libis-format 1.3.7.2 → 1.3.8.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 45a69e173c161cfa97afd3ad4b2cc9193cc883c5acaca3708466ffec56c55d8c
-  data.tar.gz: fad16f856808e6b02eaa31a4b02f2abba250b59b3baa821c5e1a27c17e842d98
+  metadata.gz: ec695a5009a2c131983af21f4c773037731907280eb229982eeacc8f7c72bf6b
+  data.tar.gz: 48139c18c6b66ddc9b25941dcbecee30294c3d7124d889dd74fae34b2a176c74
 SHA512:
-  metadata.gz: f9109cc481ff67637fac70eaa5e55ea107fea452e3ae1636d54e628b234e70d5531aafc6be480205bbbb7d41b7f1f1f0bf2ab467f7db816b201c59b6b733b3d9
-  data.tar.gz: ddb206f9fa5dc4bea55859fa0660f725eab006e01fe0d039f158ca45d430bf41b3bb18591520052a996743716b400a8221a0ce73b4a2affac42ad79cc6980333
+  metadata.gz: b9f232072ceed43e950d51869478c18e03cb79e157749a34e7580330a9e379f3008bd8f963461ea81261bdaf95b65d779b1d3c0efd1da86ef2d1fc729316cb43
+  data.tar.gz: cb40a76524a3854692b81baace242bf96a22c92bcde21192dffdaf6bd5afc187c7b4c7190866e0f0b9c17462d4392acec8b12e07fc6acaaf76192b171c37cb05

data/lib/libis/format/converter/email_converter.rb

@@ -3,6 +3,7 @@
 require_relative 'base'
 
 require 'libis/format/tool/msg_to_pdf'
+require 'libis/format/tool/eml_to_pdf'
 require 'libis/format/type_database'
 require 'rexml/document'
 
@@ -10,7 +11,6 @@ module Libis
   module Format
     module Converter
       class EmailConverter < Libis::Format::Converter::Base
-
         def self.input_types
           %i[MSG EML]
         end
@@ -28,7 +28,22 @@ module Libis
         def convert(source, target, format, opts = {})
           super
 
-          Format::Tool::MsgToPdf.run(source, target)
+          tool = tool_for_format(opts, source)
+          tool.run(source, target)
+        rescue StandardError => e
+          { command: { status: -1 }, errors: [{ error: e.message, error_class: e.class, error_trace: e.backtrace }] }
+        end
+
+        private
+
+        def tool_for_format(opts, source)
+          if opts[:source_format] == :MSG || File.extname(source).casecmp('.msg').zero?
+            Format::Tool::MsgToPdf
+          elsif opts[:source_format] == :EML || File.extname(source).casecmp('.eml').zero?
+            Format::Tool::EmlToPdf
+          else
+            raise "Unsupported file extension #{File.extname(source)}"
+          end
         end
       end
     end

