell_data
+    end
+
     def collect_html_table_tag_row(tag, table_font_opts, opts)
       cells = []
       tag.children.each do |sub|
         case sub.name
         when 'th'
           cell_data = collect_html_table_tag_cell(sub, opts.merge(table_font_opts[:header]))
-          cells.push(cell_data)
+          cells.push(indent_html_table_list_items(cell_data))
         when 'td'
           cell_data = collect_html_table_tag_cell(sub, opts.merge(table_font_opts[:cell]))
-          cells.push(cell_data)
+          cells.push(indent_html_table_list_items(cell_data))
         end
       end
       cells

lib/md_to_pdf/elements/list.rb

@@ -25,20 +25,25 @@ def draw_list(node, opts)
       end
     end
 
-    def draw_html_list_tag(tag, node, opts)
+    def data_html_list(tag, node, opts)
       level = count_list_level_html(tag)
       is_ordered = tag.name.downcase == 'ol'
-      list_style = list_style(level, is_ordered, false)
-      padding = level == 1 ? opts_padding(list_style) : {}
-      # point_inline = opt_list_point_inline?(list_style)
+      is_task_list = !tag.search("input[type=checkbox]").first.nil?
+      list_style = list_style(level, is_ordered, is_task_list)
       content_opts = opts_font(list_style, opts).merge(
         {
           force_paragraph: {
             bottom_padding: opt_list_point_spacing(list_style)
           }
         }
       )
-      points = collect_points_html(tag, level, is_ordered, content_opts)
+      points = collect_points_html(tag, level, is_ordered, is_task_list, content_opts)
+      [points, level, list_style, content_opts]
+    end
+
+    def draw_html_list_tag(tag, node, opts)
+      points, level, list_style, content_opts = data_html_list(tag, node, opts)
+      padding = level == 1 ? opts_padding(list_style) : {}
       with_block_padding_all(padding) do
         draw_points_html(points, node, content_opts)
       end
@@ -82,22 +87,31 @@ def draw_inline_points(points, opts)
       end
     end
 
-    def collect_points_html(tag, level, is_ordered, content_opts)
-      point_style = list_point_style(level, is_ordered, false)
+    def collect_points_html(tag, level, is_ordered, is_task_list, content_opts)
+      point_style = list_point_style(level, is_ordered, is_task_list)
       auto_span = opt_list_point_spanning?(point_style)
       bullet_opts = opts_font(point_style, content_opts)
       spacing = opt_list_point_spacing(point_style)
       points = []
       index = 1
       tag.children.each do |sub|
         if sub.name.downcase == 'li'
-          bullet = list_bullet(point_style, is_ordered, false, index, false)
+          checked = false
+          if is_task_list
+            checked_box_tag = tag.search("input[type=checkbox]").first
+            unless checked_box_tag.nil?
+              checked = checked_box_tag.attributes.key?('checked')
+              checked_box_tag.remove
+            end
+          end
+          bullet = list_bullet(point_style, is_ordered, is_task_list, index, checked)
           bullet_width = measure_text_width(bullet, bullet_opts) + spacing
-          points.push({ tag: sub, bullet: bullet, width: bullet_width, opts: bullet_opts })
+          space_width = measure_text_width(Prawn::Text::NBSP, bullet_opts)
+          points.push({ tag: sub, bullet: bullet, width: bullet_width, space_width: space_width, opts: bullet_opts })
           index += 1
         end
       end
-      if auto_span
+      if auto_span || is_task_list
         max_span = max_point_width(points)
         points.each { |point| point[:width] = max_span }
       end

spec/fixtures/list/tasklist_html.md

@@ -0,0 +1 @@
+<ul><li><input type="checkbox" disabled="disabled">aha</li><li><input type="checkbox" checked="checked">oho</li><li><input type="checkbox" disabled="disabled">ehe</li></ul>

spec/fixtures/table/linefeed_in_cell.md

@@ -0,0 +1,4 @@
+| Description Column 1          | 
+|-------------------------------|
+| Lorem ipsum dolor<br>sit amet | 
+| Consectetur<br/>adipiscing elit | 

spec/fixtures/table/list_in_cell.md

@@ -0,0 +1 @@
+<figure><table><tbody><tr><td><ul><li>test1</li><li>test2<br>test2cont</li><li>test3<ul><li>test3.1</li><li>test3.2<br>test3.2cont</li><li>test3.3</li></ul></li></ul></td><td><ul><li><label><input type="checkbox" disabled="disabled">aha</label></li><li><label><input type="checkbox" checked="checked">oho</label></li><li><label><input type="checkbox" disabled="disabled">ehe</label></li></ul></td><td><ol><li>wooo</li><li>waaa</li><li>wiiiii</li></ol></td></tr></tbody></table></figure>

spec/markdown_to_pdf/list_spec.rb

@@ -152,6 +152,17 @@
                  { x: 76.032, y: 553.176, text: "three" }])
   end
 
