433 lines
10 KiB
Nim
433 lines
10 KiB
Nim
{.used.}
|
|
|
|
# nim-result is also available stand-alone from https://github.com/arnetheduck/nim-result/
|
|
|
|
import ../stew/results
|
|
|
|
type R = Result[int, string]
|
|
|
|
# Basic usage, producer
|
|
|
|
block:
|
|
func works(): R = R.ok(42)
|
|
func works2(): R = result.ok(42)
|
|
func works3(): R = ok(42)
|
|
|
|
func fails(): R = R.err("dummy")
|
|
func fails2(): R = result.err("dummy")
|
|
func fails3(): R = err("dummy")
|
|
|
|
let
|
|
rOk = works()
|
|
rOk2 = works2()
|
|
rOk3 = works3()
|
|
|
|
rErr = fails()
|
|
rErr2 = fails2()
|
|
rErr3 = fails3()
|
|
|
|
doAssert rOk.isOk
|
|
doAssert rOk2.isOk
|
|
doAssert rOk3.isOk
|
|
doAssert (not rOk.isErr)
|
|
|
|
doAssert rErr.isErr
|
|
doAssert rErr2.isErr
|
|
doAssert rErr3.isErr
|
|
|
|
# Mutate
|
|
var x = rOk
|
|
x.err("failed now")
|
|
doAssert x.isErr
|
|
doAssert x.error == "failed now"
|
|
|
|
# Combine
|
|
doAssert (rOk and rErr).isErr
|
|
doAssert (rErr and rOk).isErr
|
|
doAssert (rOk or rErr).isOk
|
|
doAssert (rErr or rOk).isOk
|
|
|
|
# Fail fast
|
|
proc failFast(): int = raiseAssert "shouldn't evaluate"
|
|
proc failFastR(): R = raiseAssert "shouldn't evaluate"
|
|
|
|
doAssert (rErr and failFastR()).isErr
|
|
doAssert (rOk or failFastR()).isOk
|
|
|
|
# `and` heterogenous types
|
|
doAssert (rOk and Result[string, string].ok($rOk.get())).get() == $(rOk[])
|
|
|
|
# `or` heterogenous types
|
|
doAssert (rErr or Result[int, int].err(len(rErr.error))).error == len(rErr.error)
|
|
|
|
# Exception on access
|
|
doAssert (try: (discard rOk.tryError(); false) except ResultError[int]: true)
|
|
doAssert (try: (discard rErr.tryGet(); false) except ResultError[string]: true)
|
|
|
|
# Value access or default
|
|
doAssert rOk.get(100) == rOk.get()
|
|
doAssert rErr.get(100) == 100
|
|
|
|
doAssert rOk.get() == rOk.unsafeGet()
|
|
|
|
rOk.isOkOr: raiseAssert "should not end up in here"
|
|
rErr.isErrOr: raiseAssert "should not end up in here"
|
|
|
|
rErr.isOkOr:
|
|
doAssert error == rErr.error()
|
|
|
|
rOk.isErrOr:
|
|
doAssert value == rOk.value()
|
|
|
|
doAssert rOk.valueOr(failFast()) == rOk.value()
|
|
let rErrV = rErr.valueOr:
|
|
error.len
|
|
doAssert rErrV == rErr.error.len()
|
|
|
|
let rOkV = rOk.errorOr:
|
|
$value
|
|
doAssert rOkV == $rOk.get()
|
|
|
|
# Exceptions -> results
|
|
func raises(): int =
|
|
raise (ref CatchableError)(msg: "hello")
|
|
|
|
let c = catch:
|
|
raises()
|
|
doAssert c.isErr
|
|
|
|
# De-reference
|
|
when (NimMajor, NimMinor) >= (1, 6):
|
|
{.warning[BareExcept]:off.}
|
|
|
|
try:
|
|
echo rErr[]
|
|
doAssert false
|
|
except:
|
|
discard
|
|
|
|
when (NimMajor, NimMinor) >= (1, 6):
|
|
{.warning[BareExcept]:on.}
|
|
|
|
# Comparisons
|
|
doAssert (rOk == rOk)
|
|
doAssert (rErr == rErr)
|
|
doAssert (rOk != rErr)
|
|
|
|
# Mapping
|
|
doAssert (rOk.map(func(x: int): string = $x)[] == $rOk.value)
|
|
doAssert (rOk.map(func(x: int) = discard)).isOk()
|
|
|
|
doAssert (rOk.flatMap(
|
|
proc(x: int): Result[string, string] = Result[string, string].ok($x))[] == $rOk.value)
|
|
|
|
doAssert (rErr.mapErr(func(x: string): string = x & "no!").error == (rErr.error & "no!"))
