prawn-html 0.3.2 → 0.6.0

data/lib/prawn_html/html_parser.rb

@@ -4,12 +4,18 @@ require 'oga'
 
 module PrawnHtml
   class HtmlParser
+    REGEXP_STYLES = /\s*([^{\s]+)\s*{\s*([^}]*?)\s*}/m.freeze
+
     # Init the HtmlParser
     #
     # @param renderer [DocumentRenderer] document renderer
-    def initialize(renderer)
+    # @param ignore_content_tags [Array] array of tags (symbols) to skip their contents while preparing the PDF document
+    def initialize(renderer, ignore_content_tags: %i[script style])
       @processing = false
+      @ignore = false
+      @ignore_content_tags = ignore_content_tags
       @renderer = renderer
+      @styles = {}
     end
 
     # Processes HTML and renders it
@@ -17,17 +23,19 @@ module PrawnHtml
     # @param html [String] The HTML content to process
     def process(html)
       @processing = !html.include?('<body')
-      doc = Oga.parse_html(html)
-      traverse_nodes(doc.children)
+      @document = Oga.parse_html(html)
+      traverse_nodes(document.children)
       renderer.flush
     end
 
     private
 
-    attr_reader :processing, :renderer
+    attr_reader :document, :ignore, :processing, :renderer, :styles
 
     def traverse_nodes(nodes)
       nodes.each do |node|
+        next if node.is_a?(Oga::XML::Comment)
+
         element = node_open(node)
         traverse_nodes(node.children) if node.children.any?
         node_close(element) if element
@@ -37,21 +45,27 @@ module PrawnHtml
     def node_open(node)
       tag = node.is_a?(Oga::XML::Element) && init_element(node)
       return unless processing
+      return IgnoredTag.new(tag) if ignore
       return renderer.on_text_node(node.text) unless tag
 
-      attributes = prepare_attributes(node)
-      renderer.on_tag_open(tag, attributes)
+      renderer.on_tag_open(tag, attributes: prepare_attributes(node), element_styles: styles[node])
     end
 
     def init_element(node)
       node.name.downcase.to_sym.tap do |tag_name|
         @processing = true if tag_name == :body
-        renderer.assign_document_styles(extract_styles(node.text)) if tag_name == :style
+        @ignore = true if @processing && @ignore_content_tags.include?(tag_name)
+        process_styles(node.text) if tag_name == :style
       end
     end
 
-    def extract_styles(text)
-      text.scan(/\s*([^{\s]+)\s*{\s*([^}]*?)\s*}/m).to_h
+    def process_styles(text_styles)
+      hash_styles = text_styles.scan(REGEXP_STYLES).to_h
+      hash_styles.each do |selector, rule|
+        document.css(selector).each do |node|
+          styles[node] = rule
+        end
+      end
     end
 
     def prepare_attributes(node)
@@ -61,10 +75,21 @@ module PrawnHtml
     end
 
     def node_close(element)
-      renderer.on_tag_close(element) if @processing
+      if processing
+        renderer.on_tag_close(element) unless ignore
+        @ignore = false if ignore && @ignore_content_tags.include?(element.tag)
+      end
       @processing = false if element.tag == :body
     end
   end
 
+  class IgnoredTag
+    attr_accessor :tag
+
+    def initialize(tag_name)
+      @tag = tag_name
+    end
+  end
+
   HtmlHandler = HtmlParser
 end

data/lib/prawn_html/pdf_wrapper.rb

@@ -6,7 +6,7 @@ module PrawnHtml
   class PdfWrapper
     extend Forwardable
 
-    def_delegators :@pdf, :bounds, :start_new_page
+    def_delegators :@pdf, :start_new_page
 
     # Wrapper for Prawn PDF Document
     #
@@ -24,6 +24,46 @@ module PrawnHtml
       pdf.move_down(move_down)
     end
 
+    # Calculate the height of a buffer of items
+    #
+    # @param buffer [Array] Buffer of items
+    # @param options [Hash] Output options
+    #
+    # @return [Float] calculated height
+    def calc_buffer_height(buffer, options)
+      pdf.height_of_formatted(buffer, options)
+    end
+
+    # Calculate the width of a buffer of items
+    #
+    # @param buffer [Array] Buffer of items
+    #
+    # @return [Float] calculated width
+    def calc_buffer_width(buffer)
+      width = 0
+      buffer.each do |item|
+        font_family = item[:font] || pdf.font.name
+        pdf.font(font_family, size: item[:size] || pdf.font_size) do
+          width += pdf.width_of(item[:text], inline_format: true)
+        end
+      end
+      width
+    end
+
+    # Height of the page
+    #
+    # @return [Float] height
+    def page_height
+      pdf.bounds.height
+    end
+
+    # Width of the page
+    #
+    # @return [Float] width
+    def page_width
+      pdf.bounds.width
+    end
+
     # Draw a rectangle
     #
     # @param x [Float] left position of the rectangle
@@ -66,12 +106,12 @@ module PrawnHtml
     # @param buffer [Array] array of text items
     # @param options [Hash] hash of options
     # @param bounding_box [Array] bounding box arguments, if bounded
-    def puts(buffer, options, bounding_box: nil)
-      return pdf.formatted_text(buffer, options) unless bounding_box
+    def puts(buffer, options, bounding_box: nil, left_indent: 0)
+      return output_buffer(buffer, options, left_indent: left_indent) unless bounding_box
 
