[
https://issues.apache.org/jira/browse/XERCESJ-1400?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ludger Bünger updated XERCESJ-1400:
-----------------------------------
Description:
Ok, there are a couple of issues (bugs, non-standard-compliant behaviour and
maintenance) in Xerces Range implementation I created fixes for which I like to
provide to the community.
These Fixes are already in use in our application depending heavily upon
Ranges and have been deployed successfully at customers since more than two
years.
Bug 1:
Quite a lot of Range methods are not aware of character type Nodes other than
Text.
Described more in detail in [1], the Ranges spec differentiate between two
type of Nodes.
a) Nodes of type CharacterData (these include Text), Comment or
ProcessingInstruction which are treated the same in respect to the Ranges spec.
b) other Nodes.
The inital implementation of Ranges only considered the case of Text nodes
ignoring the other character containing node types.
With the Fixes for Bug #1228 and #1229 this has been improved for specific
cases but not in general.
The following methods of RangeImpl are not aware of all above mentioned
character containing nodes:
* RangeImpl.insertNode
* RangeImpl.surroundContent
* RangeImpl.selectNodeContent
* RangeImpl.toString
* RangeImpl.receiveSplitData
* RangeImpl.traverseNode
* RangeImpl.getSelectedNode
The following methods are correctly aware of them and have already been fixed
by Michael:
* RangeImpl.checkindex
* RangeImpl.receiveDeletedText
* RangeImpl.receiveInsertedText
* RangeImpl.receiveSplitData
* RangeImpl.traverseSameContainer
Bug 2:
selectNodeContents does not surround character containing nodes.
consider following code snippet.
Text text = doc.createNode();
text.setData("String"); // length of string: 6
range.selectNodeContent(text);
received result: (text, 0, text, 0);
correct result: (text, 0, text, 6);
Bug 3:
surroundContent(newParent) currently does not remove children of newParent as
specified in [2]:
"If the node newParent has any children, those children are removed before its
insertion. Also, if the node newParent already has a parent, it is removed from
the original parent's childNodes list."
Bug 4:
Text.textSplit does not update Ranges at all.
Solution: notify document using document.splitData after text split to cause
range update.
Bug 5:
After calling insertNode Ranges containing the same affected node are not
updated correctly.
Suggested Solution: use textSplit instead of doing a manual text split inside
RangeImpl and set content of this range afterwards.
This does work now due to Bug 4 being fixed.
However textSplit is only defined upon Nodes of type Text, while ranges might
split any character data containing nodes.
Thus I suggest to extract the text split operation to an internal method that
is implemented upon CharacterDataImpl used by both Text.textSplit and
Range.insertNode.
Maintenance 6:
Obsolete methods.
The following methods are never used and thus obsolete:
* deleteData
* insertData
* removeChild
* signalSplitData
Maintenance 7:
Obsolete fields.
After removing methods from Maintenance issue 6, the following fields are never
written and thus obsolete:
* fDeleteNode
* fInsertedFromRange
* fRemoveChild
* fSplitNode
I suggest removing these.
Range update works perfectly well without.
[1]
http://www.w3.org/TR/2000/REC-DOM-Level-2-Traversal-Range-20001113/ranges.html#Level-2-Range-Position
[2]
http://www.w3.org/TR/2000/REC-DOM-Level-2-Traversal-Range-20001113/ranges.html#Level-2-Range-Surrounding
was:
Ok, there are a couple of issues (bugs, non-standard-compliant behaviour and
maintenance) in Xerces Range implementation I created fixes for which I like to
provide to the community.
These Fixes are already in use in our application depending heavily upon
Ranges and have been deployed successfully at customers since more than two
years.
Bug 1:
Quite a lot of Range methods are not aware of character type Nodes other than
Text.
Described more in detail in [1], the Ranges spec differentiate between two
type of Nodes.
a) Nodes of type CharacterData (these include Text), Comment or
ProcessingInstruction which are treated the same in respect to the Ranges spec.
b) other Nodes.
The inital implementation of Ranges only considered the case of Text nodes
ignoring the other character containing node types.
With the Fixes for Bug #1228 and #1229 this has been improved for specific
cases but not in general.
The following methods of RangeImpl are not aware of all above mentioned
character containing nodes:
* RangeImpl.insertNode
* RangeImpl.surroundContent
* RangeImpl.selectNodeContent
* RangeImpl.toString
* RangeImpl.receiveSplitData
* RangeImpl.traverseNode
* RangeImpl.getSelectedNode
The following methods are correctly aware of them and have already been fixed
by Michael:
* RangeImpl.checkindex
* RangeImpl.receiveDeletedText
* RangeImpl.receiveInsertedText
* RangeImpl.receiveSplitData
* RangeImpl.traverseSameContainer
Bug 2:
selectNodeContents does not surround character containing nodes.
consider following code snippet.
Text text = doc.createNode();
text.setData("String"); // length of string: 6
range.selectNodeContent(text);
received result: (text, 0, text, 0);
correct result: (text, 0, text, 6);
Bug 3:
surroundContent(newParent) does remove children of newParent as sepcified in
[2].
"If the node newParent has any children, those children are removed before its
insertion. Also, if the node newParent already has a parent, it is removed from
the original parent's childNodes list."
Bug 4:
Text.textSplit does not update Ranges at all.
