v3 candidate

internal/leviathan/analysis/hash.go

@@ -0,0 +1,63 @@
+package analysis
+
+import (
+	"crypto/md5"
+	"crypto/sha1"
+	"crypto/sha256"
+	"crypto/sha512"
+	"encoding/hex"
+	"hash"
+	"hash/crc32"
+
+	"github.com/andydixon/nexus/internal/leviathan/document"
+)
+
+// HashSet contains CRC32/MD5/SHA-1/SHA-256/SHA-512 for a range.
+type HashSet struct {
+	Length int64  `json:"length"`
+	CRC32  string `json:"crc32"`
+	MD5    string `json:"md5"`
+	SHA1   string `json:"sha1"`
+	SHA256 string `json:"sha256"`
+	SHA512 string `json:"sha512"`
+}
+
+// Hashes computes all supported hashes over the document range.
+// The computation runs chunk-by-chunk so large files do not force a
+// full-buffer duplication.
+func Hashes(d *document.Document, start, length int64) (map[string]any, error) {
+	if length < 0 {
+		length = d.Size() - start
+	}
+
+	crc := crc32.NewIEEE()
+	mh := md5.New()
+	s1 := sha1.New()
+	s256 := sha256.New()
+	s512 := sha512.New()
+
+	writers := []hash.Hash{mh, s1, s256, s512}
+
+	total := int64(0)
+	err := d.ForEachChunk(start, length, func(_ int64, chunk []byte) error {
+		crc.Write(chunk)
+		for _, w := range writers {
+			w.Write(chunk)
+		}
+		total += int64(len(chunk))
+		return nil
+	})
+	if err != nil {
+		return nil, err
+	}
+	return map[string]any{
+		"length": total,
+		"crc32":  toHex(crc.Sum(nil)),
+		"md5":    toHex(mh.Sum(nil)),
+		"sha1":   toHex(s1.Sum(nil)),
+		"sha256": toHex(s256.Sum(nil)),
+		"sha512": toHex(s512.Sum(nil)),
+	}, nil
+}
+
+func toHex(b []byte) string { return hex.EncodeToString(b) }