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>
We were crashing in cases like:
var _ []byte = append([]byte{}, ""...<>)
We were type asserting the type of append's second param
to *types.Slice, but in this case it is a string (*types.Basic). Fix
by checking the type assert was successful.
Note that we still don't attempt to give string completions when
appending to a byte slice. We can add that special case later once
everyone is clamoring for it.
Change-Id: I1d2fbd7f538e580d33c2dab4ef127a88e16d7ced
Reviewed-on: https://go-review.googlesource.com/c/tools/+/219144
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Now when we expect a type name at the cursor, we omit non-type name
completion candidates. For example:
inch := 1
var foo in<> // don't offer "inch"
I also added expected type name detection for value specs:
// Expect a type name at <>
var foo <>
Fixesgolang/go#32806.
Change-Id: I32477cb286d2050bac5ccc767f8a608124fa5acd
Reviewed-on: https://go-review.googlesource.com/c/tools/+/216400
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
Now we keep a count of how many times to dereference a candidate. For
example:
var foo ***int
var _ int = f<> // Now we offer "***foo" instead of "*foo".
Change-Id: I14edc40aeec6884399eceb3dd3b4f85dc74a773c
Reviewed-on: https://go-review.googlesource.com/c/tools/+/218580
Run-TryBot: Muir Manders <muir@mnd.rs>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
When completing a composite literal value, we were returning from
candidate inference before we recorded type modifiers such as prefix
"&" or "*". This was causing funny completions like:
type myStruct struct { s *myStruct }
myStruct{s: &mySt<> // completed to "&&myStruct{}"
Now we properly pick up on the "&" prefix so we know our literal
"myStruct{}" candidate does not need a "&".
Change-Id: I908936698cfedfef81bc0c1cbcd93e14dc00e3a3
Reviewed-on: https://go-review.googlesource.com/c/tools/+/218377
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Currently we show up to ~100 unimported package names matching the
completion prefix. This isn't useful since, assuming the user even
wants an unimported package, they will just type more to narrow it
down rather than scroll through 100 options. Having so many candidates
also slows things down due to per-candidate overhead in gopls and in
the LSP client. Now we instead limit to 5 unimported package names.
Unimported package members, on the other hand, make sense to list
many. The user may want to scroll through because they don't remember
the name of what they are looking for. I left the max value at 100.
Change-Id: I00e11fa0420758f8db6c7049f80fa156773a5ee6
Reviewed-on: https://go-review.googlesource.com/c/tools/+/218879
Run-TryBot: Muir Manders <muir@mnd.rs>
Reviewed-by: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
When completing members on a type checked, unimported package, you get
fully typed members. That means you get deep completions. Because we
downrank the initial unimported package members so much, any deep
completions were dominating the rankings. For example
context.Back<>
yielded "context.Background().Err" ranked above "context.Background".
Fix by scoring context.Background in this example as
stdScore+tinyRelevanceScore instead of just tinyRelevanceScore. I also
changed untyped candidate scores in the same way so they stay
competitive when you have both imported and unimported candidates.
The other option was to propagate the score penalty into deep
candidates, but that wasn't easy. In general I think you are better off
avoiding big score penalties because they complicate the interplay
between different kinds of candidates. Scoring needs an overhaul, but
at least we are building up our test suite in the meantime.
Change-Id: Ia5d32c057b04174229686cec6ac0542c30e186e2
Reviewed-on: https://go-review.googlesource.com/c/tools/+/218378
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
We were recursing infinitely evaluating objects of recursive pointer
types such as "type foo *foo". Now we track named pointer types we
have already seen to avoid trying to dereference such objects forever.
I lazily initialized the "seen" map to avoid the allocation in the
normal case when you aren't dealing with named pointer types.
Fixesgolang/go#37104.
Change-Id: I5f294cfc5a641e7b5fd24e1d9dc55520726ea560
Reviewed-on: https://go-review.googlesource.com/c/tools/+/218579
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Completion often fails when the completion prefix happens to be a
keyword. We previously tried to fix this with AST surgery, but
often the accidental keyword is not apparent looking at the AST.
For example:
chan<>
foo()
parses as CallExpr{Fun: ChanType{Value: Ident{"foo"}}} with very few
hints that something is wrong, and:
default
foo()
is completely omitted from the AST.
Rather than look in the AST, we now instead manually look for a
keyword token that contains the completion position. If we find one,
we treat that as our surrounding identifier.
Updates golang/go#34332.
Change-Id: I68ed0dd905848c0eae61f39ecb8b73adb1e72746
Reviewed-on: https://go-review.googlesource.com/c/tools/+/216961
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
For example:
// Prefer functions that return one or two values. Previously
// we had no preference.
foo, bar := <>
// Prefer functions that return "(int)" or "(int, ??)". Previously we
// only preferred the former.
var foo int
foo, bar := <>
// Prefer functions that return "(int)" or "(int, int)". Previously we
// only preferred the former.
var foo func(int, int)
foo(<>)
In the above example, we don't handle "foo" being variadic yet.
I also took the liberty to break up matchingCandidate() into separate
functions since it was getting rather long.
Updates golang/go#36540.
Change-Id: I9140dd989dfde1ddcfcd9d2a14198045c02587f2
Reviewed-on: https://go-review.googlesource.com/c/tools/+/215537
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
Fix type inference to expect a type name for the first "make()"
parameter and an integer for later parameters. For example:
// Previously we expected "[]int{}", now we expect "[]int".
var _ []int = make(<>)
Note that we don't currently support actually completing to unnamed
type names like "[]int", but this improvement at least eliminates
nonsensical completion suggestions.
// Previously we had no expectation, now we expect an int.
var _ []int = make([]int, <>)
Change-Id: Ifd349767662ab6902d3a3ea9e52de7df70cb37c7
Reviewed-on: https://go-review.googlesource.com/c/tools/+/217310
Run-TryBot: Muir Manders <muir@mnd.rs>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ingo Oeser <nightlyone@googlemail.com>
Reviewed-by: Rebecca Stambler <rstambler@golang.org>
