Description of problem:
If you run "publican update_po" and a zero-length file exists with the same name as one of the PO files that Publican would normally create or update, Publican updates the zero-length file with the correct msgid and (empty) msgstr entries, but without the normal PO file headers, making the file unusable.

Additionally, if the XML contained non-ASCII characters, these characters are simply dropped from the corresponding msgid entries.

Version-Release number of selected component (if applicable):
1.6.2

How reproducible:
100%

Steps to Reproduce:
1. Create a document and include a non-ASCII character in the XML somewhere
2. run "publican update_pot"
3. run "publican update_po --langs de-DE" 
4. open one of the PO files in the de-DE directory and delete the contents of the file (but not the file itself)
5. run "publican update_po --langs de-DE"
  
Actual results:
Publican updates the PO file that you emptied out in step 4, but the PO file contains no headers. Additionally, any UTF-8 characters in the XML file are simply dropped. (For example, the German word "für" ("for") becomes "fr"). During PO merging, Publican displays a warning similar to:

pot/Revision_History.pot:23: warning: The following msgid contains non-ASCII characters.
                                      This will cause problems to translators who use a character encoding
                                      different from yours. Consider using a pure ASCII msgid instead.
                                      für
pot/Revision_History.pot:23: invalid multibyte sequence
pot/Revision_History.pot:23: invalid multibyte sequence
pot/Revision_History.pot:23: invalid multibyte sequence
pot/Revision_History.pot:23: invalid multibyte sequence

Expected results:
Publican creates a valid PO file and does not drop content from the XML file.