data/lib/libis/format/tool/eml_to_pdf.rb

@@ -0,0 +1,168 @@
+# rubocop:disable Style/Documentation, Metrics/*
+# frozen_string_literal: true
+
+require_relative 'mail_to_pdf'
+
+require 'mail'
+require 'word_wrap'
+
+module Libis
+  module Format
+    module Tool
+      class EmlToPdf < MailToPdf
+        protected
+
+        def open_email(source)
+          eml = File.read(source)
+          begin
+            msg = Mail.new(eml)
+          rescue StandardError => e
+            eml.force_encoding('ASCII-8BIT')
+            msg = Mail.new(eml)
+            @warnings << "Failed to parse message with default encoding, forced ASCII-8BIT: #{e.message}"
+          end
+          msg
+        rescue StandardError => e
+          raise "Failed to open message: #{e.message}"
+        end
+
+        def close_email(_msg)
+          true
+        end
+
+        def get_body_html(msg)
+          # Get the encoding
+
+          if !msg.multipart?
+            HTML_WRAPPER_TEMPLATE % WordWrap.ww(msg.decoded, 120, false)
+
+          elsif msg.html_part
+            body = msg.html_part.body.decoded
+
+            # if the encoding is not UTF-8, then the HTML may contain metadata that specifies the encoding
+            # the browser will use that metadata to render the HTML correctly, so we should not force it to UTF-8
+            encoding = body.scan(/<\?xml\s[^?>]*encoding="([^"]*)"[^?>]*\?>/).flatten.first
+            encoding ||= body.scan(/<meta\s+[^>]*charset=["']?([^"'>\s]+)["']?[^>]*>/).flatten.first
+            encoding ||= msg.html_part.charset
+            encoding ||= msg.content_type_parameters['charset'] || msg.charset || 'UTF-8'
+
+            body.force_encoding(encoding) unless encoding.casecmp(body.encoding.name).zero?
+
+            body
+          elsif msg.text_part
+            HTML_WRAPPER_TEMPLATE % msg.text_part.decoded
+
+          else
+            HTML_WRAPPER_TEMPLATE % ''
+
+          end
+        end
+
+        def get_subject(msg)
+          msg.subject || 'No Subject'
+        end
+
+        def get_headers(msg)
+          headers = {}
+          html = ''
+
+          field_list = msg.header_fields
+
+          %w[From To Cc Subject Date].each do |key|
+            value = find_hdr(field_list, key)
+            next unless value
+
+            if value.is_a? Time
+              begin
+                headers[key.downcase.to_sym] = value.iso8601
+                html += hdr_html(key, value.rfc2822)
+              rescue StandardError => e
+                logger.warn "Failed to parse date header '#{value}': #{e.message}"
+              end
+            else
+              headers[key.downcase.to_sym] = value
+              html += hdr_html(key, value)
+            end
+          end
+
+          [headers, html]
+        end
+
+        def get_inline_attachment_data(attachments, cid)
+          attachments.each do |attachment|
+            next unless attachment.has_content_id?
+            next unless attachment.cid == cid
+
+            begin
+              attachment.data.rewind
+            rescue NoMethodError
+              # do nothing, attachment.data is not a stream
+            end
+            return {
+              mime_type: attachment.mime_type,
+              base64: Base64.strict_encode64(attachment.read)
+            }
+          end
+          nil
+        end
+
+        def get_file_attachments(attachments, _used_files)
+          attachments.select do |attachment|
+            !attachment.inline? &&
+              !attachment.has_content_id? &&
+              attachment.mime_type != 'message/rfc822'
+          end
+        end
+
+        def get_mail_attachments(attachments)
+          attachments.select do |attachment|
+            !attachment.inline? &&
+              !attachment.has_content_id? &&
+              attachment.mime_type == 'message/rfc822'
+          end
+        end
+
+        def get_attachment_info(attachment)
+          if attachment.mime_type == 'message/rfc822'
+            sub_msg = Mail.new(attachment.body.to_s)
+            subject = sub_msg.subject.to_s.strip
+
+            {
+              embedded_msg: sub_msg,
+              filename: subject
+            }
+
+          elsif attachment.filename
+
+            {
+              data: attachment.decoded,
+              filename: attachment.filename
+            }
+
+          else
+            {
+              filename: attachment.content_id.to_s
+            }
+          end
+        end
+
+        private
+
+        def find_hdr(list, key)
+          hdr = list.find { |x| x.name.to_s =~ /^#{key}$/i }
+          return nil unless hdr
+
+          field = hdr.field
+          return hdr.value unless field
+
+          return field.decoded unless field.is_a? Mail::CommonDateField
+
+          return field.date_time.to_time.localtime if field.respond_to?(:date_time)
+
+          DateTime.parse(field.decoded).to_time.localtime
+        end
+      end
+    end
+  end
+end
+# rubocop:enable Style/Documentation, Metrics/*

