In cases like:
var foo []io.Writer
var buf *bytes.Buffer
foo = append(foo, <>)
we weren't giving "buf" a good score. When comparing the candidate
type *bytes.Buffer to the (variadic) expected type []io.Writer we were
turning the candidate type into []*bytes.Buffer. However, of course,
[]*bytes.Buffer is not assignable to []io.Writer, so the types didn't
match. Now we instead turn the expected type []io.Writer into
io.Writer and compare to *bytes.Buffer.
I fixed the @rank test note to check that the candidates' scores are
strictly decreasing. Previously it would allow candidates with the
same score if they happened to be in the right order. This made it
easier to right a test for this issue, but also uncovered an issue
with untyped completion logic. I fixed it to do the untyped constant
check if _either_ the expected or candidate type is
untyped (previously it required the candidate type to be untyped).
Fixesgolang/go#35625.
Change-Id: I9a837d6a781669cb7a2f1d6d3d7f360c85be49eb
Reviewed-on: https://go-review.googlesource.com/c/tools/+/207518
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
go/parser.ParseFile can return both an AST and errors. We should still
be able to do import organization even if the AST contains errors, as
long as they are below the portion of the file that contains the import
block.
Change-Id: Id6b86171fca3e15d02910d1c6f4ce25e803754d0
Reviewed-on: https://go-review.googlesource.com/c/tools/+/207261
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
We used to need our own copy of astutil.AddNamedImport to use during
completion for a variety of reasons, but I think the major one was
needing to not format the whole file. The same problem applied to using
the imports package.
Happily, that was resolved in CL 205678. Now we can use the same
implementation on both paths. In addition to removing a bunch of code,
that means that unimported completions now add their imports in the
right place, respecting goimports grouping and the local configuration
setting.
Fixesgolang/go#35519.
Change-Id: I693c2e8b5ced9bac62b1febf1e2db23c770e5a7a
Reviewed-on: https://go-review.googlesource.com/c/tools/+/206881
Run-TryBot: Heschi Kreinick <heschi@google.com>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Add a special case for append() arguments so we infer the expected
type from the append() context. For example:
var foo []int
foo = append(<>)
We now infer the expected type at <> to be []int. We also support the
variadicity of append().
Change-Id: Ie0ef0007907fcb7992f9697cb90970ce4d9a66b8
Reviewed-on: https://go-review.googlesource.com/c/tools/+/205606
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Improve candidate ranking when completing the variadic parameter of
function calls.
Using the example:
func foo(strs ...string) {}
- When completing foo(<>), we prefer candidates of type []string or
string (previously we only preferred []string).
- When completing foo("hi", <>), we prefer candidates of type
string (previously we preferred []string).
- When completing foo(<>), we use a snippet to add on the "..."
automatically to candidates of type []string.
I also fixed completion tests to work properly when you have multiple
notes referring to the same position. For example:
foo() //@rank(")", a, b),rank(")", a, c)
Previously the second "rank" was silently overwriting the first
because they both refer to the same ")".
Fixesgolang/go#34334.
Change-Id: I4f64be44a4ccbb533fb7682738c759cbca3a93cd
Reviewed-on: https://go-review.googlesource.com/c/tools/+/205117
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
In CL 205501 I thoughtlessly set import name to package name, but really
we only want to name imports when goimports would do it. For now, it's
better to not name them and let the usual imports code add a name if
necessary.
Fixesgolang/go#35397.
Change-Id: Id0df866f95e5e86ed72b25fbd1a7224c79ee8084
Reviewed-on: https://go-review.googlesource.com/c/tools/+/205657
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Unimported completions now try to pull Packages from everywhere, not
just the transitive dependencies of the current package. That confused
the import formatting code, which only looked at deps. Pass the Package
along with the import suggestion, and use it when it's present.
Also change some error messages to be different for diagnostic purposes.
Fixesgolang/go#35359.
Change-Id: Ia8ca923e46723e855ddd2da7611e6eb13c02bb4f
Reviewed-on: https://go-review.googlesource.com/c/tools/+/205501
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Packages that aren't imported in the current file will often have been
used elsewhere, which means that gopls will have their type information
available. Expose loaded packages in the Snapshot, and try to use that
information when possible for unimported packages.