|
|
|
|
# Casts and conversions
|
|
doAssert rOk.mapConvert(int64)[] == int64(42)
|
|
doAssert rOk.mapConvert(uint64)[] == uint64(42)
|
|
doAssert rOk.mapCast(int8)[] == int8(42)
|
|
|
|
doAssert (rErr.orErr(32)).error == 32
|
|
doAssert (rOk.orErr(failFast())).get() == rOk.get()
|
|
|
|
# string conversion
|
|
doAssert $rOk == "ok(42)"
|
|
doAssert $rErr == "err(dummy)"
|
|
|
|
# Exception interop
|
|
let e = capture(int, (ref ValueError)(msg: "test"))
|
|
doAssert e.isErr
|
|
doAssert e.error.msg == "test"
|
|
|
|
try:
|
|
discard rOk.tryError()
|
|
doAssert false, "should have raised"
|
|
except ValueError:
|
|
discard
|
|
|
|
try:
|
|
discard e.tryGet()
|
|
doAssert false, "should have raised"
|
|
except ValueError as e:
|
|
doAssert e.msg == "test"
|
|
|
|
# Nice way to checks
|
|
if (let v = works(); v.isOk):
|
|
doAssert v[] == v.value
|
|
|
|
# Expectations
|
|
doAssert rOk.expect("testOk never fails") == 42
|
|
|
|
# Question mark operator
|
|
func testQn(): Result[int, string] =
|
|
let x = ?works() - ?works()
|
|
ok(x)
|
|
|
|
func testQn2(): Result[int, string] =
|
|
# looks like we can even use it creatively like this
|
|
if ?fails() == 42: raise (ref ValueError)(msg: "shouldn't happen")
|
|
|
|
func testQn3(): Result[bool, string] =
|
|
# different T but same E
|
|
let x = ?works() - ?works()
|
|
ok(x == 0)
|
|
|
|
doAssert testQn()[] == 0
|
|
doAssert testQn2().isErr
|
|
doAssert testQn3()[]
|
|
|
|
proc heterOr(): Result[int, int] =
|
|
let value = ? (rErr or err(42)) # TODO ? binds more tightly than `or` - can that be fixed?
|
|
doAssert value + 1 == value, "won't reach, ? will shortcut execution"
|
|
ok(value)
|
|
|
|
doAssert heterOr().error() == 42
|
|
|
|
# Flatten
|
|
doAssert Result[R, string].ok(rOk).flatten() == rOk
|
|
doAssert Result[R, string].ok(rErr).flatten() == rErr
|
|
|
|
# Filter
|
|
doAssert rOk.filter(proc(x: int): auto = Result[void, string].ok()) == rOk
|
|
doAssert rOk.filter(proc(x: int): auto = Result[void, string].err("filter")).error == "filter"
|
|
doAssert rErr.filter(proc(x: int): auto = Result[void, string].err("filter")) == rErr
|
|
|
|
# Exception conversions - toException must not be inside a block
|
|
type
|
|
AnEnum = enum
|
|
anEnumA
|
|
anEnumB
|
|
AnException = ref object of CatchableError
|
|
v: AnEnum
|
|
|
|
func toException(v: AnEnum): AnException = AnException(v: v)
|
|
|
|
func testToException(): int =
|
|
try:
|
|
var r = Result[int, AnEnum].err(anEnumA)
|
|
r.tryGet
|
|
except AnException:
|
|
42
|
|
|
|
doAssert testToException() == 42
|
|
|
|
type
|
|
AnEnum2 = enum
|
|
anEnum2A
|
|
anEnum2B
|
|
|
|
func testToString(): int =
|
|
try:
|
|
var r = Result[int, AnEnum2].err(anEnum2A)
|
|
r.tryGet
|
|
except ResultError[AnEnum2]:
|
|
42
|
|
|
|
doAssert testToString() == 42