data/lib/libis/format/tool/mail_to_pdf.rb

@@ -0,0 +1,447 @@
+# rubocop:disable Style/Documentation, Metrics/*
+# frozen_string_literal: true
+
+require 'base64'
+require 'cgi'
+require 'pdfkit'
+require 'time'
+require 'fileutils'
+require 'pathname'
+require 'libis/format/config'
+
+module Libis
+  module Format
+    module Tool
+      HEADER_STYLE = <<~HTML
+        <style>
+          .header-table {
+            margin: 0 0 10px 0;
+            padding: 0;
+            font-family: Arial, Helvetica, sans-serif;
+          }
+          .header-table table {
+            width: 100%;
+          }
+          .header-name {
+            padding-right: 5px;
+            color: #9E9E9E;
+            text-align: right;
+            vertical-align: top;
+            font-size: 12px;
+          }
+          .header-value {
+            font-size: 12px;
+            width: 99%;
+          }
+          .header_fields {
+            background: white;
+            margin: 0;
+            border: 1px solid #DDD;
+            border-radius: 3px;
+            padding: 8px;
+            box-sizing: border-box;
+          }
+        </style>
+      HTML
+
+      HEADER_TABLE_TEMPLATE = <<~HTML
+        <div class="header-table">
+          <table class="header_fields">
+            <tbody>
+        %s
+            </tbody>
+          </table>
+        </div>
+      HTML
+
+      HEADER_FIELD_TEMPLATE = <<~HTML
+        <tr>
+          <td class="header-name">%s</td>
+          <td class="header-value">%s</td>
+        </tr>
+      HTML
+
+      HTML_WRAPPER_TEMPLATE = <<~HTML
+        <!DOCTYPE html>
+        <html>
+          <head>
+            <style>
+              body {
+                font-size: 12px;
+                }
+            </style>
+            <title>title</title>
+          </head>
+          <body>
+            <pre>
+        %s
+            </pre>
+          </body>
+        </html>
+      HTML
+
+      HTML_BODY_TEMPLATE = <<~HTML
+        <!DOCTYPE html>
+        <html>
+          <head>
+            <style>
+              body {
+                font-size: 12px;
+              }
+            </style>
+            <title>title</title>
+          </head>
+          <body>
+            %s
+          </body>
+        </html>
+      HTML
+
+      ATTACHMENT_STYLE = <<~HTML
+        <style>
+          .attachment-list {
+            border: 1px solid #DDD;
+            margin: 0 0 10px 0;
+            padding: 0;
+            font-family: Arial, Helvetica, sans-serif;
+          }
+          .attachment-list ul {
+            list-style: disclosure-closed;
+          }
+          .attachment-list li {
+            font-size: 12px;
+            padding-left: 1em;
+          }
+        </style>
+      HTML
+
+      ATTACHMENT_LIST_TEMPLATE = <<~HTML
+        <div class="attachment-list">
+          <ul>
+        %s
+          </ul>
+        </div>
+      HTML
+
+      HTML_DOCTYPE_TEMPLATE = '<!DOCTYPE html>%s'
+      ATTACHMENT_ITEM_TEMPLATE = '<li>%s</li>'
+
+      IMG_CID_PLAIN_REGEX = /\[cid:(.*?)\]/im
+      IMG_CID_HTML_REGEX = /cid:([^"]*)/im
+
+      class MailToPdf
+        include ::Libis::Tools::Logger
+
+        def self.installed?
+          File.exist?(Libis::Format::Config[:wkhtmltopdf])
+        end
+
+        def self.run(source, target, **options)
+          new.run source, target, **options
+        end
+
+        def run(source, target, **options)
+          # Preliminary checks
+          @warnings = []
+
+          # PDF creation options
+          @pdf_options = {
+            page_size: 'A4',
+            margin_top: '10mm',
+            margin_bottom: '10mm',
+            margin_left: '10mm',
+            margin_right: '10mm',
+            # image_quality: 100,
+            # viewport_size: '2480x3508',
+            dpi: 300
+          }.merge options
+
+          # Check if source file exists
+          raise "File #{source} does not exist" unless File.exist?(source)
+
+          # Open the email
+          email = open_email(source)
+
+          # Convert the email message to PDF
+          result = email_to_pdf(email, target, root_msg: true)
+
+          # Close email message
+          close_email(email)
+
+          result
+        end
+
+        protected
+
+        def email_to_pdf(msg, target, root_msg: false)
+          # Make sure the target directory exists
+          outdir = File.dirname(target)
+          FileUtils.mkdir_p(outdir)
+
+          # Process the message body
+          # ------------------------
+          body = get_body(msg)
+
+          # Process headers
+          # ---------------
+          headers, headers_html = get_headers(msg)
+
+          # Add header section to the HTML body
+          body = add_headers_to_body(body, headers_html)
+
+          # Embed inline images
+          # -------------------
+          attachments = msg.attachments
+          used_files = embed_inline_attachments(body, attachments)
+
+          # Save other attachments
+          # ----------------------
+          attachments_dir = "#{target}.attachments"
+
+          files = save_attachments(attachments, attachments_dir, used_files)
+
+          # Add attachment section to the HTML body
+          body = add_attachments_to_body(body, files, attachments_dir)
+
+          # Save the HTML body as a .html file next to the PDF for debugging purposes
+          File.open("#{target}.html", 'wb') { |f| f.write(body) }
+
+          # Create PDF
+          # ----------
+          write_target_file(body, get_subject(msg), target)
+
+          files = [target] + files if File.exist?(target)
+
+          if root_msg
+            p = Pathname(File.dirname(files.first))
+            files.drop(1).each do |f|
+              (headers[:attachments] ||= []) << Pathname.new(f).relative_path_from(p).to_s
+            end
+          end
+
+          {
+            command: { status: 0 },
+            files:,
+            headers:,
+            warnings: @warnings
+          }
+        rescue StandardError => e
+          raise unless root_msg
+
+          close_email(msg) if msg
+          {
+            command: { status: -1 },
+            files: [],
+            headers: {},
+            errors: [
+              {
+                error: e.message,
+                error_class: e.class.name,
+                error_trace: e.backtrace
+              }
+            ],
+            warnings: @warnings
+          }
+        end
+
+        def get_body(msg)
+          body = get_body_html(msg)
+
+          body = HTML_BODY_TEMPLATE % body unless /<body[^>]*>/i.match?(body)
+          body = HTML_DOCTYPE_TEMPLATE % body unless /<!DOCTYPE html/i.match?(body)
+          body.sub!(%r{<title>title</title>}, "<title>#{get_subject(msg)}</title>")
+
+          body
+        end
+
+        def add_headers_to_body(body, headers_html)
+          encoding = body.encoding
+          return body if headers_html.empty?
+
