ghc-9.14.1: GHC/HsToCore/Pmc.hs
{-# LANGUAGE FlexibleInstances #-}
{-# LANGUAGE GADTs #-}
{-# LANGUAGE LambdaCase #-}
{-# LANGUAGE StandaloneDeriving #-}
-- | This module coverage checks pattern matches. It finds
--
-- * Uncovered patterns, certifying non-exhaustivity
-- * Redundant equations
-- * Equations with an inaccessible right-hand-side
--
-- The algorithm is based on the paper
-- [Lower Your Guards: A Compositional Pattern-Match Coverage Checker"](https://dl.acm.org/doi/abs/10.1145/3408989)
--
-- There is an overview Figure 2 in there that's probably helpful.
-- Here is an overview of how it's implemented, which follows the structure of
-- the entry points such as 'pmcMatches':
--
-- 1. Desugar source syntax (like 'LMatch') to guard tree variants (like
-- 'GrdMatch'), with one of the desugaring functions (like 'desugarMatch').
-- See "GHC.HsToCore.Pmc.Desugar".
-- Follows Section 3.1 in the paper.
-- 2. Coverage check guard trees (with a function like 'checkMatch') to get a
-- 'CheckResult'. See "GHC.HsToCore.Pmc.Check".
-- The normalised refinement types 'Nabla' are tested for inhabitants by
-- "GHC.HsToCore.Pmc.Solver".
-- 3. Collect redundancy information into a 'CIRB' with a function such
-- as 'cirbsMatch'. Follows the R function from Figure 6 of the paper.
-- 4. Format and report uncovered patterns and redundant equations ('CIRB')
-- with 'formatReportWarnings'. Basically job of the G function, plus proper
-- pretty printing of the warnings (Section 5.4 of the paper).
-- 5. Return 'Nablas' reaching syntactic sub-components for
-- Note [Long-distance information]. Collected by functions such as
-- 'ldiMatch'. See Section 4.1 of the paper.
module GHC.HsToCore.Pmc (
-- Checking and printing
pmcPatBind, pmcMatches, pmcGRHSs, pmcRecSel,
isMatchContextPmChecked, isMatchContextPmChecked_SinglePat,
-- See Note [Long-distance information]
addTyCs, addCoreScrutTmCs, addHsScrutTmCs, getLdiNablas,
getNFirstUncovered
) where
import GHC.Prelude
import GHC.HsToCore.Errors.Types
import GHC.HsToCore.Pmc.Types
import GHC.HsToCore.Pmc.Utils
import GHC.HsToCore.Pmc.Desugar
import GHC.HsToCore.Pmc.Check
import GHC.HsToCore.Pmc.Solver
import GHC.Types.Basic (Origin(..), isDoExpansionGenerated)
import GHC.Core
import GHC.Driver.DynFlags
import GHC.Hs
import GHC.Types.Id
import GHC.Types.SrcLoc
import GHC.Utils.Misc
import GHC.Utils.Outputable
import GHC.Utils.Panic
import GHC.Types.Var (EvVar, Var (..))
import GHC.Types.Id.Info
import GHC.Tc.Utils.TcType (evVarPred)
import {-# SOURCE #-} GHC.HsToCore.Expr (dsLExpr)
import GHC.HsToCore.Monad
import GHC.Data.Bag
import GHC.Data.OrdList
import Control.Monad (when, unless, forM_)
import qualified Data.Semigroup as Semi
import Data.List.NonEmpty ( NonEmpty(..) )
import qualified Data.List.NonEmpty as NE
import Data.Coerce
import GHC.Tc.Utils.Monad
--
-- * Exported entry points to the checker
--
-- | A non-empty delta that is initialised from the ambient refinement type
-- capturing long-distance information, or the trivially habitable 'Nablas' if
-- the former is uninhabited.
-- See Note [Recovering from unsatisfiable pattern-matching constraints].
getLdiNablas :: DsM Nablas
getLdiNablas = do
nablas <- getPmNablas
isInhabited nablas >>= \case
True -> pure nablas
False -> pure initNablas
-- | We need to call the Hs desugarer to get the Core of a let-binding or where
-- clause. We don't want to run the coverage checker when doing so! Efficiency
-- is one concern, but also a lack of properly set up long-distance information
-- might trigger warnings that we normally wouldn't emit.
noCheckDs :: DsM a -> DsM a
noCheckDs = updTopFlags (\dflags -> foldl' wopt_unset dflags allPmCheckWarnings)
-- | Check a pattern binding (let, where) for exhaustiveness.
pmcPatBind :: DsMatchContext -> Id -> Pat GhcTc -> DsM Nablas
pmcPatBind ctxt@(DsMatchContext match_ctxt loc) var p
= mb_discard_warnings $ do
!missing <- getLdiNablas
pat_bind <- noCheckDs $ desugarPatBind loc var p
tracePm "pmcPatBind {" (vcat [ppr ctxt, ppr var, ppr p, ppr pat_bind, ppr missing])
result <- unCA (checkPatBind pat_bind) missing
let ldi = ldiGRHS $ ( \ pb -> case pb of PmPatBind grhs -> grhs) $ cr_ret result
tracePm "pmcPatBind }: " $
vcat [ text "cr_uncov:" <+> ppr (cr_uncov result)
, text "ldi:" <+> ppr ldi ]
formatReportWarnings ReportPatBind ctxt [var] result
return ldi
where
-- See Note [pmcPatBind doesn't warn on pattern guards]
mb_discard_warnings
= if want_pmc match_ctxt
then id
else discardWarningsDs
want_pmc PatBindRhs = True
want_pmc LazyPatCtx = True
want_pmc (StmtCtxt stmt_ctxt) =
case stmt_ctxt of
PatGuard {} -> False
_ -> True
want_pmc _ = False
-- | Exhaustive for guard matches, is used for guards in pattern bindings and
-- in @MultiIf@ expressions. Returns the 'Nablas' covered by the RHSs.
pmcGRHSs
:: HsMatchContextRn -- ^ Match context, for warning messages
-> GRHSs GhcTc (LHsExpr GhcTc) -- ^ The GRHSs to check
-> DsM (NonEmpty Nablas) -- ^ Covered 'Nablas' for each RHS, for long
-- distance info
pmcGRHSs hs_ctxt guards@(GRHSs _ grhss _) = do
let combined_loc = foldl1 combineSrcSpans (NE.map getLocA grhss)
ctxt = DsMatchContext hs_ctxt combined_loc
!missing <- getLdiNablas
matches <- noCheckDs $ desugarGRHSs combined_loc empty guards
tracePm "pmcGRHSs" (hang (vcat [ppr ctxt
, text "Guards:"])
2
(pprGRHSs hs_ctxt guards $$ ppr missing))
result <- unCA (checkGRHSs matches) missing
tracePm "}: " (ppr (cr_uncov result))
formatReportWarnings ReportGRHSs ctxt [] result
return (ldiGRHSs (cr_ret result))
-- | Check a list of syntactic 'Match'es (part of case, functions, etc.), each
-- with a 'Pat' and one or more 'GRHSs':
--
-- @
-- f x y | x == y = 1 -- match on x and y with two guarded RHSs
-- | otherwise = 2
-- f _ _ = 3 -- clause with a single, un-guarded RHS
-- @
--
-- Returns one non-empty 'Nablas' for 1.) each pattern of a 'Match' and 2.)
-- each of a 'Match'es 'GRHS' for Note [Long-distance information].
--
-- Special case: When there are /no matches/, then the function assumes it
-- checks an @-XEmptyCase@ with only a single match variable.
-- See Note [Checking EmptyCase].
pmcMatches
:: Origin
-> DsMatchContext -- ^ Match context, for warnings messages
-> [Id] -- ^ Match variables, i.e. x and y above
-> [LMatch GhcTc (LHsExpr GhcTc)] -- ^ List of matches
-> DsM [(Nablas, NonEmpty Nablas)] -- ^ One covered 'Nablas' per Match and
-- GRHS, for long distance info.
pmcMatches origin ctxt vars matches = {-# SCC "pmcMatches" #-} do
-- We have to force @missing@ before printing out the trace message,
-- otherwise we get interleaved output from the solver. This function
-- should be strict in @missing@ anyway!
