refactor

5 files changed, 194 insertions, 0 deletions

diff --git a/src/main/java/com/keuin/kbackupfabric/backup/incremental/identifier/FileIdentifierProvider.java b/src/main/java/com/keuin/kbackupfabric/backup/incremental/identifier/FileIdentifierProvider.java
new file mode 100644
index 0000000..e1039d6
--- /dev/null
+++ b/src/main/java/com/keuin/kbackupfabric/backup/incremental/identifier/FileIdentifierProvider.java
@@ -0,0 +1,15 @@
+package com.keuin.kbackupfabric.backup.incremental.identifier;
+
+import java.io.File;
+import java.io.IOException;
+
+public interface FileIdentifierProvider<T extends ObjectIdentifier> {
+    /**
+     * Generate file identifier from a random file. The file is not necessarily in the object base.
+     *
+     * @param file the file.
+     * @return the file identifier.
+     * @throws IOException when an I/O error occurs.
+     */
+    T fromFile(File file) throws IOException;
+}
diff --git a/src/main/java/com/keuin/kbackupfabric/backup/incremental/identifier/ObjectIdentifier.java b/src/main/java/com/keuin/kbackupfabric/backup/incremental/identifier/ObjectIdentifier.java
new file mode 100644
index 0000000..07cd390
--- /dev/null
+++ b/src/main/java/com/keuin/kbackupfabric/backup/incremental/identifier/ObjectIdentifier.java
@@ -0,0 +1,13 @@
+package com.keuin.kbackupfabric.backup.incremental.identifier;
+
+import java.io.Serializable;
+
+/**
+ * The identifier distinguishing files in the object collection.
+ * It should be based on cryptographic hash function in order to prevent possible attacks to the backup system.
+ * All identifiers should be immutable and implement their own equals method.
+ * Immutable.
+ */
+public interface ObjectIdentifier extends Serializable {
+    String getIdentification();
+}
diff --git a/src/main/java/com/keuin/kbackupfabric/backup/incremental/identifier/Sha256Identifier.java b/src/main/java/com/keuin/kbackupfabric/backup/incremental/identifier/Sha256Identifier.java
new file mode 100644
index 0000000..50e6aa4
--- /dev/null
+++ b/src/main/java/com/keuin/kbackupfabric/backup/incremental/identifier/Sha256Identifier.java
@@ -0,0 +1,88 @@
+package com.keuin.kbackupfabric.backup.incremental.identifier;
+
+import com.keuin.kbackupfabric.util.BytesUtil;
+
+import java.io.File;
+import java.io.FileInputStream;
+import java.io.IOException;
+import java.security.MessageDigest;
+import java.security.NoSuchAlgorithmException;
+import java.util.Objects;
+
+/**
+ * Identifier based on sha256.
+ * Immutable.
+ */
+public class Sha256Identifier extends SingleHashIdentifier {
+
+    private static final int SHA256_LENGTH = 32;
+    private static final Sha256Identifier DUMMY = new Sha256Identifier(new byte[SHA256_LENGTH]); // only for using its hash method
+    private static final FileIdentifierProvider<Sha256Identifier> factory = Sha256Identifier::fromFile;
+    private static final String marker = "S2";
+
+    public static Sha256Identifier fromFile(File file) throws IOException {
+        if (!file.isFile()) {
+            throw new IllegalArgumentException("file is not a file");
+        }
+        return new Sha256Identifier(DUMMY.hash(file));
+    }
+
+    /**
+     * Load sha-256 from a named file. Only used in StorageObjectLoader.
+     *
+     * @param fileName the file name.
+     * @return identifier.