+          b = body.downcase
+
+          # Insert header block styles
+          if b.include?('</head>')
+            # if head exists, append the style block
+            body.gsub!(%r{</head>}i, "#{HEADER_STYLE}</head>")
+          elsif b.include?('<head/>')
+            # empty head, replace with the style block
+            body.gsub!(%r{<head/>}i, "<head>#{HEADER_STYLE}</head>")
+          else
+            # otherwise insert a head section before the body tag
+            body.gsub!(/<body/i, "<head>#{HEADER_STYLE}</head><body")
+          end
+          # Add the headers html table as first element in the body section
+          body.gsub!(/<body[^>]*>/i) { |m| "#{m}#{HEADER_TABLE_TEMPLATE % headers_html.encode(encoding)}" }
+          body
+        end
+
+        def hdr_html(key, value)
+          if key.is_a?(String) && value.is_a?(String) && !value.empty?
+            return format(HEADER_FIELD_TEMPLATE, key,
+                          CGI.escapeHTML(value))
+          end
+
+          ''
+        end
+
+        def embed_inline_attachments(body, attachments)
+          used_files = []
+
+          # First process plaintext cid entries
+          body.gsub!(IMG_CID_PLAIN_REGEX) do |_match|
+            data = get_inline_attachment_data(attachments, ::Regexp.last_match(1))
+            if data
+              used_files << ::Regexp.last_match(1)
+              "<img src=\"data:#{data[:mime_type]};base64,#{data[:base64]}\"/>"
+            else
+              '<img src=""/>'
+            end
+          end
+
+          # Then process HTML img tags with CID entries
+          body.gsub!(IMG_CID_HTML_REGEX) do |_match|
+            data = get_inline_attachment_data(attachments, ::Regexp.last_match(1))
+            if data
+              used_files << ::Regexp.last_match(1)
+              "data:#{data[:mime_type]};base64,#{data[:base64]}"
+            else
+              ''
+            end
+          end
+
+          used_files
+        end
+
+        def save_attachments(attachments, outdir, used_files)
+          files = []
+
+          digits = ((attachments.count + 1) / 10) + 1
+          i = 1
+
+          get_attachments(attachments, used_files).each do |attachment|
+            prefix = "#{format('%0*d', digits, i)}-"
+
+            info = get_attachment_info(attachment)
+
+            if info[:embedded_msg]
+              sub_msg = info[:embedded_msg]
+              file = File.join(outdir, "#{prefix}#{info[:filename].tr('/', '_')}.msg.pdf")
+
+              result = email_to_pdf(sub_msg, file, root_msg: false)
+
+              if (e = result[:error])
+                raise e
+              end
+
+              files += result[:files]
+            elsif info[:data]
+              file = File.join(outdir, "#{prefix}#{info[:filename].tr('/', '_')}")
+              FileUtils.mkdir_p(File.dirname(file))
+              File.open(file, 'wb') { |f| f.write(info[:data]) }
+              files << file
+            else
+              @warnings << "Attachment #{info[:filename]} cannot be extracted"
+              next
+            end
+
+            i += 1
+          end
+          files
+        end
+
+        def add_attachments_to_body(body, files, attachments_dir)
+          return body if files.empty?
+
+          b = body.downcase
+
+          # Insert attachment block styles
+          if b.include?('</head>')
+            # if head exists, append the style block
+            body.gsub!(%r{</head>}i, "#{ATTACHMENT_STYLE}</head>")
+          elsif b.include?('<head/>')
+            # empty head, replace with the style block
+            body.gsub!(%r{<head/>}i, "<head>#{ATTACHMENT_STYLE}</head>")
+          else
+            # otherwise insert a head section before the body tag
+            body.gsub!(/<body/i, "<head>#{ATTACHMENT_STYLE}</head><body")
+          end
+
+          # Filter files to only include those that are in the attachments directory
+          # and map them to relative paths from the attachments directory
+          items = files.filter_map do |f|
+            Pathname.new(f).relative_path_from(Pathname.new(attachments_dir)).to_s if File.dirname(f) == attachments_dir
+          end
+
+          # Create the attachment item HTML
+          items = items.map do |f|
+            format(ATTACHMENT_ITEM_TEMPLATE, f, File.basename(f))
+          end.join("\n")
+
+          # Create the attachment list HTML
+          attachments_html = ATTACHMENT_LIST_TEMPLATE % items
+
+          # make sure the attachments_html is encoded in the same encoding as the body
+          attachments_html = attachments_html.encode(body.encoding)
+
+          # Add the attachments html list after the headers
+          # if there are no headers, then add it at the beginning of the body section
+          body.sub!(%r{<div class="header-table">.*?</div>}im) { |m| "#{m}#{attachments_html}" } ||
+            body.gsub!(/<body[^>]*>/i) { |m| "#{m}#{attachments_html}" }
+
+          body
+        end
+
+        def get_attachments(attachments, used_files)
+          get_file_attachments(attachments, used_files) + get_mail_attachments(attachments)
+        end
+
+        def write_target_file(body, title, target)
+          kit = PDFKit.new(body, title: title || 'message', **@pdf_options)
+          pdf = kit.to_pdf
+          File.open(target, 'wb') { |f| f.write(pdf) }
+        end
+
+        # ---------------------------------------
+        # Methods to be implemented by subclasses
+        # ---------------------------------------
+        def open_email(_source)
+          raise NotImplementedError, 'Subclasses must implement the open_email method'
+        end
+
+        def close_email(_email)
+          raise NotImplementedError, 'Subclasses must implement the close_email method'
+        end
+
+        def get_body_html(_msg)
+          raise NotImplementedError, 'Subclasses must implement the get_body_html method'
+        end
+
+        def get_subject(_msg)
+          raise NotImplementedError, 'Subclasses must implement the get_subject method'
+        end
+
+        def get_headers(_msg)
+          raise NotImplementedError, 'Subclasses must implement the get_headers method'
+        end
+
+        def get_inline_attachment_data(_attachments, _cid)
+          raise NotImplementedError, 'Subclasses must implement the get_inline_attachment_data method'
+        end
+
+        def get_file_attachments(_attachments, _used_files)
+          raise NotImplementedError, 'Subclasses must implement the get_file_attachments method'
+        end
+
+        def get_mail_attachments(_attachments)
+          raise NotImplementedError, 'Subclasses must implement the get_mail_attachments method'
+        end
+
+        def get_attachment_info(_attachment)
+          raise NotImplementedError, 'Subclasses must implement the get_attachment_info method'
+        end
+      end
+    end
+  end
+end
+# rubocop:enable Style/Documentation, Metrics/*

