Manual Backport to branch(3.12) : Add data loader core key and column utils (#2396)

Co-authored-by: Peckstadt <[email protected]>

Author: inv-jishnu

core/src/main/java/com/scalar/db/common/error/CoreError.java

@@ -606,6 +606,25 @@ public enum CoreError implements ScalarDbError {
       "",
       ""),
 
+  DATA_LOADER_INVALID_COLUMN_NON_EXISTENT(
+      Category.USER_ERROR,
+      "0148",
+      "Invalid key: Column %s does not exist in the table %s in namespace %s.",
+      "",
+      ""),
+  DATA_LOADER_INVALID_BASE64_ENCODING_FOR_COLUMN_VALUE(
+      Category.USER_ERROR,
+      "0149",
+      "Invalid base64 encoding for blob value for column %s in table %s in namespace %s",
+      "",
+      ""),
+  DATA_LOADER_INVALID_NUMBER_FORMAT_FOR_COLUMN_VALUE(
+      Category.USER_ERROR,
+      "0150",
+      "Invalid number specified for column %s in table %s in namespace %s",
+      "",
+      ""),
+
   //
   // Errors for the concurrency error category
   //

data-loader/core/build.gradle

@@ -9,7 +9,7 @@ archivesBaseName = "scalardb-data-loader-core"
 dependencies {
     // ScalarDB core
     implementation project(':core')
-    
+
     // for SpotBugs
     compileOnly "com.github.spotbugs:spotbugs-annotations:${spotbugsVersion}"
     testCompileOnly "com.github.spotbugs:spotbugs-annotations:${spotbugsVersion}"

data-loader/core/src/main/java/com/scalar/db/dataloader/core/ColumnInfo.java

@@ -0,0 +1,24 @@
+package com.scalar.db.dataloader.core;
+
+import lombok.Builder;
+import lombok.Value;
+
+/**
+ * Represents a column in a database table.
+ *
+ * <p>This class holds the metadata for a column, including the namespace (schema), table name, and
+ * the column name within the table.
+ */
+@Value
+@Builder
+public class ColumnInfo {
+
+  /** The namespace (schema) where the table is located. */
+  String namespace;
+
+  /** The name of the table where the column resides. */
+  String tableName;
+
+  /** The name of the column in the table. */
+  String columnName;
+}

data-loader/core/src/main/java/com/scalar/db/dataloader/core/exception/ColumnParsingException.java

@@ -0,0 +1,30 @@
+package com.scalar.db.dataloader.core.exception;
+
+/**
+ * An exception that is thrown when an error occurs while trying to create a ScalarDB column from a
+ * value.
+ *
+ * <p>This exception is typically used to indicate a problem with parsing or converting data into a
+ * format that can be used to create a column in ScalarDB.
+ */
+public class ColumnParsingException extends Exception {
+
+  /**
+   * Constructs a new {@code ColumnParsingException} with the specified detail message.
+   *
+   * @param message the detail message explaining the cause of the exception
+   */
+  public ColumnParsingException(String message) {
+    super(message);
+  }
+
+  /**
+   * Constructs a new {@code ColumnParsingException} with the specified detail message and cause.
+   *
+   * @param message the detail message explaining the cause of the exception
+   * @param cause the cause of the exception (can be {@code null})
+   */
+  public ColumnParsingException(String message, Throwable cause) {
+    super(message, cause);
+  }
+}

data-loader/core/src/main/java/com/scalar/db/dataloader/core/exception/KeyParsingException.java

@@ -0,0 +1,30 @@
+package com.scalar.db.dataloader.core.exception;
+
+/**
+ * An exception that is thrown when an error occurs while trying to create a ScalarDB key from a
+ * value.
+ *
+ * <p>This exception is typically used to indicate a problem with parsing or converting data into a
+ * format that can be used to create a key in ScalarDB.
+ */
+public class KeyParsingException extends Exception {
+
+  /**
+   * Constructs a new {@code KeyParsingException} with the specified detail message.
+   *
+   * @param message the detail message explaining the cause of the exception
+   */
+  public KeyParsingException(String message) {
+    super(message);
+  }
+
+  /**
+   * Constructs a new {@code KeyParsingException} with the specified detail message and cause.
+   *
+   * @param message the detail message explaining the cause of the exception
+   * @param cause the cause of the exception (can be {@code null})
+   */
+  public KeyParsingException(String message, Throwable cause) {
+    super(message, cause);
+  }
+}

data-loader/core/src/main/java/com/scalar/db/dataloader/core/util/ColumnUtils.java

