swordfish 0.0.5 → 0.0.6

checksums.yaml

@@ -1,15 +1,15 @@
 ---
 !binary "U0hBMQ==":
   metadata.gz: !binary |-
-    NDJiOGRjNGUyYjg3OWRjZWYwNTJhNWI5YTQ0NTgyMWUxMTdlMWJiOQ==
+    OGRlYTY4Yjg5NmY5N2QzMWU2Yzg0ZjAxMDAwM2VmZjUwNThhNDMzOA==
   data.tar.gz: !binary |-
-    ZGJkYTgxMDUzZGRiZTU3M2I1MTEwMjJhNGE1NzZmMzlhNGNhZTFlOQ==
+    NGUzNTViOGMwMTJmZmFjOGE4YTA1NzU3MGQwYTMyY2I3YmYzZjhlNQ==
 SHA512:
   metadata.gz: !binary |-
-    YmY4YzdlODI5NGUwYzFjOTliZmJmZmRlZjc4Njc4YzlhMmQ3MGZiYjA5NGZj
-    NmE4YjdlNTYxNmM1MzM1MWFmMjUxOTE2MDhlMGI1NTFkZTdlMWI5ZTQwOTA1
-    MjIyZGI1NzY4YzcyNGU1NWI3MDllNTIwOGZkNmU0OWE4OTc5OGU=
+    YWI3YmFiZGJiYzk3NTRiNTAwNjljZmQ5ODA2NjY0MjMzNjljNTVmYWU3ZmUx
+    OGYwMTljZjU3YjY3MmYzYjcwYmE3NGYzZDc3ZThlNmNmOWJlZDA2YTg2NGFm
+    YzQwMmY0ZmZkMDM4NmQ1OTE4OGZmMjdkODEwZjE3NTUxYWFhZmI=
   data.tar.gz: !binary |-
-    MjgzM2YxMTMzNDlkMDJlN2JhNWI4NzM4M2U4Y2IyOGI0MjkyOGY1MDEzZjNk
-    MjYzNGYzNzBlYzY2YzdhZGExM2M4MDMzMDBmYjYxYTRlMzNlMWQ4MmY0YWIy
-    MTllYzM2MDQwNGZiMjEwMjdmZjZhNjMxNGFlMDNhNWMzN2UwMjg=
+    YTAxYmU1NDBjNzIwNWNhZWRjY2UzMjIwMGQ5ZTk0ZjgwMmVhMzQ5YjA0Yjhh
+    NWEyYzQxZDMzYzEyOWViYjQ2Y2RjZjg3OTkzMjRkYTc0NWZmOGIyMzVlNWZj
+    M2ZlN2U3MDE0YjZhYzkwMjA0MDQ2Y2FmNDViNDMwMmU1NDM5Mzk=

data/lib/swordfish/document.rb

@@ -56,8 +56,14 @@ module Swordfish
       end
     end
 
-    def to_html
-      @nodes.map(&:to_html).join
+    def to_html(opts = {})
+      html = @nodes.map(&:to_html).join
+      
+      if opts[:pretty]
+        Nokogiri::HTML(html).to_html
+      else
+        html
+      end
     end
 
