hairyfotr / linter Goto Github PK

Consider

    @tailrec
    private def f(x: Widget, token: Option[String] = None): Boolean = {
      if (foo) true
      else Option(x.getNextToken) match { 
        case None => false
        case Some(tok) => f(x, Some(tok))
      }
    }

When a method is marked @tailrec, the linter should not trigger, since rewriting creates a non-tail-recursive function

 Option(x.getNextToken).exists(t => f(x, Some(t))

This is "morally" tail recursive, but the type checker can't tell. (I didn't look at the bytecode to see if it gets compiled to a tail recursive function.)

the plugin impacts compilation time quite severely
for me it goes up from about 300 secs to about 600
even enabling just one check, eg enable-only:UnlikelyEquality still gives me about 600 secs

The analyser incorrectly thinks that assignments in anonymous inner classes happen directly. Adding an assert(called == 0) between the two adds here works around the problem.

  "remove" should "be idempotent" in {
    val event = new Event
    var called = 0

    val id1 = event.add(new Runnable {
      override def run(): Unit = called = 1
    })
    val id2 = event.add(new Runnable {
      override def run(): Unit = called = 2 // Variable called has an unused value before this reassign.
    })

    event.remove(id1)
    event.remove(id1)

    event.run()
    assert(called == 2)
  }

Use of the parboiled2 library which uses scala macros causes a MergeNestedIfs warnings at every rule declaration

Parser.scala:6: warning: [MergeNestedIfs] These two nested ifs can be merged into one.
def Input = rule {
                 ^

This should cause a warning:

def test : Option[Type] = null

The user probably meant to return None.

Hi! Thanks for your work on the Linter library!

This method does not raise an unused method parameter by Linter:

def foo(param1: String, param2: Int) = {
    param2 + 1
}

version: linter_2.11-0.1-SNAPSHOT

[warn] /home/johannes/git/telfish/spray/spray/spray-can/src/main/scala/spray/can/server/PipeliningLimiter.scala:33: Parameter context is not used in method apply. (Add override if that's the reason)
[warn]       def apply(context: PipelineContext, commandPL: CPL, eventPL: EPL): Pipelines =
[warn]           ^

You could use typing information instead of relying on the modifiers to find out if a method is an implementation of an abstract method.

On the other hand, it's arguable if you should always use override in these cases but there are a lot of cases where adding override just adds noise (like SAMs). Maybe make it configurable if you want all method implementations to use override and other wise don't warn when you can detect that a method is an implementation.

val base = scala.collection.mutable.Map[String, Any](
  "review_id" -> "abcd"
)

linter reports:
Inferred type scala.collection.mutable.Map[String,Any]. (This might not be what you've intended)

I did certainly intend to have a Map[String, Any].

Would it be possible to publish an official release? Often I'm hacking without a network connection and have to comment out adding the linter compiler plugin because sbt tries to update the snapshot release ...

When a project has converted some data structures to e.g. shapeless records, compilation time has naturally increased. However, using linter, it spends a disproportionate amount of time in the linter-typed and linter-typed-inter. I've done some measurements using "-verbose" and the scalac-aspects project (https://github.com/gkossakowski/scalac-aspects) and for a particular file:

• Without sqltyped, without linter:
  total: 917197, typer: 360658
• without sqltyped, with linter:
  total: 1766577, typer: 513275, linter-typed: 392068, linter-typed-interpreter: 221284
• with sqltyped, without linter:
  total: 5001996, typer: 3124241
• with sqltyped, with linter:
  total: 32682659, typer: 3281671, linter-typed: 15140647, linter-typed-interpreter: 12101858

So while linter phases took about 35% of the time without sqltyped, they took a whopping 83% with sqltyped. So while the time in typer increased 8-9 times, linter phases increased 43 times!

I guess this is because shapeless generates a very deep type tree, but I didn't expect the linter time to grow so drastically. Do you have an idea how this could be improved? Perhaps a way to indicate to the traverser to prune a type branch and not process it at all? For instance, if there is a shapeless HList, this might indicate a very deep type structure and could be skipped, as there isn't much value in statically checking it. Maybe if shapeless indicates in some way that some of the members are synthetically generated?

Any tips are welcome.

is there some mechanism to integrate the finding result with sonar/jenkins or any other reporting mechenism?

Moreover, can it be configured to stop the build for any or some warning/error?
Thanks

-Chris L.

Hi, I'm getting a warning

Unless there are side-effects, this col.foldLeft can be replaced by col.exists.

which in principle would be correct, except there is no exists on my collection (scalaz.NonEmptyList).

Don't run the compiler phases that don't need to run (anything after refchecks, etc), check out Global.Run instead of IMain interpreter, where you can possibly also get positions and strings, allowing mixed tests. Non-compiling negative tests pass right now!