Solution: notify document using document.splitData after text split to cause
range update.
Bug 5:
After calling insertNode Ranges containing the same affected node are not
updated correctly.
Suggested Solution: use textSplit instead of doing a manual text split inside
RangeImpl and set content of this range afterwards.
This does work now due to Bug 4 being fixed.
However textSplit is only defined upon Nodes of type Text, while ranges might
split any character data containing nodes.
Thus I suggest to extract the text split operation to an internal method that
is implemented upon CharacterDataImpl used by both Text.textSplit and
Range.insertNode.
Maintenance 6:
Obsolete methods.
The following methods are never used and thus obsolete:
* deleteData
* insertData
* removeChild
* signalSplitData
Maintenance 7:
Obsolete fields.
After removing methods from Maintenance issue 6, the following fields are never
written and thus obsolete:
* fDeleteNode
* fInsertedFromRange
* fRemoveChild
* fSplitNode
I suggest removing.
Range update works perfectly well without these.
[1]
http://www.w3.org/TR/2000/REC-DOM-Level-2-Traversal-Range-20001113/ranges.html#Level-2-Range-Position
[2]
http://www.w3.org/TR/2000/REC-DOM-Level-2-Traversal-Range-20001113/ranges.html#Level-2-Range-Surrounding
Added missing NOT in description of Bug 3 (description did not make sense
without).
> Range implementation: several bugs (see detailed Description) and code
> Maintenance
> ----------------------------------------------------------------------------------
>
> Key: XERCESJ-1400
> URL: https://issues.apache.org/jira/browse/XERCESJ-1400
> Project: Xerces2-J
> Issue Type: Bug
> Components: DOM (Level 2 Traversal & Range)
> Affects Versions: 2.9.1
> Reporter: Ludger Bünger
> Attachments: Xerces-Range-Maintenance-Patch.txt
>
>
> Ok, there are a couple of issues (bugs, non-standard-compliant behaviour and
> maintenance) in Xerces Range implementation I created fixes for which I like
> to provide to the community.
> These Fixes are already in use in our application depending heavily upon
> Ranges and have been deployed successfully at customers since more than two
> years.
> Bug 1:
> Quite a lot of Range methods are not aware of character type Nodes other than
> Text.
> Described more in detail in [1], the Ranges spec differentiate between two
> type of Nodes.
> a) Nodes of type CharacterData (these include Text), Comment or
> ProcessingInstruction which are treated the same in respect to the Ranges
> spec.
> b) other Nodes.
> The inital implementation of Ranges only considered the case of Text nodes
> ignoring the other character containing node types.
> With the Fixes for Bug #1228 and #1229 this has been improved for specific
> cases but not in general.
> The following methods of RangeImpl are not aware of all above mentioned
> character containing nodes:
> * RangeImpl.insertNode
> * RangeImpl.surroundContent
> * RangeImpl.selectNodeContent
> * RangeImpl.toString
> * RangeImpl.receiveSplitData
> * RangeImpl.traverseNode
> * RangeImpl.getSelectedNode
> The following methods are correctly aware of them and have already been fixed
> by Michael:
> * RangeImpl.checkindex
> * RangeImpl.receiveDeletedText
> * RangeImpl.receiveInsertedText
> * RangeImpl.receiveSplitData
> * RangeImpl.traverseSameContainer
> Bug 2:
> selectNodeContents does not surround character containing nodes.
> consider following code snippet.
> Text text = doc.createNode();
> text.setData("String"); // length of string: 6
> range.selectNodeContent(text);
> received result: (text, 0, text, 0);
> correct result: (text, 0, text, 6);
> Bug 3:
> surroundContent(newParent) currently does not remove children of newParent as
> specified in [2]:
> "If the node newParent has any children, those children are removed before
> its insertion. Also, if the node newParent already has a parent, it is
> removed from the original parent's childNodes list."
> Bug 4:
> Text.textSplit does not update Ranges at all.
> Solution: notify document using document.splitData after text split to cause
> range update.
> Bug 5:
> After calling insertNode Ranges containing the same affected node are not
> updated correctly.
> Suggested Solution: use textSplit instead of doing a manual text split inside
> RangeImpl and set content of this range afterwards.
> This does work now due to Bug 4 being fixed.
> However textSplit is only defined upon Nodes of type Text, while ranges might
> split any character data containing nodes.
> Thus I suggest to extract the text split operation to an internal method that
> is implemented upon CharacterDataImpl used by both Text.textSplit and
> Range.insertNode.
> Maintenance 6:
> Obsolete methods.
> The following methods are never used and thus obsolete:
> * deleteData
> * insertData
> * removeChild
> * signalSplitData
> Maintenance 7:
> Obsolete fields.
> After removing methods from Maintenance issue 6, the following fields are
> never written and thus obsolete:
> * fDeleteNode
> * fInsertedFromRange
> * fRemoveChild
> * fSplitNode
> I suggest removing these.
> Range update works perfectly well without.
> [1]
> http://www.w3.org/TR/2000/REC-DOM-Level-2-Traversal-Range-20001113/ranges.html#Level-2-Range-Position
> [2]
> http://www.w3.org/TR/2000/REC-DOM-Level-2-Traversal-Range-20001113/ranges.html#Level-2-Range-Surrounding
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
