Question

I am looking for XML/XHTML Java library/framework that can perform the following two tasks for me.

Before going on few definitions:

• NodeOffset(Node node, int offset) marks some point in text node in the XML tree.
• nodeB, nodeI, nodeP are the corresponding Node instances of the below mentioned XHTML tree and nodeSpan is some newly created node (where Node is not necessarily org.w3c.dom.Node and may be any other abstraction)

Flattering XHTML into plain text

The library should be able to produce plaintext output (e.g. by implementing CharSequence or similar) from given XHTML and provide one-to-one mapping between chars in the output and original XHTML node tree (e.g. via the function NodeOffset getNodeOffset(int plainTextOffset)).

Example: Suppose we have the following XHTML:

<p><b>GeForce</b> 9300M GS provides powerful <i>visual computing features</i> to thin and light notebooks.</p>

Then the plaintext representation will obviously be:

GeForce 9300M GS provides powerful visual computing features to thin and light notebooks.

Then e.g.

• getNodeOffset(0) should return node NodeOffset(nodeB, 0)
• getNodeOffset(40) should return node NodeOffset(nodeI, 5)
• getNodeOffset(80) should return node NodeOffset(nodeP, 49).

I might miss the correct numbers, but I hope, you got the idea. I repeat the example, now with pseudo-markers inserted:

|GeForce 9300M GS provides powerful visua|l computing features to thin and light n|otebooks.

and

<p><b>|GeForce</b> 9300M GS provides powerful <i>visua|l computing features</i> to thin and light n|otebooks.</p>

Node manipulating

The library should provide a possibility to inject nodes into XHTML, that may span the tree possibly crossing the node boundaries e.g. via the operation NodeSet insert(Node nodeToInsert, NodeOffset start, NodeOffset end, int mode). The function works in two modes:

• mode1: Split the node to be inserted if necessary. In this case the splitted from nodeToInsert nodes are returned as operations result.
• mode2: Close the parent nodes. nodeToInsert is returned as is.

For example: the insert(nodeSpan, NodeOffset(nodeB, 2), NodeOffset(nodeP, 9), mode1) operation should produce

<p><b>Ge<span>Force</span></b><span> 9300M GS</span> provides powerful <i>visual computing features</i> to thin and light notebooks.</p>

insert(nodeSpan, NodeOffset(nodeB, 2), NodeOffset(nodeP, 9), mode2) operation should produce:

<p><b>Ge</b><span><b>Force</b> 9300M GS</span> provides powerful <i>visual computing features</i> to thin and light notebooks.</p>

It is analogue to what users do in rich editor:

GeForce 9300M GS

I wonder, if there is anything like this in OpenSource world, as I really don’t want to re-implement the wheel… I’ve checked quickly Open Source HTML Parsers in Java without success.

When you post an answer:

• Make sure the above mentioned functions are available in library API (provide a link to JavaDoc).
• The library is Java-native (no JNI) and OpenSource.

Answer 1

I wrapped code that I had already, with modifications to match your requests (WIP) in an open-source project: ShtutXML. It’s pretty documented, so I doubt you’ll have a problem using it.

The first request (Finding a node and offsets from a global position) is already built in, and splitting of text nodes in the XML is already built in (so you can easily wrap them in new nodes as you wish). Therefore, adding the logic for marking areas with an element is rather trivial. I’ll try to do it later, but this is my best effort on this request for now.

On your XML, using my example program this is my output:

************* BASE DOCUMENT *****************
DOCUMENT ROOT
|-<p >
| |-<b >
| | |-Text: GeForce
| |-Text:  9300M GS provides powerful 
| |-<i >
| | |-Text: visual computing features
| |-Text:  to thin and light notebooks.

*** Text ***
"GeForce 9300M GS provides powerful visual computing features to thin and light notebooks."

*** Node of each text segment ***
[b: null]: GeForce
[p: null]:  9300M GS provides powerful 
[i: null]: visual computing features
[p: null]:  to thin and light notebooks.


*** Offset testing ***
offset 0 is at [b: null] at 0
offset 40 is at [i: null] at 5
offset 80 is at [p: null] at 48

Asking it to split the element at the global position 4 will produce

*********** Split(4) DOCUMENT *****************
DOCUMENT ROOT
|-<p >
| |-<b >
| | |-Text: GeFo
| | |-Text: rce
| |-Text:  9300M GS provides powerful 
| |-<i >
| | |-Text: visual computing features
| |-Text:  to thin and light notebooks.

*** Node of each text segment ***
[b: null]: GeFo
[b: null]: rce
[p: null]:  9300M GS provides powerful 
[i: null]: visual computing features
[p: null]:  to thin and light notebooks.

Of course this syntactical split means nothing for the actual XML code that matches that document, but it will allow wrapping one text part at a time with any other node you wish.

Edit: The first insertion mode is already supported

Edit 2: The second insertion mode is already supported

Notes:

• Any document modification you may do, will make all the offsetts invalid. Using them later will cause corruption of the entire document. So, after each modification you must do GetOffset to retreive the offsetts again.
• I know some of the functions are not documented. Basically the only functions that should be used outside of the package are the ones you requested from the StrXML class. More documentation will be added later and you can contact me by email (see my profile page) for questions.