+  it 'creates task lists by html' do
+    generator.parse_file('list/tasklist_html.md', {})
+    expect_pdf([
+                 { x: 36.0, y: 747.384, text: "[ ]" },
+                 { x: 52.008, y: 747.384, text: "aha" },
+                 { x: 36.0, y: 733.512, text: "[x]" },
+                 { x: 52.008, y: 733.512, text: "oho" },
+                 { x: 36.0, y: 719.64, text: "[ ]" },
+                 { x: 52.008, y: 719.64, text: "ehe" }])
+  end
+
   it 'creates an ordered list with correct numbers inline' do
     generator.parse_file('list/inline.md', { ordered_list: { point_inline: true } })
     expect_pdf([

spec/markdown_to_pdf/table_spec.rb

@@ -89,14 +89,14 @@
                  { x: 421.71429, y: 744.756, text: "Header 5" },
                  { x: 498.85714, y: 744.756, text: "Header 6" },
                  { x: 36.0, y: 730.884, text: "Entry 1" },
-                 { x: 344.57143, y: 730.884, text: "[x]" },
-                 { x: 498.85714, y: 730.884, text: "[x]" },
+                 { x: 344.57143, y: 730.884, text: "[x] " },
+                 { x: 498.85714, y: 730.884, text: "[x] " },
                  { x: 36.0, y: 717.012, text: "Entry 2" },
-                 { x: 344.57143, y: 717.012, text: "[x]" },
-                 { x: 421.71429, y: 717.012, text: "[x]" },
+                 { x: 344.57143, y: 717.012, text: "[x] " },
+                 { x: 421.71429, y: 717.012, text: "[x] " },
                  { x: 36.0, y: 703.14, text: "Entry 3" },
                  { x: 36.0, y: 689.268, text: "Entry 4" },
-                 { x: 113.14286, y: 689.268, text: "[x]" }])
+                 { x: 113.14286, y: 689.268, text: "[x] " }])
   end
 
   it 'creates a table without bad wrapping with doc font style' do
@@ -163,4 +163,33 @@
                  { x: 36.0, y: 689.268, text: "Color empty cells to the" },
                  { x: 36.0, y: 675.396, text: "right" }])
   end
+
+  it 'creates a html table with linefeeds inside' do
+    generator.parse_file('table/linefeed_in_cell.md')
+    expect_pdf([
+                 { x: 36.0, y: 744.756, text: "Description Column 1" },
+                 { x: 36.0, y: 730.884, text: "Lorem ipsum dolor" },
+                 { x: 36.0, y: 717.012, text: "sit amet" },
+                 { x: 36.0, y: 703.14, text: "Consectetur" },
+                 { x: 36.0, y: 689.268, text: "adipiscing elit" }])
+  end
+
+  it 'creates a html table with lists inside' do
+    generator.parse_file('table/list_in_cell.md')
+    expect_pdf([
+                 {x:36.0, y:744.756, text:"• test1"},
+                 {x:36.0, y:730.884, text:"• test2"},
+                 {x:36.0, y:717.012, text:"  test2cont"},
+                 {x:36.0, y:703.14, text:"• test3"},
+                 {x:36.0, y:689.268, text:"  • test3.1"},
+                 {x:36.0, y:675.396, text:"  • test3.2"},
+                 {x:36.0, y:661.524, text:"    test3.2cont"},
+                 {x:36.0, y:647.652, text:"  • test3.3"},
+                 {x:216.0, y:744.756, text:"[ ] aha"},
+                 {x:216.0, y:730.884, text:"[x] oho"},
+                 {x:216.0, y:717.012, text:"[ ] ehe"},
+                 {x:396.0, y:744.756, text:"1. wooo"},
+                 {x:396.0, y:730.884, text:"2. waaa"},
+                 {x:396.0, y:717.012, text:"3. wiiiii"}])
+  end
 end