Letizia Palladino 

A bit tricky as the standard XML file types will not extract text from these tags when they are embedded within attributes or nested in an inconsistent structure.  If you don't have a developer in-house who could create a custom XML filetype for you using the Filetype API then you might be able to handle them with preprocessing techniques using XSLT or scripting to isolate the desired content, then a similar post-processing technique afterwards to put the files back together again.

Another option might be to handle all the content using the Embedded Content Plain Text processor.  But this would mean regex rules for everything and this is really only helpful if your files are relatively simple.

You could also have a play with the Multilingual XML filetype as this has a feature to handle an XML file as a monolingual file.  That way you get the benefit of its ability to have XPath processing AND embedded content processing.  The downside with this approach is it's not as flexible as the XML filetype in Studio where you can handle the XPath with a lot more flexibility, but that one has no embedded content processing.

So much depends on the files you have and which approach is likely to work best.