Specific Attribute node replacement - xquery

Question

How to do node replacement in MarkLogic for a particular attribute? For example like below:
<chapters>
<title id="primary">first primary content</title>
<title id="primary">second primary content</title>
<title id="secondary">this is amy middle content</title>
<title id="terciary">this is amy last content</title>
</chapters>
I want like below:
<chapters>
<title id="primary">third primary content</title>
<title id="secondary">this is amy middle content</title>
<title id="terciary">this is amy last content</title>
</chapters>
I mean suppose A.xml file stored in MarkLogic database server that contain data like bleow:
<chaptermetadata>
<chapters>
<title id="primary">first content</title>
<title id="primary">second content</title>
<title id="secondary">This is middle content</title>
<title id="terciary">This is last content</title>
</chapters>
<chapters>
<title id="primary">fouth content</title>
<title id="primary">fifth content</title>
<title id="primary">sixth content</title>
<title id="secondary">This is new content</title>
<title id="terciary">This is old content</title>
</chapters>
<chaptermetadata>
Now, I want to replace a node in all the element title which contain attribute #id='primary' in all chapter like below:
<chaptermetadata>
<chapters>
<title id="primary">common content</title>
<title id="secondary">This is middle content</title>
<title id="terciary">This is last content</title>
</chapters>
<chapters>
<title id="primary">common content</title>
<title id="secondary">This is new content</title>
<title id="terciary">This is old content</title>
</chapters>
<chaptermetadata>

Answer 1

If you are just getting started with XQuery and MarkLogic, http://developer.marklogic.com/learn/technical-overview and http://developer.marklogic.com/learn may help.
The best way to modify elements and attributes depends on the context, which you haven't supplied. I suppose the first question is "how do I select nodes by attribute?" A simple bit of XPath does that. For all chapters in the database:
/chapters/title[#id eq $id]
...or relative to a previously selected sequence of element(chapter)*
$chapters/title[#id eq $id]
If this is a database document, you could take it from there with the http://docs.marklogic.com/xdmp:node-replace and http://docs.marklogic.com/xdmp:node-delete functions. If the nodes are only in memory, see http://docs.marklogic.com/guide/app-dev/typeswitch for guidance and examples on using an XQuery typeswitch or XSLT. At http://developer.marklogic.com/blog/tired-of-typeswitch there are more examples and comparison of typeswitch and XSLT.

Answer 2

Based upon the examples you provided, it appears that you want to replace the text() for the first title element that has #id='primary', and remove the othertitleelements with#id='primary'`.
The following will achieve that, using xdmp:node-replace() and xdmp:node-delete() methods.
for $primary-title in doc("A.xml")/chaptermetadata/chapters/title[#id='primary']
return
if ($primary-title/preceding-sibling::title[#id='primary'])
then xdmp:node-delete($primary-title)
else xdmp:node-replace($primary-title/text(), text{"common content"})