lib/html2doc/mime.rb in html2doc-0.8.2 vs lib/html2doc/mime.rb in html2doc-0.8.3

@@ -1,9 +1,10 @@
 require "uuidtools"
 require "base64"
 require "mime/types"
 require "image_size"
+require "fileutils"
 
 module Html2Doc
   def self.mime_preamble(boundary, filename, result)
     <<~"PREAMBLE"
     MIME-Version: 1.0
@@ -18,11 +19,11 @@
     PREAMBLE
   end
 
   def self.mime_attachment(boundary, filename, item, dir)
     encoded_file = Base64.strict_encode64(
-      File.read("#{dir}/#{item}"),
+      File.read("#{dir}/#{item}", encoding: "utf-8"),
     ).gsub(/(.{76})/, "\\1\n")
     <<~"FILE"
     --#{boundary}
     Content-Location: file:///C:/Doc/#{filename}_files/#{item}
     Content-Transfer-Encoding: base64
@@ -53,11 +54,11 @@
       next if item == "." || item == ".." || /^\./.match(item) ||
         item == "filelist.xml"
       mhtml += mime_attachment(boundary, filename, item, dir)
     end
     mhtml += "--#{boundary}--"
-    File.open("#{filename}.doc", "w") { |f| f.write mhtml }
+    File.open("#{filename}.doc", "w:UTF-8") { |f| f.write mhtml }
   end
 
   # max height for Word document is 400, max width is 680
   def self.image_resize(i, maxheight, maxwidth)
     realSize = ImageSize.path(i["src"]).size
@@ -70,17 +71,18 @@
     s
   end
 
   IMAGE_PATH = "//*[local-name() = 'img' or local-name() = 'imagedata']".freeze
 
+  # only processes locally stored images
   def self.image_cleanup(docxml, dir)
     docxml.xpath(IMAGE_PATH).each do |i|
+      next if /^http/.match i["src"]
       matched = /\.(?<suffix>\S+)$/.match i["src"]
       uuid = UUIDTools::UUID.random_create.to_s
       new_full_filename = File.join(dir, "#{uuid}.#{matched[:suffix]}")
-      # presupposes that the image source is local
-      system "cp #{i['src']} #{new_full_filename}"
+      FileUtils.cp i["src"], new_full_filename
       i["width"], i["height"] = image_resize(i, 400, 680)
       i["src"] = new_full_filename
     end
     docxml
   end
@@ -98,10 +100,11 @@
       matched = / src=['"](?<src>[^"']+)['"]/.match a[1]
       matched2 = /\.(?<suffix>\S+)$/.match matched[:src]
       uuid = UUIDTools::UUID.random_create.to_s
       new_full_filename = "file:///C:/Doc/#{filename}_files/#{uuid}.#{matched2[:suffix]}"
       dest_filename = File.join(dir, "#{uuid}.#{matched2[:suffix]}")
-      system "cp #{matched[:src]} #{dest_filename}"
+      #system "cp #{matched[:src]} #{dest_filename}"
+      FileUtils.cp matched[:src], dest_filename
       a[1].sub!(%r{ src=['"](?<src>[^"']+)['"]}, " src='#{new_full_filename}'")
     end
     a.join
   end