For example, the below cause the error "Inferred type Map[String,Any]. (This might not be what you've intended)". Since the method is defined as returning Map[String, Any] I would think this is acceptable code.

val exifMap = parseExif()
def parseExif(): Map[String, Any] = {}

I am using SBT 0.13.11, Scala 2.11.8 and 0.1.14 version of the compiler plugin, and the ContainsTypeMismatch check does not catch any of the following:
Some("foo").contains(1)
Some(1).contains("foo")
List(4).contains("foo")

I haven't found a case where it is working properly. The UnlikelyEquality check is working fine, so I don't believe it's an issue with my build.

I am kinda new to Scala and this might be be an SBT thing, but I get the linter warnings only the first time I compile the code. Every subsequent compile goes thorough without any warnings. This might be because the compiled class files are not compiled again unless there are any changes, but is there a way to show the warnings every time nonetheless?

Say we have a macro for debugging:

@inline def debug(msg: => String): Unit = macro debugImpl

with debugImpl being equivalent to:

if (Logger.isDebugEnabled) Logger.debug(msg)

Then errors are reported on the call sites of this method:

if (someCondition) debug(msg) <<<

val x = 1/0 // linter:ignore DivideByZeor

should issue a warning about unrecognized check (notice the typo).

It seems Sonatype has some problems... Use 0.1.14 or 0.1-SNAPSHOT(currently same as 0.1.15) in the meantime.

When you have a range of values, this is the preffered style, and Linter shouldn't warn:

if (0 < x && x < 10) ...

I get this error when I try to generate scaladocs: [warn] dropping dependency on node with no phase object: refchecks

sbt doc

val x = 1/0 // linter:ignore DivideByZero,YodaConditions

should cause a warning about YodaConditions not present on the line.

I've found a false positive with the ContainsTypeMismatch:

val foo = Some("foo")
val bar = Some("bar")

fails with the following:

foo.exists(bar.contains(_))
foo.exists(bar.contains)

... but works with:

foo.exists(str => bar.contains(str))

Tasks:

• Change uses of "scala.this.X" to "scala.X"
• Refactor tests to a newer test library
• Fix false positives from abstract interpreter
• Fix failing tests (two left)

The UnnecessaryElseBranch warning is issued when

"the then branch always returns"

What is the reasoning behind issuing this warning? When the if condition fails, the else branch is executed, so it doesn't seem unnecessary to me.

This is one of the unit tests for UnnecessaryElseBranch:

should("""
  def test(): Any = {
    if(util.Random.nextBoolean) {
      println("foo"); return 5; println("foo2");
    } else {
      println("foo3"); println("foo4");
    }
  }""")

I don't see why the else branch would be unnecessary here.

Switch for all linter checks to be errors, and switch for info,warning,error sets of checks

There are two legitimately nested if so not sure what this is about:

[warn] /home/johannes/git/telfish/spray/spray/spray-can/src/main/scala/spray/can/server/StatsSupport.scala:45: These two ifs can be merged into one.
[warn]       if (currentMoc > moc)
[warn]       ^

Here's the complete method:

@tailrec
final def adjustMaxOpenConnections(): Unit = {
  val co = connectionsOpened.get
  val cc = connectionsClosed.get
  val moc = maxOpenConnections.get
  val currentMoc = co - cc
  if (currentMoc > moc)
    if (!maxOpenConnections.compareAndSet(moc, currentMoc)) adjustMaxOpenConnections()
}

src/scala/com/stripe/streamish/Streamish.scala:33: warning: Parameter $this is not used in method copy$extension.
case class InvariantEvents[T](evs: Events[T]) extends AnyVal {
           ^
src/scala/com/stripe/streamish/Streamish.scala:33: warning: Parameter $this is not used in method productPrefix$extension.
case class InvariantEvents[T](evs: Events[T]) extends AnyVal {
           ^
src/scala/com/stripe/streamish/Streamish.scala:33: warning: Parameter $this is not used in method productArity$extension.
case class InvariantEvents[T](evs: Events[T]) extends AnyVal {
           ^
src/scala/com/stripe/streamish/Streamish.scala:33: warning: Parameter $this is not used in method canEqual$extension.
case class InvariantEvents[T](evs: Events[T]) extends AnyVal {
           ^
warning: there were three feature warnings; re-run with -feature for details
5 warnings found

those methods are all compiler generated when using AnyVal. -P:linter:disable:UnusedParameter fixes it, but that is a pretty heavy hammer to drop on default settings.

It would be very useful to prospective users to have a section in the README discussing the other Scala linters that are available and explaining how this project relates to those and when one might choose each.

This SO answer has links to the current Scala linters: http://stackoverflow.com/a/25117125/8261

I see that you have a list of links to other Scala linters in a section starting "Rule lists from other static analysis tools", but I can't tell if this project is more or less appropriate for my needs than the alternatives.