@@ -0,0 +1,91 @@
+package com.scalar.db.dataloader.core.util;
+
+import com.scalar.db.common.error.CoreError;
+import com.scalar.db.dataloader.core.ColumnInfo;
+import com.scalar.db.dataloader.core.exception.ColumnParsingException;
+import com.scalar.db.io.BigIntColumn;
+import com.scalar.db.io.BlobColumn;
+import com.scalar.db.io.BooleanColumn;
+import com.scalar.db.io.Column;
+import com.scalar.db.io.DataType;
+import com.scalar.db.io.DoubleColumn;
+import com.scalar.db.io.FloatColumn;
+import com.scalar.db.io.IntColumn;
+import com.scalar.db.io.TextColumn;
+import java.util.Base64;
+import javax.annotation.Nullable;
+
+/**
+ * Utility class for creating and managing ScalarDB columns.
+ *
+ * <p>This class provides methods for creating ScalarDB columns based on the given data type, column
+ * information, and value. It includes handling for various data types and special cases like base64
+ * encoding for BLOB data.
+ */
+public final class ColumnUtils {
+
+  /** Restrict instantiation via private constructor */
+  private ColumnUtils() {}
+
+  /**
+   * Creates a ScalarDB column from the given data type, column information, and value.
+   *
+   * <p>Blob source values need to be base64 encoded before passing them as a value. If the value is
+   * {@code null}, the corresponding column is created as a {@code null} column.
+   *
+   * @param dataType the data type of the specified column
+   * @param columnInfo the ScalarDB table column information
+   * @param value the value for the ScalarDB column (may be {@code null})
+   * @return the ScalarDB column created from the specified data
+   * @throws ColumnParsingException if an error occurs while creating the column or parsing the
+   *     value
+   */
+  public static Column<?> createColumnFromValue(
+      DataType dataType, ColumnInfo columnInfo, @Nullable String value)
+      throws ColumnParsingException {
+    String columnName = columnInfo.getColumnName();
+    try {
+      switch (dataType) {
+        case BOOLEAN:
+          return value != null
+              ? BooleanColumn.of(columnName, Boolean.parseBoolean(value))
+              : BooleanColumn.ofNull(columnName);
+        case INT:
+          return value != null
+              ? IntColumn.of(columnName, Integer.parseInt(value))
+              : IntColumn.ofNull(columnName);
+        case BIGINT:
+          return value != null
+              ? BigIntColumn.of(columnName, Long.parseLong(value))
+              : BigIntColumn.ofNull(columnName);
+        case FLOAT:
+          return value != null
+              ? FloatColumn.of(columnName, Float.parseFloat(value))
+              : FloatColumn.ofNull(columnName);
+        case DOUBLE:
+          return value != null
+              ? DoubleColumn.of(columnName, Double.parseDouble(value))
+              : DoubleColumn.ofNull(columnName);
+        case TEXT:
+          return value != null ? TextColumn.of(columnName, value) : TextColumn.ofNull(columnName);
+        case BLOB:
+          // Source blob values need to be base64 encoded
+          return value != null
+              ? BlobColumn.of(columnName, Base64.getDecoder().decode(value))
+              : BlobColumn.ofNull(columnName);
+        default:
+          throw new AssertionError();
+      }
+    } catch (NumberFormatException e) {
+      throw new ColumnParsingException(
+          CoreError.DATA_LOADER_INVALID_NUMBER_FORMAT_FOR_COLUMN_VALUE.buildMessage(
+              columnName, columnInfo.getTableName(), columnInfo.getNamespace()),
+          e);
+    } catch (IllegalArgumentException e) {
+      throw new ColumnParsingException(
+          CoreError.DATA_LOADER_INVALID_BASE64_ENCODING_FOR_COLUMN_VALUE.buildMessage(
+              columnName, columnInfo.getTableName(), columnInfo.getNamespace()),
+          e);
+    }
+  }
+}

