From bdc5d2075876f2d087a0b415bdc3ec2e7f21ee0f Mon Sep 17 00:00:00 2001
From: Michael Dawson <nippysaurus@users.noreply.github.com>
Date: Tue, 9 May 2017 22:58:08 +1000
Subject: [PATCH] Add image support (#8)

Add support for decoding images in the RTF stream.
---
 bin/rtf_parse          | 26 +++++++++++++++++++++++
 lib/ruby-rtf/parser.rb | 11 ++++++++++
 spec/parser_spec.rb    | 47 ++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 84 insertions(+)

diff --git a/bin/rtf_parse b/bin/rtf_parse
index d8aa391..bd532a3 100755
--- a/bin/rtf_parse
+++ b/bin/rtf_parse
@@ -3,6 +3,7 @@
 
 require 'ruby-rtf'
 require 'pp'
+require 'base64'
 
 @prefix = ''
 @suffix = ''
@@ -80,6 +81,28 @@ def format(str, section)
   str << @prefix + section[:text].force_encoding('UTF-8') + @suffix
 end
 
+def process_image(section)
+  mods = section[:modifiers]
+  mime = ''
+  case mods[:picture_format]
+  when 'jpeg'
+    mime = 'image/jpg'
+  when 'png'
+    mime = 'image/png'
+  when 'bmp'
+    mime = 'image/bmp'
+  when 'wmf'
+    mime = 'image/x-wmf'
+  end
+  hex = section[:text].scan(/../).map(&:hex).pack('c*')
+  base64 = Base64.strict_encode64(hex)
+  width = 'auto'
+  width = mods[:picture_width] * (mods[:picture_scale_x] || 100) / 100 if mods[:picture_width]
+  height = 'auto'
+  height = mods[:picture_height] * (mods[:picture_scale_y] || 100) / 100 if mods[:picture_height]
+  "<img width='#{width}' height='#{height}' src=\"data:#{mime};base64,#{base64}\"/>\n"
+end
+
 doc = RubyRTF::Parser.new.parse(File.open(ARGV[0]).read)
 
 STDERR.puts doc
@@ -103,6 +126,9 @@ doc.sections.each do |section|
     end
     str << "</table>\n"
     next
+  elsif mods[:picture]
+    str << process_image(section)
+    next
   end
 
   format(str, section)
diff --git a/lib/ruby-rtf/parser.rb b/lib/ruby-rtf/parser.rb
index 3261ba8..54a375c 100644
--- a/lib/ruby-rtf/parser.rb
+++ b/lib/ruby-rtf/parser.rb
@@ -279,6 +279,15 @@ def handle_control(name, val, src, current_pos)
 
           @context_stack.pop
         end
+      when :pict then add_section!(picture: true)
+      when :jpegblip then add_section!(picture_format:'jpeg')
+      when :pngblip then add_section!(picture_format:'png')
+      when *[:dibitmap, :wbitmap] then add_section!(picture_format:'bmp')
+      when *[:wmetafile, :pmmetafile] then add_section!(picture_format:'wmf')
+      when :pich then add_section!(picture_height: RubyRTF.twips_to_points(val))
+      when :picw then add_section!(picture_width: RubyRTF.twips_to_points(val))
+      when :picscalex then add_section!(picture_scale_x: val.to_i)
+      when :picscaley then add_section!(picture_scale_y: val.to_i)
 
       else
         unless @seen[name]
@@ -514,7 +523,9 @@ def force_section!(mods = {}, text =  nil)
     #
     # @return [Nil]
     def reset_current_section!
+      paragraph = current_section[:modifiers].has_key?(:paragraph)
       current_section[:modifiers].clear
+      current_section[:modifiers][:paragraph] = true if paragraph
     end
 
     def current_context
diff --git a/spec/parser_spec.rb b/spec/parser_spec.rb
index 472fed8..741b178 100644
--- a/spec/parser_spec.rb
+++ b/spec/parser_spec.rb
@@ -94,6 +94,53 @@
       section[2][:text].should == 'Goodbye, cruel world.'
     end
 
+    context 'parses pictures' do
+      let(:src_bitmap) do
+        src = '{\rtf1 {\pict\wbitmap\picw7064\pich5292\picwgoal4005\pichgoal3000\picscalex111\picscaley109
+ffd8ffe000104a4649460001010100b400b40000ffe1158a687474703a2f2f6e732e61646f62652e636f6d2f7861702f3}}'
+      end
+      let(:src_jpeg) do
+        src = '{\rtf1 {\pict\jpegblip\picw7064\pich5292\picwgoal4005\pichgoal3000\picscalex111\picscaley109
+ffd8ffe000104a4649460001010100b400b40000ffe1158a687474703a2f2f6e732e61646f62652e636f6d2f7861702f3}}'
+      end
+
+      it 'should parse jpeg' do
+        section = parser.parse(src_jpeg).sections
+        section[0][:modifiers][:picture].should be_true
+        section[0][:modifiers][:picture_format].should == 'jpeg'
+      end
+
+      it 'should parse bmp' do
+        section = parser.parse(src_bitmap).sections
+        section[0][:modifiers][:picture].should be_true
+        section[0][:modifiers][:picture_format].should == 'bmp'
+        section = parser.parse(src_bitmap).sections
+        section[0][:modifiers][:picture].should be_true
+        section[0][:modifiers][:picture_format].should == 'bmp'
+      end
+
+      it 'should parse width' do
+        section = parser.parse(src_bitmap).sections
+        section[0][:modifiers][:picture_width].should == 7064 / 20.0
+      end
+
+      it 'should parse height' do
+        section = parser.parse(src_bitmap).sections
+        section[0][:modifiers][:picture_height].should == 5292 / 20.0
+      end
+
+      it 'should parse scale' do
+        section = parser.parse(src_bitmap).sections
+        section[0][:modifiers][:picture_scale_x].should == 111
+        section[0][:modifiers][:picture_scale_y].should == 109
+      end
+
+      it 'should parse picture data' do
+        section = parser.parse(src_bitmap).sections
+        section[0][:text].should == 'ffd8ffe000104a4649460001010100b400b40000ffe1158a687474703a2f2f6e732e61646f62652e636f6d2f7861702f3'
+      end
+    end
+
     it 'clears ul with ul0' do
       src = '{\rtf1 \ul\b Hello\b0\ul0 World}'
       section = parser.parse(src).sections