Thanks

[warn] /home/johannes/git/telfish/spray/spray/spray-io/src/main/scala/akka/io/DirectByteBufferPool.scala:43: This condition will never hold.
[warn]         try if (buffersInPool > 0) 

buffersInPool is a mutable field of the class this condition lies in.

huitseeker➜huitseeker/tmp/TEst» sbt lint:compile                      [18:43:17]
[warn] /home/huitseeker/tmp/TEst/src/main/scala/Test.scala:2: Taking the modulo by one will return zero.
[warn]   def f(x: Double) = x % 1
[warn]                        ^
[warn] one warning found
[success] Total time: 3 s, completed Nov 6, 2015 6:43:25 PM
huitseeker➜huitseeker/tmp/TEst» cat src/main/scala/Test.scala         [18:43:25]
object Test {
  def f(x: Double) = x % 1
}

Taking the modulo by 1 on a float or double is actually a perfectly valid way to take the fractional part.ty o

Note : I assume Division by 1 is guilty of the same problem, but didn't have time to check.

Often I'll see code that does _.size == 0 or _.length == 0 when _.isEmpty would be better. It's good to get people into this habit since for some collections (notably List[_]) getting the size is an O(n) operation whereas checking emptiness is O(1).

looks like linter do not find conditions like

if (false) "not linted"
if (true) { "not linted" } else { "not linted" }

Here should be a warning, that conditional expression should be omitted.

My stack is:
Scala 2.10.2, sbt 0.13.0, Play 2.2.1
Linter is added as jar and scalacOpt to Build.scala file.

I have got the following go function

import (
	"fmt"
	"github.com/northwesternmutual/grammes"
	"github.com/stretchr/testify/assert"
	"testing"
	"time"
)

func connect(t *testing.T) *grammes.Client {
	client, err := grammes.DialWithWebSocket("ws://localhost:8182", grammes.WithAuthUserPass("root", "root"), grammes.WithTimeout(time.Second))
	assert.NoError(t, err)
	return client
}

and I get the following linter error for the first function line with Google Style client, err := grammes.DialWithWebSocket("ws://localhost:8182", grammes.WithAuthUserPass("root", "root"), grammes.WithTimeout(time.Second)):

 13:73   error  '.W' should have one space.                    google.Spacing 
 13:91   error  Commas and periods go inside quotation marks.  google.Quotes  
 13:115  error  '.W' should have one space.                    google.Spacing 
 13:132  error  '.S' should have one space.                    google.Spacing

I assume that code lines are skipped.

I downloaded the .jar for Scala 2.11 from the link provided in the README.md and running it with java -jar linter_2.11-0.1-SNAPSHOT.jar gives me the error no main manifest attribute, in linter_2.11-0.1-SNAPSHOT.jar.

Linting the following code:

package pkg

sealed trait Base

case object Sub extends pkg.Base

would incorrectly yield:

[warn] /tmp/linter/src/main/scala/Test.scala:3: [UnextendedSealedTrait] This sealed trait is never extended
[warn] sealed trait Base
[warn]              ^
[warn] one warning found

Changing extends pkg.Base to extends Base would fix the lint error. Using the full name is sometimes necessary to prevent ambiguity.

It's not enough to use a string constant with BigDecimal and a large decimal expansion, you have to actually provide your own MathContext:

scala> BigDecimal("0.12958129581958395829583295823958329852395832958295382")
res0: scala.math.BigDecimal = 0.1295812958195839582958329582395833

scala> BigDecimal("0.12958129581958395829583295823958329852395832958295382", new java.math.MathContext(200))
res1: scala.math.BigDecimal = 0.12958129581958395829583295823958329852395832958295382

I'm not sure the best way to word the warning, but I think the current warning is misleading.

To demonstrate the problem:

scala> class A; class B extends A
defined class A
defined class B

scala> val b = new B; val a: A = b
b: B = B@3358f363
a: A = B@3358f363

scala> List(a).contains(b)
res9: Boolean = true

scala> List(b).contains(a)
:16: warning: [ContainsTypeMismatch] List[B].contains(A) will probably return false, since the collection and target element are of unrelated types.
List(b).contains(a)
^
res10: Boolean = true

The warning shouldn't be raised because it's possible for a List of B to contain an element of type A which is a supertype (not subtype) of B- it's obvious that it does return true.

I tried your project in our play app (2.1.1, scala 2.10) and am getting a StackOverflowError when compiling, specifically on the generated reverse routes.

The relevant output:

[error]
[error] while compiling: something/routes_reverseRouting.scala
[error] during phase: linter-typed
[error] library version: version 2.10.2
[error] compiler version: version 2.10.2
[error] reconstructed args: -encoding utf8 -deprecation -feature -bootclasspath something -Xplugin:lib/linter.jar
[error]
[error] last tree to typer: Select(This(package txt), foo)
[error] symbol: object foo in package txt (flags: )
[error] symbol definition: object foo
[error] tpe: views.txt.foo.type
[error] symbol owners: object foo -> package txt
[error] context owners: method readResolve -> object foo -> package txt
[error]
[error] == Enclosing template or block ==
[error]
[error] DefDef( // private def readResolve(): Object in object foo
[error] private
[error] "readResolve"
[error] []
[error] List(Nil)
[error] // tree.tpe=Object
[error] txt.this."foo" // object foo in package txt, tree.tpe=views.txt.foo.type
[error] )
[error]
[error] == Expanded type of tree ==
[error]
[error] TypeRef(
[error] TypeSymbol(
[error] class foo extends BaseScalaTemplate[play.api.templates.Txt,play.templates.Format[play.api.templates.Txt]] with Template4[Option[String],Option[String],Boolean,util.EnhancedRequest[_],play.api.templates.Txt]
[error]
[error] )
[error] )
[error]
[error] uncaught exception during compilation: java.lang.StackOverflowError
error java.lang.StackOverflowError

First of all, thanks for maintaining linter! I've ran it over the current spray code base (jrudolph/spray@bc7bbb7) to see what it catches. I think I've found several issues and I'm putting them into here one by one.

[warn] /home/johannes/git/telfish/spray/spray/spray-io/src/main/scala/akka/io/Tcp.scala:91: This sealed trait is never extended.
[warn]   sealed trait Message

but it definitely is

  sealed trait Message

  /// COMMANDS

  /**
   * This is the common trait for all commands understood by TCP actors.
   */
  trait Command extends Message with SelectionHandler.HasFailureMessage {
    def failureMessage = CommandFailed(this)
  }

Not sure what the exact condition is.

Is there a version that works with Scala 2.13 ?

Linter gives a false positive for "method parameter not used" if there's a java file in project.

full file http://pastebin.com/uRZcR0e5

it says this method doesn't use it's parameters

public void setAcceptedAttributeNameRegex(String acceptedAttributeNameRegex) {
    this.acceptedAttributeNameRegex = Pattern.compile(acceptedAttributeNameRegex);
}

and also this one

private boolean isAcceptedAttribute(Object key) {
    if (acceptedAttributeNameRegex == null) {
        return true;
    }
    if (!(key instanceof String)) {
        return true;
    }
    return acceptedAttributeNameRegex.matcher((String)key).matches();
}

since this is clearly false...is there a way to check if file being compiled is a java source and then just skip it(no linting)?

With version 0.1.17, I get a false positive with the following code:

def test1(s: StringBuilder): Boolean = {
    if (s != null) {
      val trimmed = s.toString.trim
      trimmed.nonEmpty
    } else {
      false
    }
  }

Warning:(123, 15) [UnnecessaryStringNonEmpty] This string will never be empty.
      trimmed.nonEmpty

If I remove the intermediate variable trimmed, the false positive disappears:

  def test2(s: StringBuilder): Boolean = {
    if (s != null) {
      s.toString.trim.nonEmpty
    } else {
      false
    }
  }

It is also not reported if we replace StringBuilder by String as the argument:

  def test3(s: String): Boolean = {
    if (s != null) {
      val trimmed = s.trim
      trimmed.nonEmpty
    } else {
      false
    }
  }

Thanks for your Linter!
Is it possible to have it published on a more default repository such as Maven Central? This way it will be possible to get it into our corporate Nexus.

Can't write this code using foreach since it is using tailrec. Can't think of another way to write this without using null.

  @tailrec
  final def tick(name: String, time: FiniteDuration): Unit =
    Option(times.putIfAbsent(name, time)) match { // linter:ignore UseOptionForeachNotPatMatch
      case None =>
      case Some(oldValue) =>
        if (!times.replace(name, oldValue, oldValue + time)) {
          tick(name, time)
        }
    }

sealed trait HttpStatusCode {
  val intValue: Int
}

object HttpStatusCode {
  private type T = HttpStatusCode
  case object BadRequest extends T { val intValue = 400 }
  case object Unauthorized extends T { val intValue = 403 }
  case object InternalServerError extends T { val intValue = 500 }
}

gives

warning: [UnextendedSealedTrait] This sealed trait is never extended
sealed trait HttpStatusCode {

Imo that's almost always a bug. Also M[Any] and M[Nothing]

Thanks for the great work, btw!

Reported by @harshad-deo on gitter:
Unused method parameter check does not consider usage of types defined in the parameter

specifically, this throws a warning:

def buildT = new Witness[EmptyDenseSet#Include[f.ValueHash]]
def build[T](f: LiteralHash[T]) = new Witness[EmptyDenseSet#Include[f.ValueHash]]