data-loader/core/src/main/java/com/scalar/db/dataloader/core/util/KeyUtils.java

@@ -0,0 +1,88 @@
+package com.scalar.db.dataloader.core.util;
+
+import com.scalar.db.api.TableMetadata;
+import com.scalar.db.common.error.CoreError;
+import com.scalar.db.dataloader.core.ColumnInfo;
+import com.scalar.db.dataloader.core.ColumnKeyValue;
+import com.scalar.db.dataloader.core.exception.ColumnParsingException;
+import com.scalar.db.dataloader.core.exception.KeyParsingException;
+import com.scalar.db.io.Column;
+import com.scalar.db.io.DataType;
+import com.scalar.db.io.Key;
+import javax.annotation.Nullable;
+
+/**
+ * Utility class for creating and managing ScalarDB keys.
+ *
+ * <p>This class provides methods to parse key-value pairs and create ScalarDB key instances. It
+ * also includes utility methods for handling data types, columns, and potential parsing exceptions.
+ */
+public final class KeyUtils {
+
+  /** Restrict instantiation via private constructor */
+  private KeyUtils() {}
+
+  /**
+   * Converts a key-value pair, in the format of <key>=<value>, into a ScalarDB Key instance for a
+   * specific ScalarDB table.
+   *
+   * <p>This method uses the provided table metadata to determine the data type for the key and
+   * creates a corresponding ScalarDB Key. If the key does not match any column in the table
+   * metadata, a {@link KeyParsingException} is thrown.
+   *
+   * @param columnKeyValue a key-value pair in the format of <key>=<value>
+   * @param namespace the name of the ScalarDB namespace
+   * @param tableName the name of the ScalarDB table
+   * @param tableMetadata metadata for the ScalarDB table
+   * @return a new ScalarDB Key instance formatted according to the data type
+   * @throws KeyParsingException if there is an error parsing the key value or if the column does
+   *     not exist
+   */
+  @Nullable
+  public static Key parseKeyValue(
+      @Nullable ColumnKeyValue columnKeyValue,
+      String namespace,
+      String tableName,
+      TableMetadata tableMetadata)
+      throws KeyParsingException {
+    if (columnKeyValue == null) {
+      return null;
+    }
+    String columnName = columnKeyValue.getColumnName();
+    DataType columnDataType = tableMetadata.getColumnDataType(columnName);
+    if (columnDataType == null) {
+      throw new KeyParsingException(
+          CoreError.DATA_LOADER_INVALID_COLUMN_NON_EXISTENT.buildMessage(
+              columnName, tableName, namespace));
+    }
+    ColumnInfo columnInfo =
+        ColumnInfo.builder()
+            .namespace(namespace)
+            .tableName(tableName)
+            .columnName(columnName)
+            .build();
+    return createKey(columnDataType, columnInfo, columnKeyValue.getColumnValue());
+  }
+
+  /**
+   * Creates a ScalarDB key based on the provided data type, column information, and value.
+   *
+   * <p>This method creates a ScalarDB Key instance by converting the column value to the
+   * appropriate data type and constructing the key using that value.
+   *
+   * @param dataType the data type of the specified column
+   * @param columnInfo the ScalarDB table column information
+   * @param value the value for the ScalarDB key
+   * @return a ScalarDB Key instance
+   * @throws KeyParsingException if there is an error while creating the ScalarDB key
+   */
+  public static Key createKey(DataType dataType, ColumnInfo columnInfo, String value)
+      throws KeyParsingException {
+    try {
+      Column<?> keyValue = ColumnUtils.createColumnFromValue(dataType, columnInfo, value);
+      return Key.newBuilder().add(keyValue).build();
+    } catch (ColumnParsingException e) {
+      throw new KeyParsingException(e.getMessage(), e);
+    }
+  }
+}