DocFetcher will do what you specified.
There are native Windows and Mac OS versions, as well as a portable version that should work on any platform for which Java is available.
On Windows, you can index a network folder either as a mapped drive or via its UNC path.
I don't think it has any "text mining" or "key word visualization" features itself. But it's built on Apache Lucene, so I'd imagine you could use other tools with the indexes that DocFetcher creates.
The source code is freely available and licensed under the Eclipse Public License, tho the project website states "Please note that due to lack of time we currently have no plans to extend DocFetcher in any significant way.".