data/lib/libis/format/tool/msg_to_pdf.rb

@@ -1,270 +1,154 @@
+# rubocop:disable Style/Documentation, Metrics/*
 # frozen_string_literal: true
 
-require 'mapi/msg'
-require 'rfc_2047'
-require 'cgi'
-require 'pdfkit'
-require 'time'
-require 'fileutils'
-require 'pathname'
-require 'libis/format/config'
+require_relative 'mail_to_pdf'
+
+require 'msg_extractor'
+require 'word_wrap'
 
 module Libis
   module Format
     module Tool
-      class MsgToPdf
-        include ::Libis::Tools::Logger
+      class MsgToPdf < MailToPdf
+        protected
 
-        HEADER_STYLE = '<style>.header-table {margin: 0 0 20 0;padding: 0;font-family: Arial, Helvetica, sans-serif;}.header-name {padding-right: 5px;color: #9E9E9E;text-align: right;vertical-align: top;font-size: 12px;}.header-value {font-size: 12px;}#header_fields {#background: white;#margin: 0;#border: 1px solid #DDD;#border-radius: 3px;#padding: 8px;#width: 100%%;#box-sizing: border-box;#}</style><script type="text/javascript">function timer() {try {parent.postMessage(Math.max(document.body.offsetHeight, document.body.scrollHeight), \'*\');} catch (r) {}setTimeout(timer, 10);};timer();</script>' # rubocop:disable Layout/LineLength
-        HEADER_TABLE_TEMPLATE = '<div class="header-table"><table id="header_fields"><tbody>%s</tbody></table></div>'
-        HEADER_FIELD_TEMPLATE = '<tr><td class="header-name">%s</td><td class="header-value">%s</td></tr>'
-        HTML_WRAPPER_TEMPLATE = '<!DOCTYPE html><html><head><style>body {font-size: 0.5cm;}</style><title>title</title></head><body>%s</body></html>' # rubocop:disable Layout/LineLength
+        def open_email(source)
+          msg = nil
 
-        IMG_CID_PLAIN_REGEX = /\[cid:(.*?)\]/m
-        IMG_CID_HTML_REGEX = /cid:([^"]*)/m
+          # Open the message
+          msg = MsgExtractor.open(source)
 
-        def self.installed?
-          File.exist?(Libis::Format::Config[:wkhtmltopdf])
-        end
+          unless msg.is_a?(MsgExtractor::Message)
+            raise "File #{File.basename(source)} is not an Outlook message but a #{msg.class.name}"
+          end
 
-        def self.run(source, target, **options)
-          new.run source, target, **options
+          msg
+        rescue StandardError => e
+          raise "Failed to open message: #{e.message}"
         end
 
-        def run(source, target, **options)
-          # Preliminary checks
-          # ------------------
-
-          @warnings = []
-
-          # Check if source file exists
-          raise "File #{source} does not exist" unless File.exist?(source)
-
-          # Retrieving the message
-          # ----------------------
-
-          # Open the message
-          msg = Mapi::Msg.open(source)
-
-          target_format = options.delete(:to_html) ? :HTML : :PDF
-          result = msg_to_pdf(msg, target, target_format, options)
-          msg.close
-          result
+        def close_email(_msg)
+          true
         end
 
-        def msg_to_pdf(msg, target, target_format, pdf_options, root_msg: true)
-          # Make sure the target directory exists
-          outdir = File.dirname(target)
-          FileUtils.mkdir_p(outdir)
-
+        def get_body_html(msg)
           # Get the body of the message in HTML
-          body = msg.properties.body_html
+          body = msg.html_body
 