|
|
|
|
block: # Result[void, E]
|
|
type VoidRes = Result[void, int]
|
|
|
|
func worksVoid(): VoidRes = VoidRes.ok()
|
|
func worksVoid2(): VoidRes = result.ok()
|
|
func worksVoid3(): VoidRes = ok()
|
|
|
|
func failsVoid(): VoidRes = VoidRes.err(42)
|
|
func failsVoid2(): VoidRes = result.err(42)
|
|
func failsVoid3(): VoidRes = err(42)
|
|
|
|
let
|
|
vOk = worksVoid()
|
|
vOk2 = worksVoid2()
|
|
vOk3 = worksVoid3()
|
|
|
|
vErr = failsVoid()
|
|
vErr2 = failsVoid2()
|
|
vErr3 = failsVoid3()
|
|
|
|
doAssert vOk.isOk
|
|
doAssert vOk2.isOk
|
|
doAssert vOk3.isOk
|
|
doAssert (not vOk.isErr)
|
|
|
|
doAssert vErr.isErr
|
|
doAssert vErr2.isErr
|
|
doAssert vErr3.isErr
|
|
|
|
vOk.get()
|
|
vOk.unsafeGet()
|
|
vOk.expect("should never fail")
|
|
vOk[]
|
|
|
|
# Comparisons
|
|
doAssert (vOk == vOk)
|
|
doAssert (vErr == vErr)
|
|
doAssert (vOk != vErr)
|
|
|
|
# Mapping
|
|
doAssert vOk.map(proc (): int = 42).get() == 42
|
|
vOk.map(proc () = discard).get()
|
|
|
|
vOk.mapErr(proc(x: int): int = 10).get()
|
|
vOk.mapErr(proc(x: int) = discard).get()
|
|
|
|
doAssert vErr.mapErr(proc(x: int): int = 10).error() == 10
|
|
|
|
# string conversion
|
|
doAssert $vOk == "ok()"
|
|
doAssert $vErr == "err(42)"
|
|
|
|
# Question mark operator
|
|
func voidF(): VoidRes =
|
|
ok()
|
|
|
|
func voidF2(): Result[int, int] =
|
|
? voidF()
|
|
|
|
ok(42)
|
|
|
|
doAssert voidF2().isOk
|
|
|
|
# flatten
|
|
doAssert Result[VoidRes, int].ok(vOk).flatten() == vOk
|
|
doAssert Result[VoidRes, int].ok(vErr).flatten() == vErr
|
|
|
|
# Filter
|
|
doAssert vOk.filter(proc(): auto = Result[void, int].ok()) == vOk
|
|
doAssert vOk.filter(proc(): auto = Result[void, int].err(100)).error == 100
|
|
doAssert vErr.filter(proc(): auto = Result[void, int].err(100)) == vErr
|
|
|
|
block: # Result[T, void] aka `Opt`
|
|
type OptInt = Result[int, void]
|
|
|
|
func worksOpt(): OptInt = OptInt.ok(42)
|
|
func worksOpt2(): OptInt = result.ok(42)
|
|
func worksOpt3(): OptInt = ok(42)
|
|
|
|
func failsOpt(): OptInt = OptInt.err()
|
|
func failsOpt2(): OptInt = result.err()
|
|
func failsOpt3(): OptInt = err()
|
|
|
|
let
|
|
oOk = worksOpt()
|
|
oOk2 = worksOpt2()
|
|
oOk3 = worksOpt3()
|
|
|
|
oErr = failsOpt()
|
|
oErr2 = failsOpt2()
|
|
oErr3 = failsOpt3()
|
|
|
|
doAssert oOk.isOk
|
|
doAssert oOk2.isOk
|
|
doAssert oOk3.isOk
|
|
doAssert (not oOk.isErr)
|
|
|
|
doAssert oErr.isErr
|
|
doAssert oErr2.isErr
|
|
doAssert oErr3.isErr
|
|
|
|
# Comparisons
|
|
doAssert (oOk == oOk)
|
|
doAssert (oErr == oErr)
|
|
doAssert (oOk != oErr)
|
|
|
|
doAssert oOk.get() == oOk.unsafeGet()
|
|
oErr.error()
|
|
oErr.unsafeError()
|
|
|
|
# Mapping
|
|
doAssert oOk.map(proc(x: int): string = $x).get() == $oOk.get()
|
|
oOk.map(proc(x: int) = discard).get()