!missing <- getLdiNablas
tracePm "pmcMatches {" $
hang (vcat [ppr origin, ppr ctxt, ppr vars, text "Matches:"])
2
((ppr matches) $$ (text "missing:" <+> ppr missing))
case NE.nonEmpty matches of
Nothing -> do
-- This must be an -XEmptyCase. See Note [Checking EmptyCase]
let var = only vars
empty_case <- noCheckDs $ desugarEmptyCase var
result <- unCA (checkEmptyCase empty_case) missing
tracePm "}: " (ppr (cr_uncov result))
formatReportWarnings ReportEmptyCase ctxt vars result
return []
Just matches -> do
matches <- {-# SCC "desugarMatches" #-}
noCheckDs $ desugarMatches vars matches
tracePm "desugared matches" (ppr matches)
result <- {-# SCC "checkMatchGroup" #-}
unCA (checkMatchGroup matches) missing
tracePm "}: " (ppr (cr_uncov result))
unless (isDoExpansionGenerated origin) -- Do expansion generated code shouldn't emit overlapping warnings
({-# SCC "formatReportWarnings" #-}
formatReportWarnings ReportMatchGroup ctxt vars result)
return (NE.toList (ldiMatchGroup (cr_ret result)))
{-
Note [Detecting incomplete record selectors]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
This Note describes the implementation of
GHC proposal 516 "Add warning for incomplete record selectors".
A **partial field** is a field that does not belong to every constructor of the
corresponding datatype.
A **partial selector occurrence** is a use of a record selector for a partial
field, either as a selector function in an expression, or as the solution to a
HasField constraint.
Partial selector occurrences desugar to case expressions which may crash at
runtime:
data T a where
T1 :: T Int
T2 {sel :: Int} :: T Bool
urgh :: T a -> Int
urgh x = sel x
===>
urgh x = case x of
T1 -> error "no record field sel"
T2 f -> f
As such, it makes sense to warn about such potential crashes.
We do so whenever -Wincomplete-record-selectors is present, and we utilise
the pattern-match coverage checker for precise results, because there are many
uses of selectors for partial fields which are in fact dynamically safe.
Pmc can detect two very common safe uses for which we will not warn:
(LDI) Ambient pattern-matches unleash Note [Long-distance information] that
render a naively flagged partial selector occurrence safe, as in
ldi :: T a -> Int
ldi T1 = 0
ldi arg = sel arg
We should not warn here, because `arg` cannot be `T1`.
(RES) Constraining the result type of a GADT such as T might render
naively flagged partial selector occurrences safe, as in
resTy :: T Bool -> Int
resTy = sel
Here, `T1 :: T Int` is ruled out because it has the wrong result type.
Additionally, we want to support incomplete -XOverloadedRecordDot access as
well, in either the (LDI) use case or the (RES) use case:
data Dot = No | Yes { sel2 :: Int }
dot d = d.sel2 -- should warn
ldiDot No = 0
ldiDot d = d.sel2 -- should not warn
resTyDot :: T Bool -> Int
resTyDot x = x.sel -- should not warn
From a user's point of view, function `ldiDot` looks very like example `ldi` and
`resTyDot` looks very like `resTy`. But from an /implementation/ point of view
they are very different: both `ldiDot` and `resTyDot` simply emit `HasField`
constraints, and it is those constraints that the implementation must use to
determine incompleteness.
Furthermore, HasField constraints allow to delay the completeness check from
the field access site to a caller, as in test cases TcIncompleteRecSel and T24891:
accessDot :: HasField "sel2" t Int => t -> Int
accessDot x = x.sel2 -- getField @Symbol @"sel2" @t @Int x
solveDot :: Dot -> Int
solveDot = accessDot
We should warn in `solveDot`, but not in `accessDot`.
Here is how we achieve all this in the implementation:
(IRS1) When renaming a record selector in `mkOneRecordSelector`,
we precompute the constructors the selector succeeds on.
That would be `T2` for `sel` because `sel (T2 42)` succeeds,
and `Yes` for `sel2` because `sel2 (Yes 13)` succeeds.
We store this information in the `sel_cons` field of `RecSelId`.