Change-Id: Icb672618a9f9ec31b9796f0c5da56ed3d2b38aa7
Reviewed-on: https://go-review.googlesource.com/c/tools/+/204824
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
When a user completes rand.<>, propose rand.Seed (from math/rand) and
rand.Prime (from crypto/rand), etc.
Because we don't necessarily have type checking information for
unimported packages, I had to add shortcut cases to a number of
functions around the completion code. Better suggestions welcome.
Change-Id: I7822dc75c86b24156963e7bdd959443f4f2748b1
Reviewed-on: https://go-review.googlesource.com/c/tools/+/204819
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Muir Manders <muir@mnd.rs>
When our expected type is a named type from another package, we now always
search that other package for completion candidates, even if it is not currently
imported.
Consider the example:
-- foo.go --
import "context"
func doSomething(ctx context.Context) {}
-- bar.go--
doSomething(<>)
"bar.go" doesn't import "context" yet, so normally you need to first import
"context" through whatever means before you get completion items from "context".
Now we notice that the expected type's package hasn't been imported yet and give
deep completions from "context".
Another use case is with literal completions. Consider:
-- foo.go --
import "bytes"
func doSomething(buf *bytes.Buffer) {}
-- bar.go--
doSomething(<>)
Now you will get a literal completion for "&bytes.Buffer{}" in "bar.go" even
though it hasn't imported "bytes" yet.
I had to pipe the import info around a bunch of places so the import is added
automatically for deep completions and literal completions.
Change-Id: Ie86af2aa64ee235038957c1eecf042f7ec2b329b
Reviewed-on: https://go-review.googlesource.com/c/tools/+/201207
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
In cases like:
type myInt int
const (
a = 1
b myInt = 2
)
var foo myInt = <>
We now prefer "b" over "a" since b's type matches the expected type
exactly.
Change-Id: I675934761cc17f6b303b63b4715b31dd1af7cea1
Reviewed-on: https://go-review.googlesource.com/c/tools/+/202737
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Now that we are using the memoize package to cache analysis results, we
can use that cache for suggested fixes.
Change-Id: I42905a6fe575f49d38979d53d58ea8ec59210ae0
Reviewed-on: https://go-review.googlesource.com/c/tools/+/203278
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
We now expect a type name when in the key or value of a *ast.MapType.
I also added an extra filter to expect a comparable type for the key.
Change-Id: I647cf4d791b2c0960ad3b12702b91b9bc168599b
Reviewed-on: https://go-review.googlesource.com/c/tools/+/197439
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
*ast.ArrayTypes are type expressions like "[]foo" or "[2]int". They
show up as standalone types (e.g. "var foo []int") and as part of
composite literals (e.g. "[]int{}"). I made the following
improvements:
- Always expect a type name for array types.
- Add a "type modifier" for array types so completions can be smart
when we know the expected type. For example:
var foo []int
foo = []i<>
we know we want a type name, but we also know the expected type is
"[]int". When evaluating type names such as "int" we turn the type
into a slice type "[]int" to match against the expected type.
- Tweak the AST fixing to add a phantom selector "_" after a naked
"[]" so you can complete directly after the right bracket.
I split out the type name related type inference bits into a separate
typeNameInference struct. It had become confusing and complicated,
especially now that you can have an expected type and expect a type
name at the same time.
Change-Id: I00878532187ee5366ab8d681346532e36fa58e5f
Reviewed-on: https://go-review.googlesource.com/c/tools/+/197438
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Remove the input type option. Now everything behaves as "symbol".
We don't use the "text" or "filename" input types, and I don't foresee
us using them. Removing them simplifies the code a bit, but simplifies
the tests a lot. It was tedious to make changes to the matcher logic
because you had to fret over test failure details that didn't actually
matter because we didn't use that functionality.
Change-Id: I651debde9e63ee283d7bc3ad718d22f4b9a127c0
Reviewed-on: https://go-review.googlesource.com/c/tools/+/202637
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
For *ast.Ident completion requests, this checks the parent node to
see if the token begins a statement and then based on the path adds
possible keyword completion candidates. The test lists some cases where
this approach cannot provide completion candidates.
The biggest thing missing is keywords for file level declarations
Updates golang/go#34009
Change-Id: I9d9c0c1eb88e362613feca66d0eea6b88705b9b0
Reviewed-on: https://go-review.googlesource.com/c/tools/+/196664
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Now we offer completion candidates for labels when completing "break",
"continue", and "goto" statements. We are reasonably smart about
filtering unusable labels, except we don't filter "goto" candidates
that jump across variable definitions.
Fixesgolang/go#33987.
Change-Id: If296a7579845aba5d86c7050ab195c35d4b147ed
Reviewed-on: https://go-review.googlesource.com/c/tools/+/197417
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
This will prevent us from panicking in cases with errors.
Fixesgolang/go#34824
Change-Id: I02c20655f6926ec00c1591a905ff5a107cc44192
Reviewed-on: https://go-review.googlesource.com/c/tools/+/200300
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Ian Cottrell <iancottrell@google.com>
In cases like "fmt.Pr<>int()" we previously would replace "Print" with
the new completion, yielding for example "fmt.Println()". Now we no
longer overwrite, yielding "fmt.Println()int()". There are some cases
where overwriting the suffix is what the user wants, but it is hard to
tell, so for now stick with the more expected behavior of not
overwriting.
Fixesgolang/go#34011.
Change-Id: I8c3ccd8948245c27b52408ad508d8e01dc163ef4
Reviewed-on: https://go-review.googlesource.com/c/tools/+/196119
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
This change does not complete the work to handle snapshots correctly,
but it does implement the behavior of re-building the snapshot on each
file invalidation.
It also moves to the approach of caching the FileHandles on the snapshot,
rather than in the goFile object, which is now not necessary.
Finally, this change shifts the logic of metadata invalidation into the
content invalidation step, so there is less logic to decide if we should
re-load a package or not.
Change-Id: I18387c385fb070da4db1302bf97035ce6328b5c3
Reviewed-on: https://go-review.googlesource.com/c/tools/+/197799
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Cottrell <iancottrell@google.com>
In CL 192137 deep fuzzy matching was enabled by default. We also have
options independent options "deepCompletion" and "fuzzyMatching" to
control this. When fuzzy matching is disabled, case insensitive prefix
matching is used.
Provide an option, "caseSensitiveCompletion", which allows for case
sensitive prefix matching when fuzzy matching is disabled.
Change-Id: I17c8fa310b2ef79e36cc2f7303e98870690b5903
Reviewed-on: https://go-review.googlesource.com/c/tools/+/194757
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Now we always expect type names inside of *ast.FieldList. This expands
the previous func signature logic to also work for *ast.StructType
and *ast.InterfaceType. For example, we will now prefer type names in
cases like:
type myStruct struct { i i<> }
Also, fix a check for anonymous fields to make sure the field is
actually embedded. This fixes cases like this to properly have no
completions:
type myStruct struct { i<> i }
where this will still give type name completions:
type myStruct struct { i<> }
I introduced a new error type source.ErrIsDefinition so source_test.go
could avoid erroring out on tests that make sure definition
identifiers have no completions.
Fixesgolang/go#34412.
Change-Id: Ib56cb52af639f2e2b132274d1f04f8074c0d9353
Reviewed-on: https://go-review.googlesource.com/c/tools/+/196560
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Fix objects defined in the function signature to only be completable
inside the function body. For example:
func (dog Dog) bark(d<>) { // Don't complete <> to "dog".
d<> // Do complete <> to "dog".
}
Change-Id: Ic9a2dc2ce6771212780f2d6af2221a67d203f35f
Reviewed-on: https://go-review.googlesource.com/c/tools/+/196559
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
source.DiagnosticSeverity and source.CompletionItemKind are duplicated
and not worth maintaining.
Change-Id: I8d6c8621a227855309c0977da59d8c9fa53617bf
Reviewed-on: https://go-review.googlesource.com/c/tools/+/197177
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Cottrell <iancottrell@google.com>
Now we will offer function literal completions when we know the
expected type is a function. For example:
sort.Slice(someSlice, <>)
will offer the completion "func(...) {}" which if selected will
insert:
func(i, j int) bool {<>}
I opted to use an abbreviated label "func(...) {}" because function
signatures can be quite long/verbose with all the type names in there.
The only interesting challenge is how to handle signatures that don't
name the parameters. For example,
func HandleFunc(pattern string, handler func(ResponseWriter, *Request)) {
does not name the "ResponseWriter" and "Request" parameters. I went
with a minimal effort approach where we try abbreviating the type
names, so the literal completion item for "handler" would look like:
func(<rw> ResponseWriter, <r> *Request) {<>}
where <> denote placeholders. The user can tab through quickly if they
like the abbreviations, otherwise they can rename them.
