Bugzilla – Full Text Bug Listing
| Summary: | FastPartitioner computePartitions / length 0 fails at starting boundary | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Product: | [Eclipse Project] Platform | Reporter: | M Dahlem <dahlem.marc> | ||||||
| Component: | Text | Assignee: | Platform-Text-Inbox <platform-text-inbox> | ||||||
| Status: | RESOLVED INVALID | QA Contact: | |||||||
| Severity: | normal | ||||||||
| Priority: | P3 | CC: | daniel_megert | ||||||
| Version: | 3.7 | Keywords: | needinfo | ||||||
| Target Milestone: | --- | ||||||||
| Hardware: | PC | ||||||||
| OS: | All | ||||||||
| Whiteboard: | |||||||||
| Attachments: |
|
||||||||
Could you attach an example or test case that allows to reproduce this? Can you tell where it actually surfaces as a problem to the user? Hi,
Regular eclipse users are not effected by this bug, it affects only developers.
In my case the code looked like this:
[..]
ITypedRegion partRegions = getPartitioning(DefaulPartitionerID, offset, false);
String partRegionName = region.getType();
}
if (!(partRegionName.equals(RelevantPartitionID)) {
//not relevant
return false;
}
ITypedRegion[] regions = computePartitioning(SubPartitionerID, offset, length, false);
//compute region names
[...]
if (regionNames.contains(RelevantSubPartID)) {
return true;
} else {
return false;
}
[...]
Lets say the document contains the default partitions "EMPTY LINE" und "TEXT LINE". Only "TEXT LINES" are relevant default partitions.
And in this textline are two SubPartitions: "CODE" and "COMMENT".
The method should return TRUE if the offset is in an textLine and the whole region (offset till offset + length) contains at minimum one subpartition "COMMENT".
But if the comment starts at an offset where the textline also starts, and the length is 0, the "getPartition(..)" gets an "TEXTLINE", but computePartition(..) returns an "DEFAULT CONTENT TYPE".
Example:
line 1: <-- EMPTY LINE
line 2:abcdefg <-- TEXTLINE (CODE)
line 3: <-- EMPTY LINE
line 4:/* hijklmn */ <-- TEXTLINE (COMMENT)
If we compute "getPartition(DEFAULTPARTITIONER, 14 (offset at begin of line 4), false)" the result is a Region with contenttype "TEXTLINE".
If we compute now "computePartitioning(SUBPARTIID, 14, 0, false);" the result is NOT "COMMENT", it is "default partition type".
I try to produce an runnable example and attach it to this bug.
Greetz
Adreamus
Created attachment 208935 [details]
partitionscanner that detects java comments
Created attachment 208936 [details]
JUnit testcase for the partitionscanner
This JUnit testcase has three tests:
(1) A test for the getPartition(partitioning, boundaryoffset, false)
(2) A test for the computePartitionings(partitioning, boundaryoffset, 0, false)
=> fails
(3) A test for checking, if the computedRegion and getRegion are the same for length 0 at an boundary offset
I looked at the test case and the current behavior is expected. If the position is between two partitions and the length is 0 we take the partition that is on the left. |
Build Identifier: M20110909-1335 Hi, By developing a plugin I found out, that the contenttype at an offset is not the same like the computed region with legthn 0 at this offset computed all with an FastPartitionner. It is not only the contenttype what differs, it is the same result by computing an partition at this offset. Like I could see in the code of FastPartitioner line 628 the for-loop fails, because startIndex and endIndex contains both the same value and therefore the partition is of the default partition type. "for (int i= startIndex; i < endIndex; i++) {[...]". Greetz Adreamus Reproducible: Always Steps to Reproduce: 1. create FastPartitionScanner and connect with an ID at an documentPorivder 2. set a comment /* */ as rule for a partition and add a comment to the document 3. compute regions at boundary 0 and length 0: ITypedRegion[] regions = computePartitioning(ID, 0, 0, false); 4. compute Partition at offset 0: ITypedRegion regtest = getPartition(ID,0,false); 5. compare to each other, you can find out, that the array contains only the default partition and the partition at offset the correct partition type.