Fix completion in the following cases:
type foo struct{}
// now we offer "&foo" instead of "foo"
var _ *foo = fo<>{}
struct { f *foo }{
// now we offer "&foo" instead of "*foo"
f: fo<>{},
}
Composite literal type names are a bit special because they are part
of an arbitrary value expression rather than just a standalone type
name expression. In particular, they can be preceded by "&", which
affects how they relate to the surrounding context. The "&" doesn't
technically apply to the type name, but we must take it into account.
I made three changes to fix the behavior:
1. When we want to make a composite literal type name into a pointer,
we use "&" instead of "*".
2. Record if a composite literal type is already has a "&" so we don't
add it again.
3. Fix "var _ *foo = fo<>{}" to properly infer expected type of "*foo"
by not stopping at *ast.CompositeLit searching up AST path when the
position is in the type name (as opposed to within the curlies).
Change-Id: Iee828f259eb939646b68f5066614ea3a262585c2
Reviewed-on: https://go-review.googlesource.com/c/tools/+/247525
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
We now rank printf operand candidates according to the corresponding
formatting verb. We follow what fmt allows for the most part, but I
omitted some things because they are uncommon and would result in many
false positives, or didn't seem worth it to support:
- We don't prefer fmt.Stringer or error types for "%x" or "%X".
- We don't prefer pointers for any verbs except "%p".
- We don't prefer recursive application of verbs (e.g. we won't prefer
[]string for "%s").
I decided against sharing code with the printf analyzer. It was
tangled somewhat with go/analysis, and I needed only a very small
subset of the format parsing.
I tweaked candidate type evaluation to accommodate the printf hints.
We now skip expected type of "interface{}" when matching candidate
types because it matches everything and would always supersede the
coarser object kind checks.
Fixesgolang/go#40485.
Change-Id: I6440702e33d5ec85d701f8be65453044b5dab746
Reviewed-on: https://go-review.googlesource.com/c/tools/+/246699
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
In this example:
p := &[]int{}
append([]int{}, *<>)
At <> we completed to "**p" instead of "*p...". There were two fixes:
1. builtinArgType() wasn't propagating the "modifiers", so we were
forgetting about the preceding "*" pointer indirection and
inserting it again with the completion. Fix by propagating
modifiers.
2. The candidate formatting responsible for adding "..." had over
simplified logic to determine if we are completing the variadic
param. Now instead the candidate evaluation code marks the
candidate as "variadic" so the formatting doesn't have to think at
all.
Change-Id: Ib71ee8ecfafb915df331f1d2e55b76f76a530243
Reviewed-on: https://go-review.googlesource.com/c/tools/+/248018
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
In the following example:
var foo []someStruct
foo = append(foo, <>)
we now downrank "foo" as a candidate at "<>". You very rarely append a
slice to itself, so having "foo" ranked highly was counterproductive.
Fixesgolang/go#40535.
Change-Id: Ic01366aeded4ba2b6b64bfddd33415499b35a323
Reviewed-on: https://go-review.googlesource.com/c/tools/+/247519
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Tweak a few things so that unnamed, non-basic types are offered as
completions in certain cases:
var _ []int = make(<>) // now properly suggests "[]int"
I also fixed type related keywords to not be offered if there is an
expected type:
var _ *int = new(<>) // don't offer "func", etc.
There are still some cases that don't work properly. For example:
var _ [][]int = make([]<>) // doesn't offer "[]int"
This would be harder to fix given the way things currently work.
Fixesgolang/go#40275, golang/go#40274.
Change-Id: I2577d5863d4757845ad3ff7dbb125106b649a6b6
Reviewed-on: https://go-review.googlesource.com/c/tools/+/246360
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Now we downrank candidates that have already been used in other switch
cases. For example:
switch time.Now().Weekday() {
case time.Monday:
case time.<> // downrank time.Monday
}
It wasn't quite as simple as tracking the seen types.Objects.
Consider this example:
type foo struct {
i int
}
var a, b foo
switch 123 {
case a.i:
case <>
}
At <> we don't want to downrank "b.i" even though "i" is represented
as the same types.Object for "a.i" and "b.i". To accommodate this, we
track having seen ["a", "i"] together. We will downrank "a.i", but not
"b.i".
I applied only a minor 0.9 downranking when the candidate has already
been used in another case clause. It is hard to know what the user is
planning. For instance, in the preceding example the user could intend
to write "a.i + 1", so we mustn't downrank "a.i" too much.
Change-Id: I62debc5be3d5d310deb69d11770cf5f8bd9add1d
Reviewed-on: https://go-review.googlesource.com/c/tools/+/247337
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Now we will filter out the types already used in other case
statements:
switch ast.Node(nil).(type) {
case *ast.Ident:
case *ast.I<> // don't offer "Ident" since it has been used
}
Note that the implementation was not able to use a map to track the
seen types.Types because we build up types.Type entries dynamically
when searching for completions (e.g. types.NewPointer() to make a
pointer type). We must use types.Identical() instead of direct pointer
equality.
Change-Id: I316638bb48bfd6802e2caea671f297d640291010
Reviewed-on: https://go-review.googlesource.com/c/tools/+/247098
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Now we prefer functions when completing "go" and "defer" statements.
Previously we had no preference for the type of object. Further, we
will now also properly invoke functions.
var f1 int
var f2 func()
go f<> // prefers "f2" and expands to "f2()"
Change-Id: I213551b74ba453c337ac89e825b5d495659e9d65
Reviewed-on: https://go-review.googlesource.com/c/tools/+/246359
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Consider this example:
var foo, bar int
if foo == 123 && b<> {
}
Completing at "<>" previously preferred the unimported
"bytes.Contains()" function because it returns a bool. You often need
to compose a boolean expression from non-boolean candidates, so
preferring only bool candidates gives unhelpful results. Now we don't
infer any expected type for "&&" and "||", which allows the example to
prefer "bar" as the top candidate.
Fixesgolang/go#37163.
Change-Id: Ic341da11dd626439cfb265d129288c5ca6008270
Reviewed-on: https://go-review.googlesource.com/c/tools/+/246362
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
I noticed an annoying completion ranking issue:
// ranks "HandlerFunc" over "HandleFunc"
http.HandleFunc<>
This was due to us downranking function calls to prefer fields/vars. I
tweaked the logic to only downrank methods (with a receiver).
Change-Id: Ia4040dc8a35f641be2d7d934bf555090831219ac
Reviewed-on: https://go-review.googlesource.com/c/tools/+/246357
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
In the example:
append([]T{}, <>)
We used to track the "objType" as "[]T", and "variadicType" separately
as "T". However, most things are more interested in "T" vs "[]T", so
they had to fiddle around with swapping types. Now instead we track
objType as T, and add a "variadic=true" flag indicating that "[]T" is
also an acceptable type.
Change-Id: I8ee3ef840917378c8406368cb5c660a377498dfd
Reviewed-on: https://go-review.googlesource.com/c/tools/+/246698
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Fix a minor completion ranking issue:
foo := func(int, int) {}
foo(123, <>)
Previously we were preferring "foo()" at "<>" even though it can't be
used. We mistakenly thought we were completing the first param because
the *ast.CallExpr appears to only have a single param.
Change-Id: Iedbbb1870a4b9eb5d5be4ed266b8bb3e313b496b
Reviewed-on: https://go-review.googlesource.com/c/tools/+/246697
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
snapshot.View().Session().Cache().FileSet() has been driving me crazy
for a while. Add it to snapshot. Along the way, discover that the Cache
interface is now totally unused and delete it.
I also changed a bunch of View arguments to Snapshot while I was in the
area.
Change-Id: I1064d0020b1567c2ed28d2d55e0f4649eb94c060
Reviewed-on: https://go-review.googlesource.com/c/tools/+/245324
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
This CL addresses completion prefix not being overwritten for completion
in comments for exported variables/functions/types etc. Instead of
setting the surrounding range as cursor position, we expand out from
cursor instead to replace the word we're currently on.
Fixesgolang/go#39262
Change-Id: I90c28562e3ef285ce6848598f8d7bd7545d5c957
Reviewed-on: https://go-review.googlesource.com/c/tools/+/246237
Run-TryBot: Danish Dua <danishdua@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Just like ParseGoHandle, PackageHandle isn't very useful as part of the
public API. Remove it.
Having PackagesForFile take a URI rather than a FileHandle seems
reasonable, and made me wonder if that logic applies to other calls like
ParseGo. For now I'm going to stop here. I could also revert that part
of the change.
Change-Id: Idba8e9fdba0b0c48e841a698eb97e47fd5f23cf5
Reviewed-on: https://go-review.googlesource.com/c/tools/+/244637
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
ParseGoHandles serve two purposes: they pin cache entries so that
redundant calculations are cached, and they allow users to obtain the
actual parsed AST. The former is an implementation detail, and the
latter turns out to just be an annoyance.
Parsed Go files are obtained from two places. By far the most common is
from a type checked package. But a type checked package must by
definition have already parsed all the files it contains, so the PGH
is already computed and cannot have failed. Type checked packages can
simply return the parsed file without requiring a separate Check
operation. We do want to pin the cache entries in this case, which I've
done by holding on to the PGH in cache.pkg.
There are some cases where we directly parse a file, such as for the
FoldingRange LSP call, which doesn't need type information. Those parses
can actually fail, so we do need an error check. But we don't need the
PGH; in all cases we are immediately using and discarding it.
So it turns out we don't actually need the PGH type at all, at least not
in the public API. Instead, we can pass around a concrete struct that
has the various pieces of data directly available.
This uncovered a bug in typeCheck: it should fail if it encounters any
real errors.
Change-Id: I203bf2dd79d5d65c01392d69c2cf4f7744fde7fc
Reviewed-on: https://go-review.googlesource.com/c/tools/+/244021
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
The logic for extracting a function is quite signficant, and the code
is expensive enough that we should only call it when requested by the
user. This means that we should support extracting through a command
rather than text edits in the code action.
To that end, we create a new struct for commands. Features like extract
variable and extract function can supply functions to determine if they
are relevant to the given range, and if so, to generate their text
edits. source.Analyzers now point to Commands, rather than
SuggestedFixFuncs. The "canExtractVariable" and "canExtractFunction"
functions still need improvements, but I think that can be done in a
follow-up.
Change-Id: I9ec894c5abdbb28505a0f84ad7c76aa50977827a
Reviewed-on: https://go-review.googlesource.com/c/tools/+/244598
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
I add a code action that triggers upon request of the user. A variable
name is generated manually for the extracted code because the LSP does
not support a user's ability to provide a name.
Change-Id: Id1ec19b49562b7cfbc2cd416378bec9bd021d82f
Reviewed-on: https://go-review.googlesource.com/c/tools/+/240182
Run-TryBot: Josh Baum <joshbaum@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
CL 239754 eagerly initialized the environment. This turns out to be a
problem for gopls, which calls ApplyFixes with no ProcessEnv.
Reinitializing it every time seriously harmed the performance of
unimported completions. Move back to lazy initialization.
Working with invalid options has caused a lot of confusion; this is only
the most recent. We have to maintain backwards compatibility in the
externally visible API, but everywhere else we can require fully
populated options. That includes the source byte slice and the options.
LocalPrefix is really more of an Option than an attribute of the
ProcessEnv, and it is needed in ApplyFixes where we really don't want to
have to pass a ProcessEnv. Move it up to Options.
Change-Id: Ib9466c375a640a521721da4587091bf93bbdaa3c
Reviewed-on: https://go-review.googlesource.com/c/tools/+/241159
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
c.pkg.GetTypeInfo().ObjectOf(node.Name) will sometimes return nil.
Check for that and a few other things that c.found also checks for.
Fixesgolang/go#40043
Change-Id: I4a2b40adbbd740323e10b3460f025b29cff74130
Reviewed-on: https://go-review.googlesource.com/c/tools/+/241019
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
We use file identities pervasively throughout gopls. Prior to this
change, the identity is the modification date of an unopened file, or
the hash of an opened file. That means that opening a file changes its
identity, which causes unnecessary churn in the cache.
Unfortunately, there isn't an easy way to fix this. Changing the
cache key to something else, such as the modification time, means that
we won't unify cache entries if a change is made and then undone. The
approach here is to read files eagerly in GetFile, so that we know their
hashes immediately. That resolves the churn, but means that we do a ton
of file IO at startup.
Incidental changes:
Remove the FileSystem interface; there was only one implementation and
it added a fair amount of cruft. We have many other places that assume
os.Stat and such work.
Add direct accessors to FileHandle for URI, Kind, and Version. Most uses
of (FileHandle).Identity were for stuff that we derive solely from the
URI, and this helped me disentangle them. It is a *ton* of churn,
though. I can revert it if you want.
Change-Id: Ia2133bc527f71daf81c9d674951726a232ca5bc9
Reviewed-on: https://go-review.googlesource.com/c/tools/+/237037
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Cached packages are probably more relevant than uncached packages, but
we still need to go in relevance order, since we'll stop adding results
after we hit the cap.
Fixesgolang/go#38461. (Hopefully.)
Change-Id: I555dd5f7568baa8d69760ed5836341a474e94346
Reviewed-on: https://go-review.googlesource.com/c/tools/+/231619
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
I made a silly mistake and checked the prefix on the import path rather
than the package name, which obviously breaks everything other than
top-level stdlib packages.
Fix that, then tweak the ranking a bit. We now get deep completions, which
is nice, but filled up the results too fast. Now instead of 5 results of
any kind, we give up after 5 packages searched.
Change-Id: I15b293f68f17531077da9ffe791a38ccc0e129f4
Reviewed-on: https://go-review.googlesource.com/c/tools/+/231617
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Assigning a slice to the appendage of itself is common and tedious
enough to warrant a special case completion candidate. We now offer
smarter "append()" candidates:
var foo []int
foo = app<> // offer "append(foo, <>)"
fo<> // offer "foo = append(foo, <>)"
The latter is only offered if the best completion candidate is a
slice. It is inserted as the second-best candidate because it seems
impossible to avoid annoying false positives if it is ranked first.
I added a new debug option to disable literal completions. This was to
clean up some test logic that was disabling snippets for all tests
just to defeat literal completions. My tests were failing mysteriously
due to having snippets disabled, and it was hard to figure out why.
Change-Id: I3e8313e00a1409840cb99d5d71c593435a7aeb71
Reviewed-on: https://go-review.googlesource.com/c/tools/+/221777
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
This fixes a bunch of fmt.Errorf calls to use %w rather than %v when wrapping
an error with additional context.
Change-Id: I03088376fbf89aa537555e825e5d02544d813ed2
Reviewed-on: https://go-review.googlesource.com/c/tools/+/231477
Run-TryBot: Ian Cottrell <iancottrell@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
This should provide simple name completions for comments
above exported vars, constants, functions, and types.
Can be activated with `ctrl+space` within a comment.
Also fixes a panic introduced in the previous commit when completing comments that occur at the end of a file.
Fixes#34010Fixes#38793
Demo: https://i.imgur.com/qN82CVA.mp4
Change-Id: If9aaec7ce03a3e085361144bce5c7a66535127d1
GitHub-Last-Rev: b9ac874c7ff3c9a164ba698d0d561141a59e2435
GitHub-Pull-Request: golang/tools#224
Reviewed-on: https://go-review.googlesource.com/c/tools/+/230215
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Untyped members from unimported packages are scored the same as typed
members from unimported packages. We depended on the unimported
package relevance to rank the probably-more-relevant typed members
higher. However, there are some unrelated score penalties that can
only be applied to typed candidates, so the untyped candidates ended
up being ranked higher. Fix by increasing the relevance coefficient so
the relevance score overpowers other less important scoring
adjustments.
Fixesgolang/go#38104.
Change-Id: Ie43f769a41511f9cc3747ce6936130be7a29cd31
Reviewed-on: https://go-review.googlesource.com/c/tools/+/231238
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
In cases like:
var v interface{}
fmt.Println(<>)
Completing to "v" would insert "v..." instead of "v". This was due to
a mixup where we were checking if the variadic type "[]interface{}"
was assignable to the candidate type "interface{}" instead of the
other way around.
Fixesgolang/go#38652.
Change-Id: I27c0b50bbf4b895924c8ed2c0c9dd6785e98cbe1
Reviewed-on: https://go-review.googlesource.com/c/tools/+/229921
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
event.Log removed
event.Print -> event.Log
event.Record -> event.Metric
event.StartSpan -> event.Start
In order to support this core now exposes the MakeEvent and Export functions.
Change-Id: Ic7550d88dbf400e32c419adbb61d1546c471841e
Reviewed-on: https://go-review.googlesource.com/c/tools/+/229238
Reviewed-by: Robert Findley <rfindley@google.com>
internal/telemetry/event was renamed to internal/event/core
Some things were partly moved from internal/telemetry/event straight to
internal/event to minimize churn in the following restructuring.
Change-Id: I8511241c68d2d05f64c52dbe04748086dd325158
Reviewed-on: https://go-review.googlesource.com/c/tools/+/229237
Run-TryBot: Ian Cottrell <iancottrell@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
I originally made this change to see if it would help with the timeouts.
Based on the TryBot results, it doesn't -- but I still think it's more
correct to have the contexts this way. It was my mistake to put the
context on the completer in the first place, I think.
Change-Id: Ib77c8f0ac0b0d0922b82db4120820fb96cb664f4
Reviewed-on: https://go-review.googlesource.com/c/tools/+/227303
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
The modifications in CL 226559 introduced a new call to
RunProcessEnvFunc that was not using the context with a timeout, so deep
completion tests were hitting the timeout early (I think?).
Also, fix a staticcheck warning that's been bugging me today.
Change-Id: Iac894f630ebfa1cbc5588ae3a5490a93fa53aba2
Reviewed-on: https://go-review.googlesource.com/c/tools/+/227057
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
When proposing unimported package members, we start with packages loaded
in-memory first, for three reasons: we know that they're fast, they're
fully typed, and they're probably pretty relevant. I hadn't bothered
doing that for package names, because the first two reasons aren't very
relevant. But the third still is -- loaded packages are a pretty good
approximation for in module scope, etc.
With this change we do package names the same as members, so relevance
should be about as good. Not perfect, but nobody's complained much yet.
Fair bit of copy-and-paste, but I don't want to extend the
getAllCandidates abstraction outside of the imports package yet.
Updates #38104.
Change-Id: Ia479181607dff898baee3cd6aa84d1ab61715d19
Reviewed-on: https://go-review.googlesource.com/c/tools/+/226559
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
This change uses the new unusedparams analyzer to remove any unused parameters from functions inside of internal/lsp/source :)
Change-Id: I220100e832971b07cd80a701cd8b293fe708af3d
Reviewed-on: https://go-review.googlesource.com/c/tools/+/225997
Run-TryBot: Rohan Challa <rohan@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
A common annoying mis-completion is as follows:
type foo struct {
field int
}
func (f foo) Field() int { return f.field }
func (f foo) logic() {
if f.f<>
}
Now at <> we prefer "field" over "Field()". Similarly:
type foo struct {
}
func (f foo) DoSomething() { }
func (f foo) doSomething() { }
func (f foo) logic() {
f.d<>
}
Now at <> we prefer "doSomething()" over "DoSomething()". All else
being equally, you normally want private objects over public objects
when the private objects are available.
The same logic is applied to deep completions so we prefer "c.foo.bar"
over "c.Foo().bar".
Change-Id: Ic91cba7721ddb1f2a30338037693ddcce8c621f7
Reviewed-on: https://go-review.googlesource.com/c/tools/+/223877
Run-TryBot: Muir Manders <muir@mnd.rs>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
When searching for deep completions, we can end up enumerating struct
types' fields a lot. Optimize fieldSelections to reduce work:
- Wait until we see an embedded field before we create the "seen"
map.
- Use a callback style to iterate over the struct's fields rather than
returning a slice of fields.
- Change "seen" checking strategy back to track struct types rather
than each individual field.
Struct with 5 non-embedded fields:
name old time/op new time/op delta
Fields-16 293ns ± 1% 20ns ± 2% -93.13% (p=0.008 n=5+5)
name old alloc/op new alloc/op delta
Fields-16 120B ± 0% 0B -100.00% (p=0.008 n=5+5)
name old allocs/op new allocs/op delta
Fields-16 4.00 ± 0% 0.00 -100.00% (p=0.008 n=5+5)
Same struct but add an embedded struct with 2 fields:
name old time/op new time/op delta
Fields-16 389ns ± 1% 142ns ± 1% -63.53% (p=0.008 n=5+5)
name old alloc/op new alloc/op delta
Fields-16 120B ± 0% 144B ± 0% +20.00% (p=0.008 n=5+5)
name old allocs/op new allocs/op delta
Fields-16 4.00 ± 0% 2.00 ± 0% -50.00% (p=0.008 n=5+5)
I think the alloc/op went up because the "seen" map is no longer
allocated on the stack. There is more room for more optimization, but
it's probably not worth making things more complicated.
Change-Id: I6f9f2124334a8594ef9d6f9b5ac4b3a8aead5f49
Reviewed-on: https://go-review.googlesource.com/c/tools/+/223419
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Now we properly offer "case" and "default" keyword completion in cases
like:
switch {
<>
}
First I had to add an AST fix for empty switch statements to move the
closing brace down. For example, say the user is completing:
switch {
ca<>
}
This gets parsed as:
switch {
}
Even if we manually recover the "ca" token, "<>" is not positioned
inside the switch statement anymore, so we don't know to offer "case"
and "default" candidates. To work around this, we move the closing
brace down one line yielding:
switch {
}
Second I had to add logic to manually extract the completion prefix
inside empty switch statements, and finally some logic to offer (only)
"case" and "default" candidates in empty switch statements.
Updates golang/go#34009.
Change-Id: I624f17da1c5e73faf91fe5f69e872d86f1cf5482
Reviewed-on: https://go-review.googlesource.com/c/tools/+/220579
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Now we offer an error-check-and-return completion candidate when
appropriate:
func myFunc() (int, error) {
f, err := os.Open("foo")
<>
}
offers the candidate:
if err != nil {
return 0, <err>
}
where <> denotes a placeholder so you can easily alter "err".
The completion will only be offered when:
1. The position is in a function that returns an error as final result
value, and
2. The statement preceding position is an assignment whose final LHS
value is an error.
The completion will contain zero values for the non-error return values
as necessary.
Using the above example, the completion will be offered after the user
has typed:
i
if
if err
Basically the candidate will be offered after every keystroke as the
user types "if err".
I call this new type of completion a statement completion - perfect
for when you want to make a statement!
Change-Id: I0a330e1c1fa81a2757d3afc84c24e853f46f26b0
Reviewed-on: https://go-review.googlesource.com/c/tools/+/221613
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Editors typically trigger completion automatically after ".". This
pops up annoying, useless completions after "..." variadic param, such
as "foo(bar...<>)". We now suppress completions in or directly after
the ellipsis.
Fixesgolang/go#37358.
Change-Id: I9fc94fbdf69429bd787bcb2c643f4f730e24154d
Reviewed-on: https://go-review.googlesource.com/c/tools/+/222200
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Offer "struct", "interface", "map", "chan", and "func" keywords when
we expect a type. For example "var foo i<>" will offer "interface".
Because "struct" and "interface" are more often used when declaring
named types, they get a higher score in type declarations. Otherwise,
"map", "chan" and "func" get a higher score.
I also got rid of the special keyword scoring. Now keywords just use
stdScore and highScore. This makes the interplay with other types of
candidates more predictable. Keywords are offered in pretty limited
contexts, so I don't think they will be annoying.
Finally, keyword candidate score is now to be scaled properly based on
how well they match the prefix. Previously they weren't penalized for
not matching well, so there were probably some situations where
keywords were ranked too high.
Updates golang/go#34009.
Change-Id: I0b659c00a8503cd72da28853dfe54fcb67f734ae
Reviewed-on: https://go-review.googlesource.com/c/tools/+/220503
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
We were panicking when a LHS assignee's type was nil, such as:
// "foo" has not been declared
foo = 123
A recent refactoring changed (*completer).typeMatches() to no longer
gracefully handle a nil types.Type for its first parameter. That
behavior seems fine, so fix the problematic caller of typeMatches to
check for nil before calling.
Change-Id: Ie11e4a2d374ab1efbf6fd13fbe214e06d359fca0
Reviewed-on: https://go-review.googlesource.com/c/tools/+/221020
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
The code to check if a candidate object matches our candidate
inference had become complicated, messy, and in some cases incorrect.
The main source of the complexity is the "derived" expected and
candidate types. When considering a candidate object "foo", we also
consider "&foo", "foo()", and "*foo", as appropriate. On the expected
side of things, when completing the a variadic function parameter we
expect either the variadic slice type and the scalar element type.
The code had grown organically to handle the expanding concerns, but
that resulted in confused code that didn't handle the interplay
between the various facets of candidate inference.
For example, we were inappropriately invoking func candidates when
completing variadic args:
func foo(...func())
func bar() {}
foo(bar<>) // oops - expanded to "bar()"
and we weren't type matching functions properly as builtin args:
func myMap() map[string]int { ... }
delete(myM<>) // we weren't preferring (or invoking) "myMap()"
We also had methods like "typeMatches" which took both a "candidate"
object and a "candType" type, which doesn't make sense because the
candidate contains the type already.
Now instead we explicitly iterate over all the derived candidate and
expected types so they are treated the same. There are still some
warts left but I think this is a step in the right direction.
Change-Id: If84a84b34a8fb771a32231f7ab64ca192f611b3d
Reviewed-on: https://go-review.googlesource.com/c/tools/+/218877
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
Add support for var/func/const/type/import keywords at the file scope.
I left out "package" because, currently, if you are completing
something that means you must already have a package declaration. The
main hurdle was that anything other than a decl keyword shows up as
an *ast.BadDecl at the file scope. To properly detect the prefix we
manually scan for the token containing the position.
I also made a couple small drive-by improvements:
- Also suggest "goto" and "type" keywords in functions.
- Allow completing directly before a comment, e.g. "foo<>//". I
needed this for a test that would have been annoying to write
otherwise.
Updates golang/go#34009.
Change-Id: I290e7bdda9e66a16f996cdc291985a54bf375231
Reviewed-on: https://go-review.googlesource.com/c/tools/+/217500
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
Currently there is no need for this because the file contents are part
of the file handle. This change is in preparation for an impending
improvement that tweaks the source code during the parse stage to fix
certain kind of terminal parse errors. Any code that wants to use
an *ast.File or *token.File in conjunction with the file contents
needs access to the doctored source code so things line up.
Change-Id: I59d83d3d6150aa1264761aa2c1f6c1269075a2ce
Reviewed-on: https://go-review.googlesource.com/c/tools/+/218979
Run-TryBot: Muir Manders <muir@mnd.rs>
Run-TryBot: Rebecca Stambler <rstambler@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
