FishEye contains a powerful query language called EyeQL. EyeQL is an intuitive SQL-like language that allows you to write your own specific queries. See examples.
EyeQL allows you to perform complex searches either within the Advanced Search or incorporated in scripts when programming the FishEye API.
(from (dir|directory) word)?
(order by date (asc | desc)? )?
Notes: asc produces 'ascending order'.
desc produces 'descending order'.
(group by (file|dir|directory|csid|changeset))?
clause ((or|and|,) clause)*
and binds more tightly than or.
',' (comma) means 'and'.
( clauses )
path (not)? like word
word is an Antglob.
path = word
Defines an exact path without wildcards or variables. path must represent a complete (hard-coded) path.
path != word
Defines an exact path exclusion without wildcards or variables. path must represent a complete (hard-coded) path.
date in ( ( | [ ) dateExp, dateExp ( ) | ] )
Notes: The edges are
] is used.
) is used.
date dateop dateExp
dateop can be
<, >, <=, >=, =, == or != .
author = word
author in (word-list)
comment matches word
Does a full-text search.
comment = string
Matches string exactly.
Most comments end in a new line, so remember to add
\n at the end of your string.
comment =~ string
string is a regular expression.
content matches word
Does a full-text search.
At this time searches are restricted to HEAD revisions.
(modified|added|deleted)? on branch word
Selects all revisions on a branch.
modified excludes the branch-point of a branch.
added selects all revisions on the branch if any revision was added on the branch.
deleted selects all revisions on the branch if any revision was deleted on the branch.
tagged op? word
op can be
<, >, <=, >=, =, == or != .
op defaults to
== if omitted.
These operators are 'positional' and select revisions that appear on, after, and/or before the given tag.
between tags tag-range
after tag word
before tag word
is head (on word)?
This selects the top-most revision on any branch, if no branch is specified.
is ( dead | deleted )
Means the revision was removed/deleted.
Means the revision was added (or re-added).
csid = word
Selects all revisions for the given changeset ID.
p4:jobid = word
Notes: finds revisions whose Perforce jobid is word.
p4:jobid =~ word
Notes: finds revisions whose Perforce jobid matches regex word.
Notes: (applies to ) alias for in or before any closed review.
(in | before | in or before) review word
(in | before | in or before) any (review states)? review
word is a review key.
in selects reviewed revisions. If a review contains a diff, then only the most recent revision is considered in the review.
before selects all revisions in a file prior to the revision in the review.
review states is a comma-separated list of open, closed, draft.
( ( | [ ) T1:word, T2:word ( ) | ] )
A range of revisions between those tagged T1 and T2.
The edges are:
] is used.
) is used.
You can mix edge types. These are all valid:
Having trouble with Subversion tags? See How Tags Work in Subversion for more information.
Any string, or any non-quoted word that does not contain white space or any other separators.
A sequence enclosed in either
" (double quotes) or
' (single quotes).
The following escapes work:
\' \" \n \r \t \b \f.
Unicode characters can be escaped with
You can also specify strings in 'raw' mode like
r"foo". (Similar to Python's raw strings. See Python's own documentation).
return-clause (, return-clause)*
A return clause signifies that you want control over what data is returned/displayed.
( path | dir | directory | revision | author | date | comment | csid | isBinary | totalLines | linesAdded | linesRemoved | isAdded | isDeleted | isCopied | isMoved | tags | reviews | aggregate)
( as word )?
The attribute to return, optionally followed by a name to use for the column.
Notes: reviews applies to.
( count(revisions ) | count( binary-field ) | count(distinct other-field ) | sum( numeric-field ) | average( numeric-field ) | max( numeric-field ) | min( numeric-field ) )
The aggregate field to return.
binary-fields are isBinary, isAdded, isDeleted, isCopied, isMoved. e.g. count(isAdded) will return the number of added files.
numeric-fields are totalLines, linesAdded, linesRemoved.
other-field can be path, dir, author, date, csid, tags or reviews. e.g. count(distinct path) will return the number of unique paths. count(distinct tags) will return the number of unique tags.
group by is given, give sub-totals for each group.
group by clause, you can have:
group by changeset|csid clause:
return csid, comment, date, author, aggregates
group by file|path clause:
return path, aggregates
group by dir|directory clause:
return dir, aggregates
i.e. The EyeQL can contain a returns clause that contains all non-aggregate columns, or all aggregate columns.
Non-aggregate and aggregate columns can only be mixed if the columns are unique for the grouping.
( length | offset, length | length offset offset )
Notes: Limits the number of results to return. offset specifies the starting point of the truncated result set and length specifies the set length. offset is zero-based.
The following examples demonstrate using EyeQL to extract information from your repository.
Find files removed on the Ant 1.5 branch:
select revisions where modified on branch ANT_15_BRANCH and is dead group by changeset
As above, but just return the person and time the files were deleted:
select revisions where modified on branch ANT_15_BRANCH and is dead return path, author, date
Find files on branch and exclude delete files:
select revisions where modified on branch ANT_15_BRANCH and not is deleted group by changeset
Find changes made to Ant 1.5.x after 1.5FINAL:
select revisions where on branch ANT_15_BRANCH and after tag ANT_MAIN_15FINAL group by changeset
Find changes made between Ant 1.5 and 1.5.1:
select revisions where between tags (ANT_MAIN_15FINAL, ANT_151_FINAL] group by changeset
As above, but show the history of each file separately:
select revisions where between tags (ANT_MAIN_15FINAL, ANT_151_FINAL] group by file
Find Java files that are tagged ANT_151_FINAL and are head on the ANT_15_BRANCH: (i.e. files that haven't changed in 1.5.x since 1.5.1)
select revisions from dir /src/main where is head and tagged ANT_151_FINAL and on branch ANT_15_BRANCH and path like *.java group by changeset
Find changes made by conor to Ant 1.5.x since 1.5.0:
select revisions where between tags (ANT_MAIN_15FINAL, ANT_154] and author = conor group by changeset
Find commits that do not have comments:
select revisions from dir / where comment = "" group by changeset
Find the 10 most recent revisions:
select revisions order by date desc limit 10
Find the 5th, 6th & 7th revisions:
select revisions order by date limit 4, 3
Find commits between two dates:
select revisions where date in [2008-03-08, 2008-04-08]
Find revisions that do not have any associated review:
select revisions where (not in any review)
Return number of matched revisions, the number of files modified, authors who modified code, changesets, tags, and reviews:
As Sub-totals for each distinct changeset, Return csid, the author, date, comment, number of matched revisions, the number of files modified, the lines added/removed:
For each matched file, return the file name, number of matched revisions, the lines added/removed:
Show all the changesets with no review: