The Word importer does not escape wiki-markup-like characters within tables leading to a dissimilar export artifact in Confluence.

Reproduction procedure:

• install Confluence 6.0.5 / 6.2.3
• import Test-page.docx
• the import result will be equal to the left side of 
  

Let's look at the problems more closely (see: ):

• 1. a bold Text starting with http will be transferred to a link with * at the beginning and at the end
• 2. the commands text will be c ommands
• 3. the code area with square brackets will be transferred to a corrupt unmigrated-wiki-markup macro
• 4. the command with pipe will be transferred to two columns and the pipe symbol has been removed
• 5. the carriage returns has been removed
• 6. the Oracle command line printout will be transferred to a unsorted list.

Workaround

1. Workaround - Character replacement

Search and replace all wiki-markup-like characters within the word file with the html code of the character.
e.g (see https://dev.w3.org/html5/html-author/charref for the full html code list)

The changed word file ( Test-page-workaround-html-code.docx ) can be handled by the word importer

(See https://jira.atlassian.com/browse/CONFSERVER-41626 and https://jira.atlassian.com/browse/CONFSERVER-41138 for carriage returns problem)

2. Workaround - Change the structure or fix the problems afterwards

The workaround is based on recommendations by getsupport.atlassian.com:
".. workaround is to change the structure of Office document to align with the limitation of Office Connector." and/or ".. edit or reconstruct the essential information in Editor after Word file imported .."
see: Test-page-workaround-recommendation-atlassian-support.docx and https://confluence.atlassian.com/confkb/office-connector-limitations-and-known-issues-170492994.html