diff options
| author | Gareth Rees <gareth@mysociety.org> | 2014-03-17 11:15:40 +0000 |
|---|---|---|
| committer | Gareth Rees <gareth@mysociety.org> | 2014-03-28 09:39:04 +0000 |
| commit | d57ca2a22579df4c634d554989c0ee9e4ebb5165 (patch) | |
| tree | e1d11c626cedf57373be95b6b1ec6ce4dc22ea30 /lib/attachment_to_html/adapters/rtf.rb | |
| parent | 0adf9399cbef42054809479c8f1b64dad7bbf8ca (diff) | |
Add AttachmentToHTML library
Extracts the attachment processing from FoiAttachment#body_to_html
AttachmentToHTML contains adapters which convert
- text/plain
- application/pdf
- application/rtf
Results are returned as an AttachmentHTML::HTML instance which contains
the raw HTML and other metadata about the conversion.
Diffstat (limited to 'lib/attachment_to_html/adapters/rtf.rb')
| -rw-r--r-- | lib/attachment_to_html/adapters/rtf.rb | 120 |
1 files changed, 120 insertions, 0 deletions
diff --git a/lib/attachment_to_html/adapters/rtf.rb b/lib/attachment_to_html/adapters/rtf.rb new file mode 100644 index 000000000..f38e5e381 --- /dev/null +++ b/lib/attachment_to_html/adapters/rtf.rb @@ -0,0 +1,120 @@ +module AttachmentToHTML + module Adapters + # Convert application/rtf documents in to HTML + class RTF + + attr_reader :attachment, :wrapper, :tmpdir + + # Public: Initialize a RTF converter + # + # attachment - the FoiAttachment to convert to HTML + # opts - a Hash of options (default: {}): + # :wrapper - String id of the div that wraps the + # attachment body + # :tmpdir - String name of directory to store the + # converted document + def initialize(attachment, opts = {}) + @attachment = attachment + @wrapper = opts.fetch(:wrapper, 'wrapper') + @tmpdir = opts.fetch(:tmpdir, ::Rails.root.join('tmp')) + end + + # Public: Convert the attachment to HTML + # + # Returns a String + def to_html + @html ||= generate_html + end + + # Public: Was the document conversion successful? + # + # Returns a Boolean + def success? + has_content? || contains_images? + end + + private + + def generate_html + html = "<!DOCTYPE html>" + html += "<html>" + html += "<head>" + html += "<title>#{ title }</title>" + html += "</head>" + html += "<body>" + html += "<div id=\"#{ wrapper }\">" + html += "<div id=\"view-html-content\">" + html += body + html += "</div>" + html += "</div>" + html += "</body>" + html += "</html>" + end + + def title + @title ||= attachment.display_filename + end + + def body + parsed_body + end + + # Parse the output of the converted attachment so that we can pluck + # the parts we need and insert in to our own sensible template + # + # Returns a Nokogiri::HTML::Document + def parsed + @parsed ||= Nokogiri::HTML.parse(convert) + end + + def parsed_body + parsed.css('body').inner_html + end + + # Does the body element have any content, excluding HTML tags? + # + # Returns a Boolean + def has_content? + !parsed.css('body').inner_text.empty? + end + + def contains_images? + parsed.css('body img').any? + end + + def convert + @converted ||= Dir.chdir(tmpdir) do + tempfile = create_tempfile + write_attachment_body_to_tempfile(tempfile) + + html = AlaveteliExternalCommand.run("unrtf", "--html", + tempfile.path, :timeout => 120 + ) + + cleanup_tempfile(tempfile) + + html + end + end + + def create_tempfile + if RUBY_VERSION.to_f >= 1.9 + Tempfile.new('foiextract', '.', :encoding => attachment.body.encoding) + else + Tempfile.new('foiextract', '.') + end + end + + def write_attachment_body_to_tempfile(tempfile) + tempfile.print(attachment.body) + tempfile.flush + end + + def cleanup_tempfile(tempfile) + tempfile.close + tempfile.delete + end + + end + end +end |