       current_y = pdf.cursor
       pdf.bounding_box(*bounding_box) do
-        pdf.formatted_text(buffer, options)
+        output_buffer(buffer, options, left_indent: left_indent)
       end
       pdf.move_cursor_to(current_y)
     end
@@ -90,5 +130,12 @@ module PrawnHtml
     private
 
     attr_reader :pdf
+
+    def output_buffer(buffer, options, left_indent:)
+      formatted_text = proc { pdf.formatted_text(buffer, options) }
+      return formatted_text.call if left_indent == 0
+
+      pdf.indent(left_indent, 0, &formatted_text)
+    end
   end
 end

data/lib/prawn_html/tag.rb

@@ -2,7 +2,16 @@
 
 module PrawnHtml
   class Tag
-    TAG_CLASSES = %w[A B Body Br Del Div H Hr I Img Li Mark Ol P Small Span U Ul].freeze
+    extend Forwardable
+
+    CALLBACKS = {
+      'Background' => Callbacks::Background,
+      'StrikeThrough' => Callbacks::StrikeThrough
+    }.freeze
+
+    TAG_CLASSES = %w[A B Blockquote Body Br Code Del Div H Hr I Img Li Mark Ol P Pre Small Span Sub Sup U Ul].freeze
+
+    def_delegators :@attrs, :styles, :update_styles
 
     attr_accessor :parent
     attr_reader :attrs, :tag
@@ -11,12 +20,11 @@ module PrawnHtml
     #
     # @param tag [Symbol] tag name
     # @param attributes [Hash] hash of element attributes
-    # @param document_styles [Hash] hash of document styles
-    def initialize(tag, attributes = {}, document_styles = {})
+    # @param options [Hash] options (container width/height/etc.)
+    def initialize(tag, attributes: {}, options: {})
       @tag = tag
-      element_styles = attributes.delete(:style)
+      @options = options
       @attrs = Attributes.new(attributes)
-      process_styles(document_styles, element_styles)
     end
 
     # Is a block tag?
@@ -35,6 +43,16 @@ module PrawnHtml
       block_styles
     end
 
+    # Process tag styles
+    #
+    # @param element_styles [String] extra styles to apply to the element
+    def process_styles(element_styles: nil)
+      attrs.merge_text_styles!(tag_styles, options: options) if respond_to?(:tag_styles)
+      attrs.merge_text_styles!(element_styles, options: options) if element_styles
+      attrs.merge_text_styles!(attrs.style, options: options)
+      attrs.merge_text_styles!(extra_styles, options: options) if respond_to?(:extra_styles)
+    end
+
     # Styles to apply on tag closing
     #
     # @return [Hash] hash of styles to apply
@@ -42,13 +60,6 @@ module PrawnHtml
       styles.slice(*Attributes::STYLES_APPLY[:tag_close])
     end
 
-    # Styles hash
-    #
-    # @return [Hash] hash of styles
-    def styles
-      attrs.styles
-    end
-
     # Styles to apply on tag opening
     #
     # @return [Hash] hash of styles to apply
@@ -74,23 +85,6 @@ module PrawnHtml
 
     private
 
-    def evaluate_document_styles(document_styles)
-      selectors = [
-        tag.to_s,
-        attrs['class'] ? ".#{attrs['class']}" : nil,
-        attrs['id'] ? "##{attrs['id']}" : nil
-      ].compact!
-      document_styles.each_with_object({}) do |(sel, attributes), res|
-        res.merge!(attributes) if selectors.include?(sel)
-      end
-    end
-
-    def process_styles(document_styles, element_styles)
-      attrs.merge_styles!(attrs.process_styles(tag_styles)) if respond_to?(:tag_styles)
-      doc_styles = evaluate_document_styles(document_styles)
-      attrs.merge_styles!(doc_styles)
-      el_styles = Attributes.parse_styles(element_styles)
-      attrs.merge_styles!(attrs.process_styles(el_styles)) if el_styles
-    end
+    attr_reader :options
   end
 end

data/lib/prawn_html/tags/a.rb

@@ -5,8 +5,15 @@ module PrawnHtml
     class A < Tag
       ELEMENTS = [:a].freeze
 