           # Embed plain body in HTML as a fallback
-          body ||= HTML_WRAPPER_TEMPLATE % msg.properties.body
-
-          # Check and fix the character encoding
-          begin
-            # Try to encode into UTF-8
-            body.encode!('UTF-8', universal_newline: true)
-          rescue Encoding::InvalidByteSequenceError, Encoding::UndefinedConversionError
-            begin
-              # If it fails, the text may be in Windows' Latin1 (ISO-8859-1)
-              body.force_encoding('ISO-8859-1').encode!('UTF-8', universal_newline: true)
-            rescue Encoding::InvalidByteSequenceError, Encoding::UndefinedConversionError => e
-              # If that fails too, log a warning and replace the invalid/unknown with a ? character.
-              @warnings << "#{e.class}: #{e.message}"
-              body.encode!('UTF-8', universal_newline: true, invalid: :replace, undef: :replace)
-            end
-          end
+          body ||= HTML_WRAPPER_TEMPLATE % WordWrap.ww(msg.body, 120, false)
+
+          # Worst case, just create empty body
+          body ||= HTML_WRAPPER_TEMPLATE % ''
+
+          # # Check and fix the character encoding
+          # begin
+          #   # Try to encode into UTF-8
+          #   body.encode!('UTF-8', universal_newline: true)
+          # rescue Encoding::InvalidByteSequenceError, Encoding::UndefinedConversionError
+          #   begin
+          #     # If it fails, the text may be in Windows' Latin1 (ISO-8859-1)
+          #     body.force_encoding('ISO-8859-1').encode!('UTF-8', universal_newline: true)
+          #   rescue Encoding::InvalidByteSequenceError, Encoding::UndefinedConversionError => e
+          #     # If that fails too, log a warning and replace the invalid/unknown with a ? character.
+          #     @warnings << "#{e.class}: #{e.message}"
+          #     body.encode!('UTF-8', universal_newline: true, invalid: :replace, undef: :replace)
+          #   end
+          # end
+
+          body
+        end
 
-          # Process headers
-          # ---------------
+        def get_subject(msg)
+          msg.subject || ''
+        end
+
+        def get_headers(msg)
           headers = {}
-          hdr_html = ''
+          html = ''
 
           %w[From To Cc Subject Date].each do |key|
-            value = find_hdr(msg.headers, key)
-            if value
-              headers[key.downcase.to_sym] = value
-              hdr_html += hdr_html(key, value)
-            end
-          end
-
-          [:date].each do |key|
-            next unless headers[key]
-
-            headers[key] = DateTime.parse(headers[key]).to_time.localtime.iso8601
-          end
+            value = find_hdr(msg, key)
+            next unless value
 
-          # Add header section to the HTML body
-          unless hdr_html.empty?
-            # Insert header block styles
-            if body =~ %r{</head>}
-              # if head exists, append the style block
-              body.gsub!(%r{</head>}, "#{HEADER_STYLE}</head>")
-            elsif body =~ %r{<head/>}
-              # empty head, replace with the style block
-              body.gsub!(%r{<head/>}, "<head>#{HEADER_STYLE}</head>")
+            if key.casecmp('Date').zero? && value.is_a?(Time)
+              headers[key.downcase.to_sym] = value.iso8601
+              html += hdr_html(key, value.rfc2822)
             else
-              # otherwise insert a head section before the body tag
-              body.gsub!(/<body/, "<head>#{HEADER_STYLE}</head><body")
+              headers[key.downcase.to_sym] = value
+              html += hdr_html(key, value)
             end
-            # Add the headers html table as first element in the body section
-            body.gsub!(/<body[^>]*>/) { |m| "#{m}#{HEADER_TABLE_TEMPLATE % hdr_html}" }
           end
 
-          # Embed inline images
-          # -------------------
-          attachments = msg.attachments
-          used_files = []
-
-          # First process plaintext cid entries
-          body.gsub!(IMG_CID_PLAIN_REGEX) do |_match|
-            data = get_attachment_data(attachments, ::Regexp.last_match(1))
-            if data
-              used_files << ::Regexp.last_match(1)
-              "<img src=\"data:#{data[:mime_type]};base64,#{data[:base64]}\"/>"
-            else
-              '<img src=""/>'
-            end
-          end
+          [headers, html]
+        end
 
-          # Then process HTML img tags with CID entries
-          body.gsub!(IMG_CID_HTML_REGEX) do |_match|
-            data = get_attachment_data(attachments, ::Regexp.last_match(1))
-            if data
-              used_files << ::Regexp.last_match(1)
-              "data:#{data[:mime_type]};base64,#{data[:base64]}"
-            else
-              ''
-            end
-          end
+        def get_inline_attachment_data(attachments, cid)
+          attachments.each do |attachment|
+            next unless attachment.content_id == cid
 
