interface fully working: conversion from docx is handdled by py library mammoth

cgi-bin/pandoc.cgi

@@ -1,7 +1,7 @@
 #!/usr/bin/env python
 
 import cgitb; cgitb.enable()
-import cgi, os, sys, subprocess, shlex
+import cgi, os, sys, subprocess, shlex, mammoth
 from tempfile import NamedTemporaryFile as NTF 
 from settings import PANDOC 
 
@@ -83,36 +83,21 @@ if method == "POST":
             if download:
                 print "Content-Disposition: attachment;filename=\"hype_conversion.{0}\"".format(to_type['ext'])
             print
-            # p = subprocess.check_output('pandoc --from {0} --to {1} "{2}"'.format(ffrom, to, tmp.name), shell=True, stderr=subprocess.STDOUT)
             
-            # read content of tmp docx
-            cmd_head = "head {}".format(tmp.name)
-            cmd_head_list = shlex.split( cmd_head )
-            p_head = subprocess.Popen(cmd_head_list, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
 
-            # convert docx to html: error. No idea why
+            docx_file =  open(tmp.name, "rb")
-            cmd_unoconv = "unoconv --stdout -f html {}".format(tmp.name)
+            result = mammoth.convert_to_markdown(docx_file)
-            cmd_unoconv_list = shlex.split( cmd_unoconv )
+            out = (result.value).encode('utf-8')
-            p_unoconv = subprocess.Popen(cmd_unoconv_list, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+            err = result.messages
 
-            # convert docx to html: depends on most recente versions of pandoc
+            print out#, err
-            # cmd = "{} --from {} --to {} {}".format(PANDOC, from_format, to_format, tmp.name)
-            # cmd_list = shlex.split( cmd )
-            # p = subprocess.Popen(cmd_list, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
-            
-            # print p
-            # p = subprocess.Popen([PANDOC, '--from', from_format, '--to', to_format, tmp.name], stdout=subprocess.PIPE, stderr=subprocess.PIPE)
-
-            out, err = p_head.communicate() # p_unoconv.communicate()
-            #print tmp.name, from_format
-            print out, err
 
         except subprocess.CalledProcessError, e:
             print "Content-type: text/html;charset=utf-8"
             print
             print u"<span style='font-family:monospace'>An error occurred, pandoc said: {0}</span>".format(e.output).format("utf-8")
 
-#        tmp.unlink(tmp.name) # remove file
+        tmp.unlink(tmp.name) # remove file
 
     ###################################
     # FORM/TEXT INPUT
@@ -136,7 +121,7 @@ if method == "POST":
 
                 
                 out, err = p.communicate(text)
-                print cmd+' '+ out
+                print out
 
             except subprocess.CalledProcessError, e:
                 print "Content-type: text/html;charset=utf-8"