(Remember, the same field may occur in several constructors of the data
type; hence the selector may succeed on more than one constructor.)
We generate warnings for incomplete record selectors in two places:
* Mainly: in GHC.HsToCore.Expr.ds_app (see (IRS2-5) below)
* Plus: in GHC.Tc.Instance.Class.matchHassField (see (IRS6-7) below)
(IRS2) In function `ldi`, we have a record selector application `sel arg`.
This situation is detected `GHC.HsToCore.Expr.ds_app_rec_sel`, when the
record selector is applied to at least one argument. We call out to the
pattern-match checker to determine whether use of the selector is safe,
by calling GHC.HsToCore.Pmc.pmcRecSel, passing the `RecSelId` `sel` as
well as `arg`.
The pattern-match checker reduces the partial-selector-occurrence problem to
a complete-match problem by adding a negative constructor constraint such as
`arg /~ T2` for every constructor in the precomputed `rsi_def . sel_cons` of
`sel`. (Recall that these were exactly the constructors which define a field
`sel`.) `pmcRecSel` then tests
case arg of {}
for completeness. Any incomplete match, such as in the original `urgh`, must
reference a constructor that does not have field `sel`, such as `T1`.
In case of `urgh`, `T1` is indeed the case that we report as inexhaustive.
However, in function `ldi`, we have *both* the result type of
`arg::T a` (boring, but see (IRS3)) as well as Note [Long-distance information]
about `arg` from the ambient match, and the latter lists the constraint
`arg /~ T1`. Consequently, since `arg` is neither `T1` nor `T2` in the
reduced problem, the match is exhaustive and the use of the record selector
safe.
(IRS3) In function `resTy`, the record selector is unsaturated, but the result type
ensures a safe use of the selector.
This situation is also detected in `GHC.HsToCore.Expr.ds_app_rec_sel`.
THe selector is elaborated with its type arguments; we simply match on
desugared Core `sel @Bool :: T Bool -> Int` to learn the result type `T Bool`.
We again call `pmcRecSel`, but this time with a fresh dummy Id `ds::T Bool`.
(IRS4) In case of an unsaturated record selector that is *not* applied to any type
argument after elaboration (e.g. in `urgh2 = sel2 :: Dot -> Int`), we simply
produce a warning about all `sel_cons`; no need to call `pmcRecSel`.
This happens in `ds_app_rec_sel`
Finally, there are two more items addressing -XOverloadedRecordDot:
(IRS5) With -XOverloadedDot, all occurrences of (r.x), such as in `ldiDot` and
`accessDot` above, are warned about as follows. `r.x` is parsed as
`HsGetField` in `HsExpr`; which is then expanded (in `rnExpr`) to a call to
`getField`. For example, consider:
ldiDot No = 0
ldiDot x = x.sel2 -- should not warn
The `d.sel2` in the RHS generates
getField @GHC.Types.Symbol @"sel2" @Dot @Int
($dHasField :: HasField "sel2" Dot Int) x
where
$dHasField = sel2 |> (co :: Dot -> Int ~R# HasField "sel2" Dot Int)
We spot this `getField` application in `GHC.HsToCore.Expr.ds_app_var`,
and treat it exactly like (IRS2) and (IRS3).
Note carefully that doing this in the desugarer allows us to account for the
long-distance info about `x`; even though `sel2` is partial, we don't want
to warn about `x.sel2` in this example.
(IRS6) Finally we have
solveDot :: Dot -> Int
solveDot = accessDot
No field-accesses or selectors in sight! From the RHS we get the constraint
[W] HasField @"sel2" @Dot @Int`
The only time we can generate a warning is when we solve this constraint,
in `GHC.Tc.Instance.Class.matchHasField`, generating a call to the (partial)
selector. We have no hope of exploiting long-distance info here.
(IRS7) BUT, look back at `ldiDot`. Doesn't `matchHasField` /also/ generate a
warning for the `HasField` constraint arising from `x.sel2`? We don't
want that, because the desugarer will catch it: see (IRS5). So we suppress
the (IRS6) warning in the typechecker for a `HasField` constraint that
arises from a record-dot HsGetField occurrence. Happily, this is easy to do
by looking at its `CtOrigin`. Tested in T24891.