|
|
|
|
doAssert oOk.mapErr(proc(): int = 10).get() == oOk.get()
|
|
doAssert oOk.mapErr(proc() = discard).get() == oOk.get()
|
|
|
|
doAssert oErr.mapErr(proc(): int = 10).error() == 10
|
|
|
|
# string conversion
|
|
doAssert $oOk == "ok(42)"
|
|
doAssert $oErr == "none()"
|
|
|
|
proc optQuestion(): OptInt =
|
|
let v = ? oOk
|
|
ok(v)
|
|
|
|
doAssert optQuestion().isOk()
|
|
|
|
# Flatten
|
|
doAssert Result[OptInt, void].ok(oOk).flatten() == oOk
|
|
doAssert Result[OptInt, void].ok(oErr).flatten() == oErr
|
|
|
|
# Filter
|
|
doAssert oOk.filter(proc(x: int): auto = Result[void, void].ok()) == oOk
|
|
doAssert oOk.filter(proc(x: int): auto = Result[void, void].err()).isErr()
|
|
doAssert oErr.filter(proc(x: int): auto = Result[void, void].err()) == oErr
|
|
|
|
doAssert oOk.filter(proc(x: int): bool = true) == oOk
|
|
doAssert oOk.filter(proc(x: int): bool = false).isErr()
|
|
doAssert oErr.filter(proc(x: int): bool = true) == oErr
|
|
|
|
doAssert Opt.some(42).get() == 42
|
|
doAssert Opt.none(int).isNone()
|
|
|
|
block: # `cstring` dangling reference protection
|
|
type CSRes = Result[void, cstring]
|
|
|
|
func cstringF(s: string): CSRes =
|
|
when compiles(err(s)):
|
|
doAssert false
|
|
|
|
discard cstringF("test")
|
|
|
|
block: # Experiments
|
|
# Can formalise it into a template (https://github.com/arnetheduck/nim-result/issues/8)
|
|
template `?=`(v: untyped{nkIdent}, vv: Result): bool =
|
|
(let vr = vv; template v: auto {.used.} = unsafeGet(vr); vr.isOk)
|
|
|
|
if f ?= Result[int, string].ok(42):
|
|
doAssert f == 42
|
|
|
|
# TODO there's a bunch of operators that one could lift through magic - this
|
|
# is mainly an example
|
|
template `+`(self, other: Result): untyped =
|
|
## Perform `+` on the values of self and other, if both are ok
|
|
type R = type(other)
|
|
if self.isOk:
|
|
if other.isOk:
|
|
R.ok(self.value + other.value)
|
|
else:
|
|
R.err(other.error)
|
|
else:
|
|
R.err(self.error)
|
|
|
|
let rOk = Result[int, string].ok(42)
|
|
# Simple lifting..
|
|
doAssert (rOk + rOk)[] == rOk.value + rOk.value
|
|
|
|
iterator items[T, E](self: Result[T, E]): T =
|
|
## Iterate over result as if it were a collection of either 0 or 1 items
|
|
## TODO should a Result[seq[X]] iterate over items in seq? there are
|
|
## arguments for and against
|
|
if self.isOk:
|
|
yield self.value
|
|
|
|
# Iteration
|
|
var counter2 = 0
|
|
for v in rOk:
|
|
counter2 += 1
|
|
|
|
doAssert counter2 == 1, "one-item collection when set"
|
|
|
|
block: # Constants
|
|
# TODO https://github.com/nim-lang/Nim/issues/20699
|
|
type
|
|
WithOpt = object
|
|
opt: Opt[int]
|
|
const
|
|
noneWithOpt =
|
|
WithOpt(opt: Opt.none(int))
|
|
proc checkIt(v: WithOpt) =
|
|
doAssert v.opt.isNone()
|
|
checkIt(noneWithOpt)
|