+     */
+    static Sha256Identifier fromFileName(String fileName) {
+        if (!fileName.matches(marker + "-[0-9A-Fa-f]{32}"))
+            return null;
+        String hexString = fileName.substring(marker.length() + 1);
+        return new Sha256Identifier(BytesUtil.hexToBytes(hexString));
+    }
+
+    public static FileIdentifierProvider<Sha256Identifier> getFactory() {
+        return factory;
+    }
+
+    protected Sha256Identifier(byte[] hash) {
+        super(hash, marker);
+        Objects.requireNonNull(hash);
+        if (hash.length != SHA256_LENGTH) {
+            throw new IllegalStateException(String.format("SHA256 must be %d bytes", SHA256_LENGTH));
+        }
+    }
+
+    @Override
+    protected byte[] hash(File file) throws IOException {
+        try {
+            MessageDigest digest = MessageDigest.getInstance("SHA-256");
+
+            try (FileInputStream inputStream = new FileInputStream(file)) {
+                // This does not work. I don't know why
+//            FileChannel channel = inputStream.getChannel();
+//            ByteBuffer buffer = ByteBuffer.allocate(128);
+//            int readLength;
+//            while ((readLength = channel.read(buffer)) > 0)
+//                digest.update(buffer);
+
+                // This also works, without warnings
+                byte[] readBuffer = new byte[1024 * 1024];
+                int readLength;
+                while ((readLength = inputStream.read(readBuffer)) > 0)
+                    digest.update(readBuffer, 0, readLength);
+
+                // The below lines also works, but the IDE will complain about the while loop
+//            DigestInputStream digestInputStream = new DigestInputStream(inputStream, digest);
+//            while(digestInputStream.read() > 0)
+//                ;
+
+                return digest.digest();
+            }
+
+        } catch (NoSuchAlgorithmException ignored) {
+            // this shouldn't happen
+            return new byte[SHA256_LENGTH];
+        }
+    }
+
+}
diff --git a/src/main/java/com/keuin/kbackupfabric/backup/incremental/identifier/SingleHashIdentifier.java b/src/main/java/com/keuin/kbackupfabric/backup/incremental/identifier/SingleHashIdentifier.java
new file mode 100644
index 0000000..9fd61c8
--- /dev/null
+++ b/src/main/java/com/keuin/kbackupfabric/backup/incremental/identifier/SingleHashIdentifier.java
@@ -0,0 +1,53 @@
+package com.keuin.kbackupfabric.backup.incremental.identifier;
+
+import com.keuin.kbackupfabric.util.BytesUtil;
+
+import java.io.File;
+import java.io.IOException;
+import java.util.Arrays;
+import java.util.Objects;
+
+/**
+ * A simple identifier based on a single hash function.
+ * Immutable.
+ */
+public abstract class SingleHashIdentifier implements ObjectIdentifier {
+
+    private final byte[] hash;
+    private final String type;
+
+    protected SingleHashIdentifier(byte[] hash, String type) {
+        Objects.requireNonNull(hash);
+        Objects.requireNonNull(type);
+        this.hash = Arrays.copyOf(hash, hash.length);
+        this.type = type;
+    }
+
+    /**
+     * The hash function.
+     *
+     * @param file the file to be hashed.
+     * @return the hash bytes.
+     */
+    protected abstract byte[] hash(File file) throws IOException;
+
+    @Override
+    public String getIdentification() {
+        return type + "-" + BytesUtil.bytesToHex(hash);
+    }
+
+    @Override
+    public boolean equals(Object obj) {
+        if (!(obj instanceof SingleHashIdentifier)) {
+            return false;
+        }
+        return Arrays.equals(hash, ((SingleHashIdentifier) obj).hash);
+    }
+
+    @Override
+    public int hashCode() {
+        int result = Objects.hash(type);
+        result = 31 * result + Arrays.hashCode(hash);
+        return result;
+    }
+}
diff --git a/src/main/java/com/keuin/kbackupfabric/backup/incremental/identifier/StorageObjectLoader.java b/src/main/java/com/keuin/kbackupfabric/backup/incremental/identifier/StorageObjectLoader.java
new file mode 100644
index 0000000..55dd6bd
--- /dev/null
+++ b/src/main/java/com/keuin/kbackupfabric/backup/incremental/identifier/StorageObjectLoader.java
@@ -0,0 +1,25 @@
+package com.keuin.kbackupfabric.backup.incremental.identifier;
+
+import java.io.File;
+import java.util.Objects;
+
+public class StorageObjectLoader {
+    /**
+     * Get identifier from storage file.
+     *
+     * @param file storage file.
+     * @return identifier. If failed, return null.
+     */
+    public static ObjectIdentifier asIdentifier(File file) {
+        Objects.requireNonNull(file);
+        String fileName = file.getName();
+        ObjectIdentifier identifier;
+
+        identifier = Sha256Identifier.fromFileName(fileName);
+        if (identifier != null)
+            return identifier;
+
+        // Add more identifiers.
+        return null;
+    }
+}