-}
pmcRecSel :: Id -- ^ Id of the selector
-> CoreExpr -- ^ Core expression of the argument to the selector
-> DsM ()
-- See (IRS2,3) in Note [Detecting incomplete record selectors]
pmcRecSel sel_id arg
| RecSelId{ sel_cons = rec_sel_info } <- idDetails sel_id
, RSI { rsi_def = cons_w_field, rsi_undef = cons_wo_field } <- rec_sel_info
, not (null cons_wo_field)
= do { !missing <- getLdiNablas
; tracePm "pmcRecSel {" (ppr sel_id)
; CheckResult{ cr_ret = PmRecSel{ pr_arg_var = arg_id }, cr_uncov = uncov_nablas }
<- unCA (checkRecSel (PmRecSel () arg cons_w_field)) missing
; tracePm "}: " $ ppr uncov_nablas
; inhabited <- isInhabited uncov_nablas
; when inhabited $ warn_incomplete arg_id uncov_nablas }
| otherwise
= return ()
where
sel_name = varName sel_id
warn_incomplete arg_id uncov_nablas = do
dflags <- getDynFlags
let maxPatterns = maxUncoveredPatterns dflags
unc_examples <- getNFirstUncovered MinimalCover [arg_id] (maxPatterns + 1) uncov_nablas
let cons = [con | unc_example <- unc_examples
, Just (PACA (PmAltConLike con) _ _) <- [lookupSolution unc_example arg_id]]
tracePm "unc-ex" (ppr cons $$ ppr unc_examples)
diagnosticDs $ DsIncompleteRecordSelector sel_name cons maxPatterns
{- Note [pmcPatBind doesn't warn on pattern guards]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
@pmcPatBind@'s main purpose is to check vanilla pattern bindings, like
@x :: Int; Just x = e@, which is in a @PatBindRhs@ context.
But its caller is also called for individual pattern guards in a @StmtCtxt@.
For example, both pattern guards in @f x y | True <- x, False <- y = ...@ will
go through this function. It makes no sense to report pattern match warnings
for these pattern guards:
* Pattern guards may well fail. Fall-through is not an unrecoverable panic,
but rather behavior the programmer expects, so inexhaustivity should not be
reported.
* Redundancy is already reported for the whole GRHS via one of the other
exported coverage checking functions. Also, reporting individual redundant
guards is... redundant. See #17646.
However, we should not skip pattern-match checking altogether, as it may reveal
important long-distance information. One example is described in
Note [Long-distance information in do notation] in GHC.HsToCore.Expr.
Instead, we simply discard warnings when in pattern-guards, by using the function
discardWarningsDs.
-}
--
-- * Collecting long-distance information
--
ldiMatchGroup :: PmMatchGroup Post -> NonEmpty (Nablas, NonEmpty Nablas)
ldiMatchGroup (PmMatchGroup matches) = ldiMatch <$> matches
ldiMatch :: PmMatch Post -> (Nablas, NonEmpty Nablas)
ldiMatch (PmMatch { pm_pats = red, pm_grhss = grhss }) =
(rs_cov red, ldiGRHSs grhss)
ldiGRHSs :: PmGRHSs Post -> NonEmpty Nablas
ldiGRHSs (PmGRHSs { pgs_grhss = grhss }) = ldiGRHS <$> grhss
ldiGRHS :: PmGRHS Post -> Nablas
ldiGRHS (PmGRHS { pg_grds = red }) = rs_cov red
--
-- * Collecting redundancy information
--
-- | The result of redundancy checking:
-- * RHSs classified as /C/overed, /I/naccessible and /R/edundant
-- * And redundant /B/ang patterns. See Note [Dead bang patterns].
data CIRB
= CIRB
{ cirb_cov :: !(OrdList SrcInfo) -- ^ Covered clauses
, cirb_inacc :: !(OrdList SrcInfo) -- ^ Inaccessible clauses
, cirb_red :: !(OrdList SrcInfo) -- ^ Redundant clauses
, cirb_bangs :: !(OrdList SrcInfo) -- ^ Redundant bang patterns
}
instance Semigroup CIRB where
CIRB a b c d <> CIRB e f g h = CIRB (a <> e) (b <> f) (c <> g) (d <> h)
where (<>) = (Semi.<>)
instance Monoid CIRB where
mempty = CIRB mempty mempty mempty mempty
-- See Note [Determining inaccessible clauses]
ensureOneNotRedundant :: CIRB -> CIRB
ensureOneNotRedundant ci = case ci of
CIRB { cirb_cov = NilOL, cirb_inacc = NilOL, cirb_red = ConsOL r rs }
-> ci { cirb_inacc = unitOL r, cirb_red = rs }
_ -> ci
-- | Only adds the redundant bangs to the @CIRB@ if there is at least one
-- non-redundant 'SrcInfo'. There is no point in remembering a redundant bang
-- if the whole match is redundant!
addRedundantBangs :: OrdList SrcInfo -> CIRB -> CIRB
addRedundantBangs _red_bangs cirb@CIRB { cirb_cov = NilOL, cirb_inacc = NilOL } =
cirb
addRedundantBangs red_bangs cirb =
cirb { cirb_bangs = cirb_bangs cirb Semi.<> red_bangs }
-- | Checks the 'Nablas' in a 'RedSets' for inhabitants and returns
-- 1. Whether the Covered set was inhabited
-- 2. Whether the Diverging set was inhabited
-- 3. All source bangs whose 'Nablas' were empty, which means they are
-- redundant.
testRedSets :: RedSets -> DsM (Bool, Bool, OrdList SrcInfo)
testRedSets RedSets { rs_cov = cov, rs_div = div, rs_bangs = bangs } = do
is_covered <- isInhabited cov
may_diverge <- isInhabited div
red_bangs <- flip mapMaybeM (fromOL bangs) $ \(nablas, bang) ->
isInhabited nablas >>= \case
True -> pure Nothing
False -> pure (Just bang)
pure (is_covered, may_diverge, toOL red_bangs)
cirbsMatchGroup :: PmMatchGroup Post -> DsM CIRB
cirbsMatchGroup (PmMatchGroup matches) =
Semi.sconcat <$> traverse cirbsMatch matches
cirbsMatch :: PmMatch Post -> DsM CIRB
cirbsMatch PmMatch { pm_pats = red, pm_grhss = grhss } = do
(_is_covered, may_diverge, red_bangs) <- testRedSets red
-- Don't look at is_covered: If it is True, all children are redundant anyway,
-- unless there is a 'considerAccessible', which may break that rule
-- intentionally. See Note [considerAccessible] in "GHC.HsToCore.Pmc.Check".
cirb <- cirbsGRHSs grhss
pure $ addRedundantBangs red_bangs
-- See Note [Determining inaccessible clauses]
$ applyWhen may_diverge ensureOneNotRedundant
$ cirb
cirbsGRHSs :: PmGRHSs Post -> DsM CIRB
cirbsGRHSs (PmGRHSs { pgs_grhss = grhss }) = Semi.sconcat <$> traverse cirbsGRHS grhss
cirbsGRHS :: PmGRHS Post -> DsM CIRB
cirbsGRHS PmGRHS { pg_grds = red, pg_rhs = info } = do
(is_covered, may_diverge, red_bangs) <- testRedSets red
let cirb | is_covered = mempty { cirb_cov = unitOL info }
| may_diverge = mempty { cirb_inacc = unitOL info }
| otherwise = mempty { cirb_red = unitOL info }
pure (addRedundantBangs red_bangs cirb)
cirbsEmptyCase :: PmEmptyCase -> DsM CIRB
cirbsEmptyCase _ = pure mempty
cirbsPatBind :: PmPatBind Post -> DsM CIRB
cirbsPatBind = coerce cirbsGRHS
{- Note [Determining inaccessible clauses]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Consider
f _ True = ()
f () True = ()
f _ _ = ()
Is f's second clause redundant? The perhaps surprising answer is, no, it isn't!
@f (error "boom") False@ will force the error with clause 2, but will return
() if it was deleted, so clearly not redundant. Yet for now combination of
arguments we can ever reach clause 2's RHS, so we say it has inaccessible RHS
(as opposed to being completely redundant).
We detect an inaccessible RHS simply by pretending it's redundant, until we see
-}
--
-- * Formatting and reporting warnings
--
-- | A datatype to accommodate the different call sites of
-- 'formatReportWarnings'. Used for extracting 'CIRB's from a concrete 'ann'
-- through 'collectInMode'. Since this is only possible for a couple of
-- well-known 'ann's, this is a GADT.
data FormatReportWarningsMode ann where
ReportPatBind :: FormatReportWarningsMode (PmPatBind Post)
ReportGRHSs :: FormatReportWarningsMode (PmGRHSs Post)
ReportMatchGroup:: FormatReportWarningsMode (PmMatchGroup Post)
ReportEmptyCase:: FormatReportWarningsMode PmEmptyCase
deriving instance Eq (FormatReportWarningsMode ann)
-- | A function collecting 'CIRB's for each of the different
-- 'FormatReportWarningsMode's.
collectInMode :: FormatReportWarningsMode ann -> ann -> DsM CIRB
collectInMode ReportPatBind = cirbsPatBind
collectInMode ReportGRHSs = cirbsGRHSs
collectInMode ReportMatchGroup = cirbsMatchGroup
collectInMode ReportEmptyCase = cirbsEmptyCase
-- | Given a 'FormatReportWarningsMode', this function will emit warnings
-- for a 'CheckResult'.
formatReportWarnings :: FormatReportWarningsMode ann -> DsMatchContext -> [Id] -> CheckResult ann -> DsM ()
formatReportWarnings report_mode ctx vars cr@CheckResult { cr_ret = ann } = do
cov_info <- collectInMode report_mode ann
dflags <- getDynFlags
reportWarnings dflags report_mode ctx vars cr{cr_ret=cov_info}
-- | Issue all the warnings
-- (redundancy, inaccessibility, exhaustiveness, redundant bangs).
reportWarnings :: DynFlags -> FormatReportWarningsMode ann -> DsMatchContext -> [Id] -> CheckResult CIRB -> DsM ()
reportWarnings dflags report_mode (DsMatchContext kind loc) vars
CheckResult { cr_ret = CIRB { cirb_inacc = inaccessible_rhss
, cirb_red = redundant_rhss
, cirb_bangs = redundant_bangs }
, cr_uncov = uncovered
, cr_approx = precision }
= when (flag_i || flag_u || flag_b) $ do
unc_examples <- getNFirstUncovered gen_mode vars (maxPatterns + 1) uncovered
let exists_r = flag_i && notNull redundant_rhss
exists_i = flag_i && notNull inaccessible_rhss
exists_u = flag_u && notNull unc_examples
exists_b = flag_b && notNull redundant_bangs
approx = precision == Approximate
when (approx && (exists_u || exists_i)) $
putSrcSpanDs loc (diagnosticDs (DsMaxPmCheckModelsReached (maxPmCheckModels dflags)))
when exists_b $ forM_ redundant_bangs $ \(SrcInfo (L l q)) ->
putSrcSpanDs l (diagnosticDs (DsRedundantBangPatterns kind q))
when exists_r $ forM_ redundant_rhss $ \(SrcInfo (L l q)) ->
putSrcSpanDs l (diagnosticDs (DsOverlappingPatterns kind q))
when exists_i $ forM_ inaccessible_rhss $ \(SrcInfo (L l q)) ->
putSrcSpanDs l (diagnosticDs (DsInaccessibleRhs kind q))
when exists_u $
putSrcSpanDs loc (diagnosticDs (DsNonExhaustivePatterns kind check_type maxPatterns vars unc_examples))
where
flag_i = overlapping dflags kind
flag_u = exhaustive dflags kind
flag_b = redundantBang dflags
check_type = ExhaustivityCheckType (exhaustiveWarningFlag kind)
gen_mode = case report_mode of -- See Note [Case split inhabiting patterns]
ReportEmptyCase -> CaseSplitTopLevel
_ -> MinimalCover
maxPatterns = maxUncoveredPatterns dflags
getNFirstUncovered :: GenerateInhabitingPatternsMode -> [Id] -> Int -> Nablas -> DsM [Nabla]
getNFirstUncovered mode vars n (MkNablas nablas) = go n (bagToList nablas)
where
go 0 _ = pure []
go _ [] = pure []
go n (nabla:nablas) = do
front <- generateInhabitingPatterns mode vars n nabla
back <- go (n - length front) nablas
pure (front ++ back)
--
-- * Adding external long-distance information
--
-- | Locally update 'dsl_nablas' with the given action, but defer evaluation
-- with 'unsafeInterleaveM' in order not to do unnecessary work.
locallyExtendPmNablas :: DsM a -> (Nablas -> DsM Nablas) -> DsM a
locallyExtendPmNablas k ext = do
nablas <- getLdiNablas
nablas' <- unsafeInterleaveM $ ext nablas
updPmNablas nablas' k
-- | Add in-scope type constraints if the coverage checker might run and then
-- run the given action.
addTyCs :: Origin -> Bag EvVar -> DsM a -> DsM a
addTyCs origin ev_vars thing_inside
| isEmptyBag ev_vars
= thing_inside -- Very common fast path
| otherwise
= do { dflags <- getDynFlags
; if needToRunPmCheck dflags origin
then locallyExtendPmNablas thing_inside $ \nablas ->
addPhiCtsNablas nablas (PhiTyCt . evVarPred <$> ev_vars)
else thing_inside }
-- | Add equalities for the 'CoreExpr' scrutinees to the local 'DsM' environment,
-- e.g. when checking a case expression:
-- case e of x { matches }
-- When checking matches we record that (x ~ e) where x is the initial
-- uncovered. All matches will have to satisfy this equality.
-- This is also used for the Arrows \cases command, where these equalities have
-- to be added for multiple scrutinees rather than just one.
addCoreScrutTmCs :: [CoreExpr] -> [Id] -> DsM a -> DsM a
addCoreScrutTmCs [] _ k = k
addCoreScrutTmCs (scr:scrs) (x:xs) k
= locallyExtendPmNablas (addCoreScrutTmCs scrs xs k) $ \nablas ->
addPhiCtsNablas nablas (unitBag (PhiCoreCt x scr))
addCoreScrutTmCs _ _ _ = panic "addCoreScrutTmCs: numbers of scrutinees and match ids differ"
-- | 'addCoreScrutTmCs', but desugars the 'LHsExpr's first.
addHsScrutTmCs :: [LHsExpr GhcTc] -> [Id] -> DsM a -> DsM a
addHsScrutTmCs scrs vars k = do
scr_es <- traverse dsLExpr scrs
addCoreScrutTmCs scr_es vars k
{- Note [Long-distance information]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Consider
data Color = R | G | B
f :: Color -> Int
f R = …
f c = … (case c of
G -> True
B -> False) …
Humans can make the "long-distance connection" between the outer pattern match
and the nested case pattern match to see that the inner pattern match is
exhaustive: @c@ can't be @R@ anymore because it was matched in the first clause
of @f@.
To achieve similar reasoning in the coverage checker, we keep track of the set
of values that can reach a particular program point (often loosely referred to
as "Covered set") in 'GHC.HsToCore.Monad.dsl_nablas'.
We fill that set with Covered Nablas returned by the exported checking
functions, which the call sites put into place with
'GHC.HsToCore.Monad.updPmNablas'.
Call sites also extend this set with facts from type-constraint dictionaries,
case scrutinees, etc. with the exported functions 'addTyCs', 'addCoreScrutTmCs'
and 'addHsScrutTmCs'.
Note [Recovering from unsatisfiable pattern-matching constraints]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Consider the following code (see #12957 and #15450):
f :: Int ~ Bool => ()
f = case True of { False -> () }
We want to warn that the pattern-matching in `f` is non-exhaustive. But GHC
used not to do this; in fact, it would warn that the match was /redundant/!
This is because the constraint (Int ~ Bool) in `f` is unsatisfiable, and the
coverage checker deems any matches with unsatisfiable constraint sets to be
unreachable.
We make sure to always start from an inhabited 'Nablas' by calling
'getLdiNablas', which falls back to the trivially inhabited 'Nablas' if the
long-distance info returned by 'GHC.HsToCore.Monad.getPmNablas' is empty.
-}