internal/pkg: garbage collection

This destroys cache entries not referred to by user-specified artifacts and optionally their inputs.

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

1 files changed, 138 insertions, 0 deletions

diff --git a/internal/pkg/clean.go b/internal/pkg/clean.go
new file mode 100644
index 00000000..addcd32d
--- /dev/null
+++ b/internal/pkg/clean.go
@@ -0,0 +1,138 @@
+package pkg
+
+import (
+	"errors"
+	"os"
+	"unique"
+)
+
+// Clean destroys checksum backing entries without any identifier or substitute
+// entry referring to it. If at least one keep [Artifact] is specified,
+// identifier and substitute entries not kept alive by them are destroyed first.
+func (c *Cache) Clean(dry, inputs bool, keep ...Artifact) (
+	[]unique.Handle[ID],
+	[]unique.Handle[Checksum],
+	error,
+) {
+	c.identMu.Lock()
+	defer c.identMu.Unlock()
+
+	c.checksumMu.Lock()
+	defer c.checksumMu.Unlock()
+
+	dents, err := os.ReadDir(c.base.Append(dirChecksum).String())
+	if err != nil {
+		return nil, nil, err
+	}
+	checksums := make(map[unique.Handle[Checksum]]string, len(dents))
+	var buf Checksum
+	for _, dent := range dents {
+		name := dent.Name()
+		if err = Decode(&buf, name); err != nil {
+			return nil, nil, err
+		}
+		checksums[unique.Make(buf)] = name
+	}
+
+	type identPair struct {
+		id   unique.Handle[ID]
+		name string
+	}
+	dents, err = os.ReadDir(c.base.Append(dirIdentifier).String())
+	if err != nil {
+		return nil, nil, err
+	}
+
+	keepIdents := make(map[unique.Handle[ID]]struct{})
+	if inputs {
+		for _, id := range Inputs((*Collect)(&keep)) {
+			keepIdents[id] = struct{}{}
+		}
+	} else {
+		for _, a := range keep {
+			keepIdents[c.Ident(a)] = struct{}{}
+		}
+	}
+
+	idents := make([]identPair, 0, len(dents))
+	for _, dent := range dents {
+		name := dent.Name()
+		if err = Decode(&buf, name); err != nil {
+			return nil, nil, err
+		}
+		id := unique.Make(ID(buf))
+
+		if _, ok := keepIdents[id]; len(keep) == 0 || ok {
+			if err = readlinkChecksum(c.base.Append(
+				dirIdentifier,
+				name,
+			), &buf); err != nil {
+				return nil, nil, err
+			}
+			delete(checksums, unique.Make(buf))
+			continue
+		}
+
+		c.msg.Verbosef("arranging for destruction of %s...", name)
+		idents = append(idents, identPair{id, name})
+	}
+
+	destroyedIdents := make([]unique.Handle[ID], 0, len(idents))
+	for _, pair := range idents {
+		if !dry {
+			if err = os.Remove(c.base.Append(
+				dirIdentifier,
+				pair.name,
+			).String()); err != nil {
+				return destroyedIdents, nil, err
+			}
+		}
+		destroyedIdents = append(destroyedIdents, pair.id)
+	}
+
+	destroyedChecksums := make([]unique.Handle[Checksum], 0, len(checksums))
+	for checksum, name := range checksums {
+		if err = c.parent.Err(); err != nil {
+			return destroyedIdents, destroyedChecksums, err
+		}
+		c.msg.Verbosef("destroying checksum %s...", name)
+		if !dry {
+			if err = errors.Join(removeAll(c.base.Append(
+				dirChecksum,
+				name,
+			))); err != nil {
+				return destroyedIdents, destroyedChecksums, err
+			}
+		}
+		destroyedChecksums = append(destroyedChecksums, checksum)
+	}
+
+	dents, err = os.ReadDir(c.base.Append(dirSubstitute).String())
+	if err != nil {
+		return destroyedIdents, destroyedChecksums, err
+	}
+	for _, dent := range dents {
+		name := dent.Name()
+		if err = readlinkChecksum(c.base.Append(
+			dirSubstitute,
+			name,
+		), &buf); err != nil {
+			return destroyedIdents, destroyedChecksums, err
+		}
+		if _, ok := checksums[unique.Make(buf)]; !ok {
+			continue
+		}
+
+		c.msg.Verbosef("destroying substitute %s...", name)
+		if !dry {
+			if err = os.Remove(c.base.Append(
+				dirSubstitute,
+				name,
+			).String()); err != nil {
+				return destroyedIdents, destroyedChecksums, err
+			}
+		}
+	}
+
+	return destroyedIdents, destroyedChecksums, nil
+}