Search does not return expected results
Platform Notice: Data Center Only - This article only applies to Atlassian products on the Data Center platform.
Note that this KB was created for the Data Center version of the product. Data Center KBs for non-Data-Center-specific features may also work for Server versions of the product, however they have not been tested. Support for Server* products ended on February 15th 2024. If you are running a Server product, you can visit the Atlassian Server end of support announcement to review your migration options.
*Except Fisheye and Crucible
Problem
When performing a search in Fisheye the results returned are not what is expected.
Example replication steps
Commit code to a repo such as the following:
1@SomeClass(Arg, Arg.Scanning.C, Arg2Scanned)Commit more code to the repo that build off the previous revision:
1[@SomeClass(Arg, Arg.Scanning.C, Arg2Scanned)]Use
@in Quick search to look for all reposUse Simple search within the repo and search for
@Use Simple search within the repo and search for
[@
Results
Quick search returns both revisions
Simple search for
@returns no contentSimple search for
[@returns only the second revision
Explanation
Fisheye/Crucible during indexing process creates special tokens, that are used in search. Tokens for content of files usually represents single words. Basically in order to find anything in text, user has to type whole word in search area.
To find a new token, each single character is checked:
each character is put into one of three groups:
standard - it contains all standard letters and digits
operators - these are chars, that can have special meaning (currently it is a set of:
+ - * / % \ = < > ! ~ . & & | ^[ ]_ @)pauses - all characters that separate different words - spaces, commas, semicolons, brackets etc.
each token contains characters from the same group
if next character is from another group, search ends and token is created
characters from pause group are omitted
When we consider @SomeClass(Arg, Arg.Scanning.C, Arg2Scanned) it will be divided into tokens: '@', 'SomeClass', 'Arg', '.', 'Scanning', '.', 'C', 'Arg2Scanned'.
If we add '[' at the beginning, the first token will look like '[@', because both characters are operators and characters from the same group are merged into one token.
All tokens have to be found in the content in order to display it. So if user types only '@', content which contains '[@' will not be displayed. If user types '[@', content which has only '@' will not be displayed as well.
Was this helpful?