Instant Search query syntax

Instant Search is the full-text search over your source code, reached from the global search bar's File Contents category (%). See Find things fast for the search bar itself and the other categories; this page is the query language.

All searching is case insensitive.

What gets indexed

Instant Search reads an index built after parsing, updated incrementally as files change. Indexing runs in the background and doesn't block other work. The index spans all projects, so a new project over already-indexed code doesn't pay to index it again — but results are limited to the current project while one is open.

For C, C++, Java and Ada, code is split into terms following the language's own syntax and lexical rules. Other languages are split on whitespace and punctuation. So in this line:

foreach (i=1, i<10, i++)

foreach, i, 1 and 10 are each searchable terms.

Fields

Every term is filed under one of three fields. A bare query searches all three; a prefix restricts it to one.

Query	Matches
test	test in any field
string:test	test only inside string literals
identifier:test	identifiers named test
comment:test	comments containing test

Wildcards and fuzzy matching

Query	Matches
test*	anything beginning with test
te?t	te, any single character, then t
test*bug	anything beginning with test and ending with bug
test~	terms like test — a fuzzy match, so also best, quest

Boolean operators

Query	Matches
test this	test or this
test OR bug	same as test bug
test AND bug	both test and bug
+test bug	must contain test, may contain bug
test this NOT that	test this, excluding anything with that
test this -that	same as above
(test this) AND that	test or this, and also that

Notes

The syntax is based on Lucene's query syntax, which is richer than the subset above — most Lucene constructs work.

If results look stale or wrong, the index rather than the query is usually at fault; see Search results are wrong or stale.