     private

data/lib/swordfish/formats/docx.rb

@@ -157,11 +157,16 @@ module Swordfish
       # The 'run' is the basic unit of text in Office OpenXML. A paragraph, table cell, or other
       # block element may contain one or more runs, and each run has an associated set of styles.
       texts = []
-      node.children.each do |run_xml|
+      # A complex field is a special type of node spanning multiple runs, where most of the runs
+      # designate a special control flow rather than normal text.
+      complex_field = nil
+
+      nodes = node.is_a?(Array) ? node : node.children
+      nodes.each_with_index do |run_xml, idx|
         case run_xml.name
           when 'r'
-            # A true run node
-            if run_xml.xpath('./w:t').length > 0
+            if run_xml.xpath('./w:t').length > 0 && complex_field.nil?
+              # A True run node
               # Only examine the run if it includes text codes. The run may also include
               # things like comment nodes, which should be ignored.
               text = Swordfish::Node::Text.new
@@ -177,6 +182,36 @@ module Swordfish
                 @swordfish_doc.images[image.original_name] = read_image(image.original_name)
                 texts << image
               end
+            elsif run_xml.xpath('./w:fldChar').length > 0 || complex_field
+              # A complex field
+              case
+                when run_xml.xpath('./w:fldChar').length > 0 && run_xml.xpath('./w:fldChar')[0]['w:fldCharType'] == 'begin'
+                  # Start the complex field
+                  complex_field = true
+                when run_xml.xpath('./w:instrText').length > 0
+                  # An instruction run, defining the complex field's behavior
+                  instruction = run_xml.xpath('./w:instrText')[0].content
+                  if instruction =~ /^\s*HYPERLINK/
+                    # A hyperlink
+                    complex_field = Swordfish::Node::Hyperlink.new
+                    complex_field.href = instruction.match(/^\s*HYPERLINK "([^"]+)"/).captures[0]
+                  else
+                    # Anything else
+                    complex_field = Swordfish::Node::Text.new
+                  end
+                when run_xml.xpath('./w:t').length > 0 && complex_field.children.length.zero?
+                  # The textual content
+                  complex_field.append(_node_parse_runs(nodes.to_a[idx..-1]))
+                when run_xml.xpath('./w:fldChar').length > 0 && run_xml.xpath('./w:fldChar')[0]['w:fldCharType'] == 'end'
+                  # End the complex field
+                  if complex_field
+                    texts << complex_field
+                    complex_field = nil
+                  else
+                    # Handle the case where _node_parse_runs gets called from within a complex field
+                    return texts
+                  end
+              end
             end
           when 'hyperlink'
             # Hyperlink nodes are placed amongst other run nodes, but
@@ -189,16 +224,15 @@ module Swordfish
         end
       end
       # Clean up runs by merging them if they have identical styles
-      to_delete = []
-      texts.each_with_index do |text, idx|
-        if idx > 0
-          if text.is_a?(Swordfish::Node::Text) && texts[idx-1].is_a?(Swordfish::Node::Text) && text.style == texts[idx-1].style
-            texts[idx-1].content += text.content
-            to_delete << text
-          end
+      texts = texts.reduce([]) do |memo, run|
+        if memo.length > 0 && memo.last.is_a?(Swordfish::Node::Text) && run.is_a?(Swordfish::Node::Text) && memo.last.style == run.style
+          memo.last.content += run.content
+        else
+          memo << run
         end
+        memo
       end
-      texts.reject! {|t| to_delete.include?(t) }
+
       texts
     end
 

data/lib/swordfish/nodes/paragraph.rb

@@ -12,7 +12,7 @@ module Swordfish
           @children.map(&:to_html).join
         else
           text = @children.map(&:to_html).join
-          "<p>#{text}</p>" unless text.length.zero?
+          "<p>#{text}</p>" unless text =~ /^[[:space:]]*$/
         end
       end
 

data/lib/swordfish/nodes/text.rb

@@ -11,15 +11,21 @@ module Swordfish
 
       def to_html
         @content ||= ""
+        @content.gsub!(/[[:space:]]/, ' ')
+        leading_space = !!@content.lstrip!  # If there is a leading or trailing space,
+        trailing_space = !!@content.rstrip! # shift it outside of any formatting tags
         html = CGI::escapeHTML(@content)
-        html = "<i>#{html}</i>" if @style.italic?
-        html = "<b>#{html}</b>" if @style.bold?
-        html = "<u>#{html}</u>" if @style.underline?
-        html = "<strike>#{html}</strike>" if @style.strikethrough?
-        html = "<sup>#{html}</sup>" if @style.superscript?
-        html = "<sub>#{html}</sub>" if @style.subscript?
-        html = "<strong>#{html}</strong>" if @style.strong?
-        html = "<em>#{html}</em>" if @style.emphasis?
+        if html.length > 0
+          html = "<i>#{html}</i>" if @style.italic?
+          html = "<b>#{html}</b>" if @style.bold?
+          html = "<u>#{html}</u>" if @style.underline?
+          html = "<strike>#{html}</strike>" if @style.strikethrough?
+          html = "<sup>#{html}</sup>" if @style.superscript?
+          html = "<sub>#{html}</sub>" if @style.subscript?
+          html = "<strong>#{html}</strong>" if @style.strong?
+          html = "<em>#{html}</em>" if @style.emphasis?
+        end
+        html = "#{' ' if leading_space}#{html}#{' ' if trailing_space}"
         html
       end
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: swordfish
 version: !ruby/object:Gem::Version
-  version: 0.0.5
+  version: 0.0.6
 platform: ruby
 authors:
 - Martin Posthumus
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-06-03 00:00:00.000000000 Z
+date: 2014-06-05 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler