internal/pkg: make checksum available to cure

This enables deduplication by value as implemented in execArtifact.

Signed-off-by: Ophestra <cat@gensokyo.uk>

1 files changed, 54 insertions, 40 deletions

diff --git a/internal/pkg/pkg.go b/internal/pkg/pkg.go
index 65bf5e17..738eb5c0 100644
--- a/internal/pkg/pkg.go
+++ b/internal/pkg/pkg.go
@@ -187,7 +187,7 @@ type FContext struct {
 	TContext
 
 	// Cured top-level dependencies looked up by Pathname.
-	deps map[Artifact]*check.Absolute
+	deps map[Artifact]cureRes
 }
 
 // InvalidLookupError is the identifier of non-dependency [Artifact] looked up
@@ -200,15 +200,17 @@ func (e InvalidLookupError) Error() string {
 
 var _ error = InvalidLookupError{}
 
-// Pathname returns the identifier pathname of an [Artifact]. Calling Pathname
-// with an [Artifact] not part of the slice returned by [Artifact.Dependencies]
-// panics.
-func (f *FContext) Pathname(a Artifact) *check.Absolute {
-	if p, ok := f.deps[a]; ok {
-		return p
-	} else {
-		panic(InvalidLookupError(f.cache.Ident(a).Value()))
+// GetArtifact returns the identifier pathname and checksum of an [Artifact].
+// Calling Pathname with an [Artifact] not part of the slice returned by
+// [Artifact.Dependencies] panics.
+func (f *FContext) GetArtifact(a Artifact) (
+	pathname *check.Absolute,
+	checksum unique.Handle[Checksum],
+) {
+	if res, ok := f.deps[a]; ok {
+		return res.pathname, res.checksum
 	}
+	panic(InvalidLookupError(f.cache.Ident(a).Value()))
 }
 
 // An Artifact is a read-only reference to a piece of data that may be created
@@ -378,6 +380,12 @@ const (
 	checksumLinknamePrefix = "../" + dirChecksum + "/"
 )
 
+// cureRes are the non-error results returned by [Cache.Cure].
+type cureRes struct {
+	pathname *check.Absolute
+	checksum unique.Handle[Checksum]
+}
+
 // A pendingArtifactDep is a dependency [Artifact] pending concurrent curing,
 // subject to the cures limit. Values pointed to by result addresses are safe
 // to access after the [sync.WaitGroup] associated with this pendingArtifactDep
@@ -389,7 +397,7 @@ type pendingArtifactDep struct {
 
 	// Address of result pathname populated during [Cache.Cure] and dereferenced
 	// if curing succeeds.
-	resP **check.Absolute
+	resP *cureRes
 
 	// Address of result error slice populated during [Cache.Cure], dereferenced
 	// after acquiring errsMu if curing fails. No additional action is taken,
@@ -437,7 +445,7 @@ type Cache struct {
 	checksumMu sync.RWMutex
 
 	// Identifier to content pair cache.
-	ident map[unique.Handle[ID]]Checksum
+	ident map[unique.Handle[ID]]unique.Handle[Checksum]
 	// Identifier to error pair for unrecoverably faulted [Artifact].
 	identErr map[unique.Handle[ID]]error
 	// Pending identifiers, accessed through Cure for entries not in ident.
@@ -650,7 +658,7 @@ func (c *Cache) Scrub(checks int) error {
 	c.checksumMu.Lock()
 	defer c.checksumMu.Unlock()
 
-	c.ident = make(map[unique.Handle[ID]]Checksum)
+	c.ident = make(map[unique.Handle[ID]]unique.Handle[Checksum])
 	c.identErr = make(map[unique.Handle[ID]]error)
 	c.artifact.Clear()
 
@@ -848,7 +856,7 @@ func (c *Cache) Scrub(checks int) error {
 // identifier is stored in identPending and a non-nil channel is returned.
 func (c *Cache) loadOrStoreIdent(id unique.Handle[ID]) (
 	done chan<- struct{},
-	checksum Checksum,
+	checksum unique.Handle[Checksum],
 	err error,
 ) {
 	var ok bool
@@ -887,14 +895,14 @@ func (c *Cache) loadOrStoreIdent(id unique.Handle[ID]) (
 func (c *Cache) finaliseIdent(
 	done chan<- struct{},
 	id unique.Handle[ID],
-	checksum *Checksum,
+	checksum unique.Handle[Checksum],
 	err error,
 ) {
 	c.identMu.Lock()
 	if err != nil {
 		c.identErr[id] = err
 	} else {
-		c.ident[id] = *checksum
+		c.ident[id] = checksum
 	}
 	delete(c.identPending, id)
 	c.identMu.Unlock()
@@ -1091,7 +1099,7 @@ func (e DependencyError) Error() string {
 // calls to Cure are not subject to the cures limit.
 func (c *Cache) Cure(a Artifact) (
 	pathname *check.Absolute,
-	checksum Checksum,
+	checksum unique.Handle[Checksum],
 	err error,
 ) {
 	if c.threshold > 0 {
@@ -1181,7 +1189,7 @@ func (e *DependencyCureError) Error() string {
 // cure implements Cure without checking the full dependency graph.
 func (c *Cache) cure(a Artifact) (
 	pathname *check.Absolute,
-	checksum Checksum,
+	checksum unique.Handle[Checksum],
 	err error,
 ) {
 	id := c.Ident(a)
@@ -1193,7 +1201,7 @@ func (c *Cache) cure(a Artifact) (
 	defer func() {
 		if err != nil {
 			pathname = nil
-			checksum = Checksum{}
+			checksum = unique.Handle[Checksum]{}
 		}
 	}()
 
@@ -1202,7 +1210,7 @@ func (c *Cache) cure(a Artifact) (
 	if done == nil {
 		return
 	} else {
-		defer func() { c.finaliseIdent(done, id, &checksum, err) }()
+		defer func() { c.finaliseIdent(done, id, checksum, err) }()
 	}
 
 	_, err = os.Lstat(pathname.String())
@@ -1211,7 +1219,12 @@ func (c *Cache) cure(a Artifact) (
 		if name, err = os.Readlink(pathname.String()); err != nil {
 			return
 		}
-		err = Decode(&checksum, path.Base(name))
+		buf := c.getIdentBuf()
+		err = Decode((*Checksum)(buf[:]), path.Base(name))
+		if err == nil {
+			checksum = unique.Make(Checksum(buf[:]))
+		}
+		c.putIdentBuf(buf)
 		return
 	}
 	if !errors.Is(err, os.ErrNotExist) {
@@ -1234,8 +1247,8 @@ func (c *Cache) cure(a Artifact) (
 	var checksumPathname *check.Absolute
 	var checksumFi os.FileInfo
 	if kc, ok := a.(KnownChecksum); ok {
-		checksum = kc.Checksum()
-		checksums = Encode(checksum)
+		checksum = unique.Make(kc.Checksum())
+		checksums = Encode(checksum.Value())
 		checksumPathname = c.base.Append(
 			dirChecksum,
 			checksums,
@@ -1288,8 +1301,11 @@ func (c *Cache) cure(a Artifact) (
 		if checksumPathname == nil {
 			h := sha512.New384()
 			h.Write(data)
-			h.Sum(checksum[:0])
-			checksums = Encode(checksum)
+			buf := c.getIdentBuf()
+			h.Sum(buf[:0])
+			checksum = unique.Make(Checksum(buf[:]))
+			checksums = Encode(Checksum(buf[:]))
+			c.putIdentBuf(buf)
 			checksumPathname = c.base.Append(
 				dirChecksum,
 				checksums,
@@ -1297,10 +1313,10 @@ func (c *Cache) cure(a Artifact) (
 		} else if c.IsStrict() {
 			h := sha512.New384()
 			h.Write(data)
-			if got := Checksum(h.Sum(nil)); got != checksum {
+			if got := Checksum(h.Sum(nil)); got != checksum.Value() {
 				err = &ChecksumMismatchError{
 					Got:  got,
-					Want: checksum,
+					Want: checksum.Value(),
 				}
 				return
 			}
@@ -1355,11 +1371,11 @@ func (c *Cache) cure(a Artifact) (
 
 	case FloodArtifact:
 		deps := a.Dependencies()
-		f := FContext{t, make(map[Artifact]*check.Absolute, len(deps))}
+		f := FContext{t, make(map[Artifact]cureRes, len(deps))}
 
 		var wg sync.WaitGroup
 		wg.Add(len(deps))
-		res := make([]*check.Absolute, len(deps))
+		res := make([]cureRes, len(deps))
 		errs := make(DependencyCureError, 0, len(deps))
 		var errsMu sync.Mutex
 		for i, d := range deps {
@@ -1422,20 +1438,18 @@ func (c *Cache) cure(a Artifact) (
 	}
 
 	if checksumPathname == nil {
-		checksum = gotChecksum
-		checksums = Encode(checksum)
+		checksum = unique.Make(gotChecksum)
+		checksums = Encode(gotChecksum)
 		checksumPathname = c.base.Append(
 			dirChecksum,
 			checksums,
 		)
-	} else {
-		if gotChecksum != checksum {
-			err = &ChecksumMismatchError{
-				Got:  gotChecksum,
-				Want: checksum,
-			}
-			return
+	} else if gotChecksum != checksum.Value() {
+		err = &ChecksumMismatchError{
+			Got:  gotChecksum,
+			Want: checksum.Value(),
 		}
+		return
 	}
 
 	if err = os.Chmod(t.work.String(), 0700); err != nil {
@@ -1470,9 +1484,9 @@ func (c *Cache) cure(a Artifact) (
 func (pending *pendingArtifactDep) cure(c *Cache) {
 	defer pending.Done()
 
-	pathname, _, err := c.cure(pending.a)
+	var err error
+	pending.resP.pathname, pending.resP.checksum, err = c.cure(pending.a)
 	if err == nil {
-		*pending.resP = pathname
 		return
 	}
 
@@ -1536,7 +1550,7 @@ func open(
 		msg:  msg,
 		base: base,
 
-		ident:        make(map[unique.Handle[ID]]Checksum),
+		ident:        make(map[unique.Handle[ID]]unique.Handle[Checksum]),
 		identErr:     make(map[unique.Handle[ID]]error),
 		identPending: make(map[unique.Handle[ID]]<-chan struct{}),
 	}