-          # Create PDF
-          # ----------
-          files = []
-
-          if target_format == :PDF
-            # PDF creation options
-            pdf_options = {
-              page_size: 'A4',
-              margin_top: '10mm',
-              margin_bottom: '10mm',
-              margin_left: '10mm',
-              margin_right: '10mm',
-              # image_quality: 100,
-              # viewport_size: '2480x3508',
-              dpi: 300
-            }.merge pdf_options
-
-            subject = find_hdr(msg.headers, 'Subject')
-            kit = PDFKit.new(body, title: (subject || 'message'), **pdf_options)
-            pdf = kit.to_pdf
-            File.open(target, 'wb') { |f| f.write(pdf) }
-          else
-            File.open(target, 'wb') { |f| f.write(body) }
-          end
-          files << target if File.exist?(target)
-
-          # Save attachments
-          # ----------------
-          outdir = File.join(outdir, "#{File.basename(target)}.attachments")
-          digits = ((attachments.count + 1) / 10) + 1
-          i = 1
-          attachments.delete_if { |a| a.properties.attachment_hidden }.each do |a|
-            prefix = "#{format('%0*d', digits, i)}-"
-            if (sub_msg = a.instance_variable_get(:@embedded_msg))
-              subject = a.properties[:display_name] || sub_msg.subject || ''
-              file = File.join(outdir, "#{prefix}#{subject.gsub('/', '_')}.msg.#{target_format.to_s.downcase}")
-              result = msg_to_pdf(sub_msg, file, target_format, pdf_options, root_msg: false)
-              if (e = result[:error])
-                raise e
-              end
-
-              files += result[:files]
-            elsif a.filename
-              next if used_files.include?(a.filename)
-
-              file = File.join(outdir, "#{prefix}#{a.filename.gsub('/', '_')}")
-              FileUtils.mkdir_p(File.dirname(file))
-              File.open(file, 'wb') { |f| a.save(f) }
-              files << file
-            else
-              @warnings << "Attachment #{a.properties[:display_name]} cannot be extracted"
-              next
-            end
-            i += 1
+            return {
+              mime_type: attachment.mime_type,
+              base64: Base64.encode64(attachment.data).gsub(/[\r\n]/, '')
+            }
           end
+          nil
+        end
 
-          if root_msg
-            p = Pathname(File.dirname(files.first))
-            files[1..].each do |f|
-              (headers[:attachments] ||= []) << Pathname.new(f).relative_path_from(p).to_s
-            end
+        def get_file_attachments(attachments, _used_files)
+          attachments.select do |attachment|
+            !attachment.content_id && !attachment.embedded_message? && attachment.filename
           end
+        end
 
-          {
-            command: { status: 0 },
-            files:,
-            headers:,
-            warnings: @warnings
-          }
-        rescue Exception => e
-          raise unless root_msg
-
-          msg.close
-          {
-            command: { status: -1 },
-            files: [],
-            headers: {},
-            errors: [
-              {
-                error: e.message,
-                error_class: e.class.name,
-                error_trace: e.backtrace
-              }
-            ],
-            warnings: @warnings
-          }
+        def get_mail_attachments(attachments)
+          attachments.select(&:embedded_message?)
         end
 
-        protected
+        def get_attachment_info(attachment)
+          if attachment.embedded_message?
 
-        def eml_to_html; end
+            {
+              embedded_msg: attachment.data,
+              filename: attachment.message&.subject || 'email'
+            }
 
-        private
+          elsif attachment.filename
 
-        def find_hdr(list, key)
-          keys = list.keys
-          if (k = keys.find { |x| x.to_s =~ /^#{key}$/i })
-            v = list[k]
-            v = v.first if v.is_a? Array
-            v = Rfc2047.decode(v).strip if v.is_a? String
-            return v
+            {
+              data: attachment.data,
+              filename: attachment.filename
+            }
+
+          else
+            {
+              filename: attachment.mime_type || 'unknown'
+            }
           end
-          nil
         end
 
-        def hdr_html(key, value)
-          return format(HEADER_FIELD_TEMPLATE, key, CGI.escapeHTML(value)) if key.is_a?(String) && value.is_a?(String) && !value.empty?
+        private
 
-          ''
-        end
+        def find_hdr(msg, key)
+          value = if key.casecmp('From').zero?
+                    msg.sender
+                  else
+                    msg.send(key.downcase.to_sym)
+                  end
 
-        def get_attachment_data(attachments, cid)
-          attachments.each do |attachment|
-            next unless attachment.properties.attach_content_id == cid
+          if value.is_a?(Array)
+            value.compact.empty? ? nil : value.compact.map(&:to_s).join(', ')
+
+          elsif value.is_a?(Time)
+            value.localtime
+
+          else
+            value.to_s
 
-            attachment.data.rewind
-            return {
-              mime_type: attachment.properties.attach_mime_tag,
-              base64: Base64.encode64(attachment.data.read).gsub(/[\r\n]/, '')
-            }
           end
-          nil
         end
       end
     end
   end
 end
+# rubocop:enable Style/Documentation, Metrics/*

data/lib/libis/format/version.rb

@@ -2,6 +2,6 @@
 
 module Libis
   module Format
-    VERSION = '1.3.7.2'
+    VERSION = '1.3.8.1'
   end
 end

data/libis-format.gemspec

@@ -16,7 +16,7 @@ Gem::Specification.new do |spec|
   spec.homepage      = ''
   spec.license       = 'MIT'
 
-  spec.platform      = Gem::Platform::JAVA if defined?(RUBY_ENGINE) && RUBY_ENGINE == 'jruby'
+  # spec.platform      = Gem::Platform::JAVA if defined?(RUBY_ENGINE) && RUBY_ENGINE == 'jruby'
   spec.required_ruby_version = '>= 3.2'
 
   spec.files = `git ls-files -z`.split("\x0").select do |f|
@@ -27,14 +27,18 @@ Gem::Specification.new do |spec|
 
   spec.add_runtime_dependency 'chromaprint', '~> 0.0.2'
   spec.add_runtime_dependency 'deep_dive', '~> 0.3'
-  spec.add_runtime_dependency 'libis-mapi', '~> 0.3'
+  # spec.add_runtime_dependency 'libis-mapi', '~> 0.3'
   spec.add_runtime_dependency 'libis-tools', '~> 1.1'
+  spec.add_runtime_dependency 'mail', '~> 2.9'
   spec.add_runtime_dependency 'mini_magick', '~> 5.0.1'
+  spec.add_runtime_dependency 'msg_extractor', '~> 0.1'
   spec.add_runtime_dependency 'naturally', '~> 2.2'
   spec.add_runtime_dependency 'new_rfc_2047', '~> 1.0'
   spec.add_runtime_dependency 'os', '~> 1.1'
   spec.add_runtime_dependency 'pdfinfo', '~> 1.4'
   spec.add_runtime_dependency 'pdfkit', '~> 0.8'
+  # spec.add_runtime_dependency 'ruby-msg', '~> 1.5.3.1'
+  spec.add_runtime_dependency 'word_wrap', '~> 1.0'
 
   spec.add_development_dependency 'awesome_print'
   spec.add_development_dependency 'equivalent-xml'

metadata

@@ -1,14 +1,13 @@
 --- !ruby/object:Gem::Specification
 name: libis-format
 version: !ruby/object:Gem::Version
-  version: 1.3.7.2
+  version: 1.3.8.1
 platform: ruby
 authors:
 - Kris Dekeyser
-autorequire:
 bindir: bin
 cert_chain: []
-date: 2025-06-18 00:00:00.000000000 Z
+date: 1980-01-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: chromaprint
@@ -39,33 +38,33 @@ dependencies:
       - !ruby/object:Gem::Version
         version: '0.3'
 - !ruby/object:Gem::Dependency
-  name: libis-mapi
+  name: libis-tools
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '0.3'
+        version: '1.1'
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '0.3'
+        version: '1.1'
 - !ruby/object:Gem::Dependency
-  name: libis-tools
+  name: mail
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '1.1'
+        version: '2.9'
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '1.1'
+        version: '2.9'
 - !ruby/object:Gem::Dependency
   name: mini_magick
   requirement: !ruby/object:Gem::Requirement
@@ -80,6 +79,20 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: 5.0.1
+- !ruby/object:Gem::Dependency
+  name: msg_extractor
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '0.1'
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '0.1'
 - !ruby/object:Gem::Dependency
   name: naturally
   requirement: !ruby/object:Gem::Requirement
@@ -150,6 +163,20 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: '0.8'
+- !ruby/object:Gem::Dependency
+  name: word_wrap
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '1.0'
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '1.0'
 - !ruby/object:Gem::Dependency
   name: awesome_print
   requirement: !ruby/object:Gem::Requirement
@@ -273,12 +300,14 @@ files:
 - lib/libis/format/identifier.rb
 - lib/libis/format/tool.rb
 - lib/libis/format/tool/droid.rb
+- lib/libis/format/tool/eml_to_pdf.rb
 - lib/libis/format/tool/extension_identification.rb
 - lib/libis/format/tool/ff_mpeg.rb
 - lib/libis/format/tool/fido.rb
 - lib/libis/format/tool/file_tool.rb
 - lib/libis/format/tool/fop_pdf.rb
 - lib/libis/format/tool/identification_tool.rb
+- lib/libis/format/tool/mail_to_pdf.rb
 - lib/libis/format/tool/msg_to_pdf.rb
 - lib/libis/format/tool/office_to_pdf.rb
 - lib/libis/format/tool/pdf_merge.rb
@@ -352,7 +381,6 @@ homepage: ''
 licenses:
 - MIT
 metadata: {}
-post_install_message:
 rdoc_options: []
 require_paths:
 - lib
@@ -367,8 +395,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.4.19
-signing_key:
+rubygems_version: 4.0.8
 specification_version: 4
 summary: LIBIS File format format services.
 test_files: []