fix: combine outputs when using bit -D deps

alecthomas · alecthomas · commit 1dcaf8fff833 · 2023-12-06T17:42:27.000+11:00
Previously each output had a complete copy of identical inputs for the
sam target.
diff --git a/cmd/bit/main.go b/cmd/bit/main.go
@@ -8,6 +8,7 @@ import (
 	"path/filepath"
 	"runtime"
 	"runtime/pprof"
+	"sort"
 	"strings"
 
 	"github.com/alecthomas/kong"
@@ -81,6 +82,7 @@ func main() {
 
 	case cli.Describe == "deps":
 		deps := eng.Deps()
+		deps = mergeDepsWithCommonInputs(deps)
 		for in, deps := range deps {
 			w := len(in) + 1
 			fmt.Printf("%s:", in)
@@ -93,6 +95,7 @@ func main() {
 				fmt.Printf(" %s", dep)
 			}
 			fmt.Println()
+			fmt.Println()
 		}
 
 	case cli.Describe == "files":
@@ -153,3 +156,18 @@ func reportError(file *os.File, logger *logging.Logger, err error) {
 	}
 	os.Exit(1)
 }
+func mergeDepsWithCommonInputs(deps map[string][]string) map[string][]string {
+	inputHashes := map[engine.Hasher][]string{}
+	for output, inputs := range deps {
+		sort.Strings(inputs)
+		hasher := engine.NewHasher()
+		engine.HashSlice(&hasher, inputs)
+		inputHashes[hasher] = append(inputHashes[hasher], output)
+	}
+	merged := map[string][]string{}
+	for _, outputs := range inputHashes {
+		key := strings.Join(outputs, " \\\n")
+		merged[key] = deps[outputs[0]]
+	}
+	return merged
+}
diff --git a/engine/engine.go b/engine/engine.go
@@ -542,7 +542,7 @@ func (e *Engine) dbRefHasher(target *Target, ref *parser.Ref) (Hasher, error) {
 // Hash real files.
 func (e *Engine) realRefHasher(target *Target, ref *parser.Ref) (Hasher, error) {
 	h := NewHasher()
-	h.string(ref.Text)
+	h.Str(ref.Text)
 
 	// If we have a hash function, use that for every reference.
 	if target.hashFunc != nil {
diff --git a/engine/hasher.go b/engine/hasher.go
@@ -10,9 +10,33 @@ const prime64 = 1099511628211
 
 type Hasher uint64
 
+// HashSlice hashes a slice of strings, uint64s, byte slices or Hashers.
+func HashSlice[T string | uint64 | []byte | Hasher](hasher *Hasher, slice []T) {
+	switch slice := any(slice).(type) {
+	case []string:
+		for _, s := range slice {
+			hasher.Str(s)
+		}
+	case []uint64:
+		for _, s := range slice {
+			hasher.Int(s)
+		}
+	case [][]byte:
+		for _, s := range slice {
+			hasher.Bytes(s)
+		}
+	case []Hasher:
+		for _, s := range slice {
+			hasher.Update(s)
+		}
+	default:
+		panic(fmt.Sprintf("unexpected type %T", slice))
+	}
+}
+
 func NewHasher() Hasher { return offset64 }
 
-// Update the hash with a uint64.
+// Int updates the hash with a uint64.
 func (h *Hasher) Int(data uint64) {
 	f := *h
 	f ^= Hasher(data)
@@ -28,8 +52,8 @@ func (h *Hasher) Update(other Hasher) {
 	*h = f
 }
 
-// Update the hash with a string.
-func (h *Hasher) string(data string) {
+// Str updates the hash with a Str.
+func (h *Hasher) Str(data string) {
 	f := *h
 	for _, c := range data {
 		f ^= Hasher(c)
@@ -38,7 +62,7 @@ func (h *Hasher) string(data string) {
 	*h = f
 }
 
-// Update the hash with a byte slice.
+// Bytes updates the hash with a byte slice.
 func (h *Hasher) Bytes(data []byte) {
 	f := *h
 	for _, c := range data {
