Marcel breakage not recognising XML mime type correctly, forced workaround: metanorma/metanorma#473

Author: opoudjis

lib/html2doc/mime.rb

@@ -41,8 +41,16 @@ def mime_attachment(boundary, _filename, item, dir)
   end
 
   def mime_type(item)
-    type = Marcel::MimeType.for Pathname.new(item) ||
-      'text/plain; charset="utf-8"'
+    abs_path = File.absolute_path(item)
+    type = Marcel::MimeType.for(Pathname.new(abs_path))
+    
+    # Marcel sometimes fails to detect XML files and returns application/octet-stream
+    # Override for .xml files when detection fails
+    if type == "application/octet-stream" && File.extname(abs_path).downcase == ".xml"
+      type = "application/xml"
+    end
+    
+    type ||= 'text/plain; charset="utf-8"'
     type = %(#{type} charset="utf-8") if /^text/.match?(type)
     type
   end

spec/html2doc_spec.rb

@@ -82,7 +82,7 @@ def word_body(xml, footnote)
   Content-ID: <filelist.xml>
   Content-Disposition: inline; filename="filelist.xml"
   Content-Transfer-Encoding: base64
-  Content-Type: #{Html2Doc.new({}).mime_type('filelist.xml')}
+  Content-Type: application/xml
 
   PHhtbCB4bWxuczpvPSJ1cm46c2NoZW1hcy1taWNyb3NvZnQtY29tOm9mZmljZTpvZmZpY2UiPgog
   ICAgICAgIDxvOk1haW5GaWxlIEhSZWY9Ii4uL3Rlc3QuaHRtIi8+ICA8bzpGaWxlIEhSZWY9ImZp
@@ -96,7 +96,7 @@ def word_body(xml, footnote)
   Content-ID: <filelist.xml>
   Content-Disposition: inline; filename="filelist.xml"
   Content-Transfer-Encoding: base64
-  Content-Type: #{Html2Doc.new({}).mime_type('filelist.xml')}
+  Content-Type: application/xml
   PHhtbCB4bWxuczpvPSJ1cm46c2NoZW1hcy1taWNyb3NvZnQtY29tOm9mZmljZTpvZmZpY2UiPgog
   ICAgICAgIDxvOk1haW5GaWxlIEhSZWY9Ii4uL3Rlc3QuaHRtIi8+ICA8bzpGaWxlIEhSZWY9ImZp
   bGVsaXN0LnhtbCIvPgogIDxvOkZpbGUgSFJlZj0iaGVhZGVyLmh0bWwiLz4KPC94bWw+Cg==
@@ -108,7 +108,7 @@ def word_body(xml, footnote)
   Content-ID: <filelist.xml>
   Content-Disposition: inline; filename="filelist.xml"
   Content-Transfer-Encoding: base64
-  Content-Type: #{Html2Doc.new({}).mime_type('filelist.xml')}
+  Content-Type: application/xml
 
   PHhtbCB4bWxuczpvPSJ1cm46c2NoZW1hcy1taWNyb3NvZnQtY29tOm9mZmljZTpvZmZpY2UiPgog
   ICAgICAgIDxvOk1haW5GaWxlIEhSZWY9Ii4uL3Rlc3QuaHRtIi8+ICA8bzpGaWxlIEhSZWY9IjFh