+      def extra_styles
+        attrs.href ? "href: #{attrs.href}" : nil
+      end
+
       def tag_styles
-        attrs.href ? { 'href' => attrs.href } : {}
+        <<~STYLES
+          color: #00E;
+          text-decoration: underline;
+        STYLES
       end
     end
   end

data/lib/prawn_html/tags/b.rb

@@ -6,9 +6,7 @@ module PrawnHtml
       ELEMENTS = [:b, :strong].freeze
 
       def tag_styles
-        {
-          'font-weight' => 'bold'
-        }
+        'font-weight: bold'
       end
     end
   end

data/lib/prawn_html/tags/blockquote.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+module PrawnHtml
+  module Tags
+    class Blockquote < Tag
+      ELEMENTS = [:blockquote].freeze
+
+      MARGIN_BOTTOM = 12.7
+      MARGIN_LEFT = 40.4
+      MARGIN_TOP = 12.7
+
+      def block?
+        true
+      end
+
+      def tag_styles
+        <<~STYLES
+          margin-bottom: #{MARGIN_BOTTOM}px;
+          margin-left: #{MARGIN_LEFT}px;
+          margin-top: #{MARGIN_TOP}px;
+        STYLES
+      end
+    end
+  end
+end

data/lib/prawn_html/tags/body.rb

@@ -4,6 +4,10 @@ module PrawnHtml
   module Tags
     class Body < Tag
       ELEMENTS = [:body].freeze
+
+      def block?
+        true
+      end
     end
   end
 end

data/lib/prawn_html/tags/br.rb

@@ -5,14 +5,14 @@ module PrawnHtml
     class Br < Tag
       ELEMENTS = [:br].freeze
 
-      BR_SPACING = Utils.convert_size('12')
+      BR_SPACING = Utils.convert_size('17')
 
       def block?
         true
       end
 
       def custom_render(pdf, context)
-        return if context.last_text_node
+        return if context.last_text_node || context.previous_tag != :br
 
         pdf.advance_cursor(BR_SPACING)
       end

data/lib/prawn_html/tags/code.rb

@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module PrawnHtml
+  module Tags
+    class Code < Tag
+      ELEMENTS = [:code].freeze
+
+      def tag_styles
+        'font-family: Courier'
+      end
+    end
+  end
+end

data/lib/prawn_html/tags/del.rb

@@ -6,9 +6,7 @@ module PrawnHtml
       ELEMENTS = [:del, :s].freeze
 
       def tag_styles
-        {
-          'callback' => Callbacks::StrikeThrough
-        }
+        'text-decoration: line-through'
       end
     end
   end

data/lib/prawn_html/tags/h.rb

@@ -6,30 +6,30 @@ module PrawnHtml
       ELEMENTS = [:h1, :h2, :h3, :h4, :h5, :h6].freeze
 
       MARGINS_TOP = {
-        h1: 25.5,
+        h1: 25,
         h2: 20.5,
-        h3: 19,
-        h4: 20,
+        h3: 18,
+        h4: 21.2,
         h5: 21.2,
-        h6: 23.5
+        h6: 22.8
       }.freeze
 
       MARGINS_BOTTOM = {
-        h1: 18.2,
-        h2: 17.5,
-        h3: 17.5,
-        h4: 22,
-        h5: 22,
-        h6: 26.5
+        h1: 15.8,
+        h2: 15.8,
+        h3: 15.8,
+        h4: 20,
+        h5: 21.4,
+        h6: 24.8
       }.freeze
 
       SIZES = {
-        h1: 31,
-        h2: 23.5,
-        h3: 18.2,
-        h4: 16,
+        h1: 31.5,
+        h2: 24,
+        h3: 18.7,
+        h4: 15.7,
         h5: 13,
-        h6: 10.5
+        h6: 10.8
       }.freeze
 
       def block?
@@ -37,12 +37,12 @@ module PrawnHtml
       end
 
       def tag_styles
-        @tag_styles ||= {
-          'font-size' => SIZES[tag].to_s,
-          'font-weight' => 'bold',
-          'margin-bottom' => MARGINS_BOTTOM[tag].to_s,
-          'margin-top' => MARGINS_TOP[tag].to_s
-        }
+        <<~STYLES
+          font-size: #{SIZES[tag]}px;
+          font-weight: bold;
+          margin-bottom: #{MARGINS_BOTTOM[tag]}px;
+          margin-top: #{MARGINS_TOP[tag]}px;
+        STYLES
       end
     end
   end