erp5_hr: get correct document's creator

bt5/erp5_hr/SkinTemplateItem/portal_skins/erp5_hr/PositionApplicationModule_addPositionApplication.py

-from erp5.component.module.DateUtils import addToDate
+from cStringIO import StringIO
+import zipfile
+import re
 from DateTime import DateTime
 requester = None
 if not first_name or not last_name:
@@ -22,6 +24,19 @@ if link:
   if requester.default_link.getUrlString() != link:
     requester.default_link.edit(url_string=link)
 
+def getGenerator(text_document):
+  zipbuffer = StringIO(str(text_document.getData()))
+  zipfileobj = zipfile.ZipFile(zipbuffer)
+  try:
+    meta = zipfileobj.read('meta.xml')
+    generator_list  = re.findall('<meta:generator>.*?</meta:generator>', meta)
+    if generator_list:
+      generator = generator_list[0]
+      return generator[len('<meta:generator>') : len(generator) - len('</meta:generator>')]
+    return ''
+  except KeyError:
+    return ''
+
 position_application = context.position_application_module.newContent(
   portal_type='Position Application',
   destination_decision_value = requester,
@@ -86,6 +101,17 @@ for line in listbox:
                   }
     document = context.Base_contribute(**document_kw)
     content_information = document.getContentInformation()
+    # text is converted by cloudooo, so creator is changed to LibreOffice
+    # check content type
+    if document.getPortalType() == 'Text':
+      content_type = document.getContentTypeFromContent()
+      if content_type.endswith('msword'):
+        content_information['Creator'] = 'Microsoft Word'
+      else:
+        generator = getGenerator(document)
+        if generator:
+          content_information['Creator'] = generator
+
     creator = content_information.pop('Creator', content_information.pop('Generator', None))
     document_type = line['document_type']
     description=document_type