For unnamed types or if the abbreviation would duplicate a previous
abbreviation, we fall back to "_" as the parameter name. The user will
have to rename the parameter before they can use it.
One side effect of this is that we cannot support function literal
completions with unnamed parameters unless the user has enabled
snippet placeholders.
Change-Id: Ic0ec81e85cd8de79bff73314e80e722f10f8c84c
Reviewed-on: https://go-review.googlesource.com/c/tools/+/193699
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Add support for literal completion candidates such as "[]int{}" or
"make([]int, 0)". We support both named and unnamed types. I used the
existing type matching logic, so, for example, if the expected type is
an interface, we will suggest literal candidates that implement the
interface.
The literal candidates have a lower score than normal matching
candidates, so they shouldn't be disruptive in cases where you don't
want a literal candidate.
This commit adds support for slice, array, struct, map, and channel
literal candidates since they are pretty similar. Functions will be
supported in a subsequent commit.
I also added support for setting a snippet's final tab stop. This is
useful if you want the cursor to end up somewhere other than the
character after the snippet.
Change-Id: Id3b74260fff4d61703989b422267021b00cec005
Reviewed-on: https://go-review.googlesource.com/c/tools/+/193698
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Parse errors need to be treated separately from actual errors when
parsing a file. Parse errors are treated more like values, whereas
actual errors should not be propagated to the user. This enables us to
delete some of the special handling for context.Canceled errors.
Change-Id: I93a02f22b3f54beccbd6bcf26f04bb8da0202c25
Reviewed-on: https://go-review.googlesource.com/c/tools/+/195997
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Cottrell <iancottrell@google.com>
We had too many options for functions to use to get type information for
a package. Now we stick with having one option to get the check package
handles, and then the caller can refine the results as needed.
Change-Id: I81f69a670e1539854ee23b6f364159a7de9b782f
Reviewed-on: https://go-review.googlesource.com/c/tools/+/194457
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Cottrell <iancottrell@google.com>
Tweak a couple things to improve how we reduce our search scope based
on remaining time budget:
- Check our budget on the first candidate rather than waiting for the
1000th candidate. If type checking is slow you can be out of budget
before you even begin.
- Reduce our budget check interval from 1000 candidates to 100
candidates. This just helps us adjust our search scope faster.
The first tweak required me to raise the completion budget for tests
because 100ms is not always enough. I moved the budget into the
completion options so that tests can raise it.
Change-Id: I1aa7909d7baf9c998bc830c960dc579eb33db12a
Reviewed-on: https://go-review.googlesource.com/c/tools/+/195419
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Revert my previous change to include fuzzy matches with a score of
zero. Zero scorers have some characters that match, but they are
pretty poor overall. Pulling in all the extra junk candidates was
slowing things down in certain cases.
Change-Id: I560f46903281f21b0628c9b14848cddf1e3c0a38
Reviewed-on: https://go-review.googlesource.com/c/tools/+/195418
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
A mapper is always uniquely tied to a file at a specific version, so
just build it when we get a new *ast.File. We build the mapper using the
*token.File associated with the particular *ast.File, which is why there
is one per ParseGoHandle instead of FileHandle.
Change-Id: Ida40981ef91f6133cdd07e9793337fcd67510fba
Reviewed-on: https://go-review.googlesource.com/c/tools/+/194517
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Ian Cottrell <iancottrell@google.com>
In the case of documentation items for completion items, we should make
sure to use the ASTs and type information for the originating package.
To do this while avoiding race conditions, we have to do this by
breadth-first searching the top-level package and its dependencies.
Change-Id: Id657be969ca3e400bb2bbd769a82d88e91865764
Reviewed-on: https://go-review.googlesource.com/c/tools/+/194477
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Cottrell <iancottrell@google.com>
I moved the "usePlaceholders" config field on to CompletionOptions.
This way the completion code generates a single snippet with a little
conditional logic based on the "WantPlaceholders" option instead of
juggling the generation of two almost identical "plain" and
"placeholder" snippets at the same time. It also reduces the work done
generating completion candidates a little.
I also made a minor tweak to the snippet builder where empty
placeholders are now always represented as e.g "${1:}" instead of
"${1}" or "${1:}", depending on if you passed a callback to
WritePlaceholder() or not.
Change-Id: Ib84cc0cd729a11b9e13ad3ac4b6fd2d82460acd5
Reviewed-on: https://go-review.googlesource.com/c/tools/+/193697
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Downranking builtins causes weird interplay with other completion
candidates due to fuzzy matching. For example:
notNil := 123
var foo *int = nil<>
ranks "notNil" before "nil" in the builtin list, which is counter
productive.
Change it to not downrank builtins. In my testing with this change,
builtins never were ranked above lexical items with similar names. I
think this is because the "natural" order of completion items puts
builtins last, and we stable sort items by score, so their relative
order is preserved.
Change-Id: Ifbad02be205e3cb26c1d4ce500b77690e7ac5b04
Reviewed-on: https://go-review.googlesource.com/c/tools/+/193897
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
This cl is the first in a set that change the configuration behaviour.
This one should have no behaviour differences, but makes a lot of preparatory changes.
The same options are set to the same values in the same places.
The options are now stored on the Session instead of the Server
The View supports options, but does not have any yet.
Change-Id: Ie966cceca6878861686a1766d63bb8a78021259b
Reviewed-on: https://go-review.googlesource.com/c/tools/+/193726
Run-TryBot: Ian Cottrell <iancottrell@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
This feature has been in an experimental state for a long enough time
that I think we can enable it by default at master.
Change-Id: I9bbb8b41377719f0e97f608f6e5163a883a176b3
Reviewed-on: https://go-review.googlesource.com/c/tools/+/192259
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Cottrell <iancottrell@google.com>
The existing implementation did not suggest struct field names
when running completion from within a slice literal of
pointers. Now, struct field names are suggested in that case.
Fixesgolang/go#33211
Change-Id: I6028420a9a789846b070fcc6e45ec89dc4d898d4
Reviewed-on: https://go-review.googlesource.com/c/tools/+/192277
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Invert "useDeepCompletions" config flag to "disableDeepCompletion" and
separate out "disableFuzzyMatching" which reverts to the previous
prefix matching behavior.
I separated fuzzy matching tests out to a separate file so they aren't
entangled with deep completion tests. In coming up with representative
test cases I found a couple issues which I fixed:
- We were treating a fuzzy matcher score of 0 as no match, but the
matcher returns 0 for candidates that match but have no bonuses. I
changed the matcher interface so that a score of 0 counts as a
match. For example, this was preventing a pattern of "o" from
matching "foo".
- When we lower a candidate's score based on its depth, we were
subtracting a static multiplier which could result in the score
going negative. A negative score messes up future score weighting
because multiplying it by a value in the range [0, 1) makes it
bigger instead of smaller. Fix by scaling a candidate's score based
on its depth rather than subtracting a constant factor.
Updates golang/go#32754
Change-Id: Ie6f9111f1696b0d067d08f7eed7b0a338ad9cd67
Reviewed-on: https://go-review.googlesource.com/c/tools/+/192137
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Deep completions can take a long time (500ms+) if there are many
large, deeply nested structs in scope. To make sure we return
completion results in a timely manner we now notice if we have spent
"too long" searching for deep completions and reduce the search scope.
In particular, our overall completion budget is 100ms. This value is
often cited as the longest latency that still feels instantaneous to
most people. As we spend 25%, 50%, and 75% of our budget we limit our
deep completion candidate search depth to 4, 3, and 2,
respectively. If we hit 90% of our budget, we disable deep completions
entirely.
In my testing, limiting the search scope to 4 normally makes even
enormous searches finish in a few milliseconds. Of course, you can
have arbitrarily many objects in scope with arbitrarily many fields,
so to cover our bases we continue to dial down the search depth as
needed.
I replaced the "enabled" field with a "maxDepth" field that disables
deep search when set to 0.
Change-Id: I9b5a07de70709895c065503ae6082d1ea615d1af
Reviewed-on: https://go-review.googlesource.com/c/tools/+/190978
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Cottrell <iancottrell@google.com>
Optimize a few things to speed up deep completions:
- item() is slow, so don't call it unless the candidate's name matches
the input.
- We only end up returning the top 3 deep candidates, so skip deep
candidates early if they are not in the top 3 scores we have seen so
far. This greatly reduces calls to item(), but also avoids a
humongous sort in lsp/completion.go.
- Get rid of error return value from found(). Nothing checked for this
error, and we spent a lot of time allocating the only possible error
"this candidate is not accessible", which is not unexpected to begin
with.
- Cache the call to types.NewMethodSet in methodsAndFields(). This is
relatively expensive and can be called many times for the same type
when searching for deep completions.
- Avoid calling deepState.chainString() twice by calling it once and
storing the result on the candidate.
These optimizations sped up my slow completion from 1.5s to
0.5s. There were around 200k deep candidates examined for this one
completion. The remaining time is dominated by the fuzzy
matcher. Obviously 500ms is still unacceptable under any
circumstances, so there will be subsequent improvements to limit the
deep completion search scope to make sure we always return completions
in a reasonable amount of time.
I also made it so there is always a "matcher" set on the
completer. This makes the matching logic a bit simpler.
Change-Id: Id48ef7031ee1d4ea04515c828277384562b988a8
Reviewed-on: https://go-review.googlesource.com/c/tools/+/190522
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
this moves the actual diff algorithm into a different package and then provides hooks so it can be easily replaced with an alternate algorithm.
Change-Id: Ia0359f58878493599ea0e0fda8920f21100e16f1
Reviewed-on: https://go-review.googlesource.com/c/tools/+/190898
Run-TryBot: Ian Cottrell <iancottrell@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Unimported packages may be suggested as completion items. Since these
are not yet imported, they should be ranked lower than other candidates.
They also require an additional import statement to be valid, which is
provided as an AdditionalTextEdit.
Adding this import does not use astutil.AddNamedImport, to avoid
editing the current ast and work even if there are errors. Additionally,
it can be hard to determine what changes need to be made to the source
document from the ast, as astutil.AddNamedImport includes a merging
pass. Instead, the completion item simply adds another import
declaration.
Change-Id: Icbde226d843bd49ee3713cafcbd5299d51530695
Reviewed-on: https://go-review.googlesource.com/c/tools/+/190338
Run-TryBot: Suzy Mueller <suzmue@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
This is a straight move of some code with no changes.
It splits the part of the telemetry code that will become a standalone library from the bit that belongs in the lsp.
Change-Id: Icedb6bf1f3711da9251450531729984df6df7787
Reviewed-on: https://go-review.googlesource.com/c/tools/+/190403
Run-TryBot: Ian Cottrell <iancottrell@google.com>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Make use of the existing fuzzy matcher to perform server side fuzzy
completion matching. Previously the server did exact prefix matching
for completion candidates and left fancy filtering to the
client. Having the server do fuzzy matching has two main benefits:
- Deep completions now update as you type. The completion candidates
returned to the client are marked "incomplete", causing the client
to refresh the candidates after every keystroke. This lets the
server pick the most relevant set of deep completion candidates.
- All editors get fuzzy matching for free. VSCode has fuzzy matching
out of the box, but some editors either don't provide it, or it can
be difficult to set up.
I modified the fuzzy matcher to allow matches where the input doesn't
match the final segment of the candidate. For example, previously "ab"
would not match "abc.def" because the "b" in "ab" did not match the
final segment "def". I can see how this is useful when the text
matching happens in a vacuum and candidate's final segment is the most
specific part. But, in our case, we have various other methods to
order candidates, so we don't want to exclude them just because the
final segment doesn't match. For example, if we know our candidate
needs to be type "context.Context" and "foo.ctx" is of the right type,
we want to suggest "foo.ctx" as soon as the user starts inputting
"foo", even though "foo" doesn't match "ctx" at all.
Note that fuzzy matching is behind the "useDeepCompletions" config
flag for the time being.
Change-Id: Ic7674f0cf885af770c30daef472f2e3c5ac4db78
Reviewed-on: https://go-review.googlesource.com/c/tools/+/190099
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
This change eliminates the need for the package cache map, and instead
stores package type information in the store. We still have to maintain
invalidation logic because the key is not computed correctly.
Change-Id: I1c2a7502b99491ef0ff68d68c9f439503d531ff1
Reviewed-on: https://go-review.googlesource.com/c/tools/+/185438
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Cottrell <iancottrell@google.com>
This relates to https://github.com/golang/go/issues/31374 and should switch all instances within `gopls` to use `x/errors` instead of `fmt` to create new errors.
Change-Id: I18339b75d12418d852e0dcc2ba0ed6c2970783b3
GitHub-Last-Rev: f4a55d9b79e7458ef1f1e06cb5eabbabd884f321
GitHub-Pull-Request: golang/tools#108
Reviewed-on: https://go-review.googlesource.com/c/tools/+/179880
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
