无法使用Spoon等工具时，如何用Python执行Pentaho Kettle的.ktr文件？

Great question! When you can't use Spoon or other native Kettle tools, executing a .ktr transformation file via Python is totally feasible—here are two reliable approaches I’ve used in real-world projects:

This is the simplest approach, requiring no extra Python libraries beyond the standard library. It leverages Kettle's built-in command-line executor pan (designed specifically for transformations; use kitchen for jobs).

步骤： #

1. 确认Kettle安装路径: Locate the pan executable in your Kettle installation:
   • Windows: <KETTLE_INSTALL_DIR>\data-integration\pan.bat
   • Linux/macOS: <KETTLE_INSTALL_DIR>/data-integration/pan.sh
2. 用Python的subprocess执行命令:

import subprocess
import os

# 配置路径（根据你的系统修改）
KETTLE_PAN_PATH = r"C:\pentaho\data-integration\pan.bat"  # Windows示例
# KETTLE_PAN_PATH = "/opt/pentaho/data-integration/pan.sh"  # Linux/macOS示例
KTR_FILE_PATH = r"C:\projects\my_transform.ktr"
LOG_FILE_PATH = r"C:\logs\transform_execution.log"

# 构建命令参数
cmd_args = [
    KETTLE_PAN_PATH,
    "-file", KTR_FILE_PATH,
    "-log", LOG_FILE_PATH,
    # 可选：给转换传递参数
    # "-param:DB_HOST=localhost",
    # "-param:DB_USER=my_user"
]

try:
    # 执行命令并捕获输出
    execution_result = subprocess.run(
        cmd_args,
        check=True,
        stdout=subprocess.PIPE,
        stderr=subprocess.PIPE,
        text=True
    )
    print("转换执行成功！")
    print("标准输出:\n", execution_result.stdout)
except subprocess.CalledProcessError as e:
    print(f"转换执行失败，错误码: {e.returncode}")
    print("错误输出:\n", e.stderr)

常用参数说明： #

• -file: 指定要执行的.ktr文件路径
• -log: 指定执行日志的保存路径
• -param:KEY=VALUE: 给转换传递参数（要和你KTR中定义的参数名匹配）
• -level: 设置日志详细程度（比如Basic、Detailed、Debug）

如果需要更精细的控制（比如动态修改转换配置、实时监控执行进度、深度整合Python工作流），可以用JPype直接调用Kettle底层的Java API。

步骤： #

1. 安装JPype: pip install JPype1
2. 确保Java版本兼容: Kettle对Java版本有要求（比如Kettle 9.x适配Java 8或11）
3. 找到Kettle的依赖jar包: 所有jar都在<KETTLE_INSTALL_DIR>/data-integration/lib目录下

import jpype
import jpype.imports
from jpype import JClass
import os

# 配置路径
KETTLE_LIB_DIR = r"C:\pentaho\data-integration\lib"  # Windows示例
# KETTLE_LIB_DIR = "/opt/pentaho/data-integration/lib"  # Linux/macOS示例
KTR_FILE_PATH = r"C:\projects\my_transform.ktr"

# 构建classpath，包含所有Kettle的jar包
classpath_entries = [os.path.join(KETTLE_LIB_DIR, jar) for jar in os.listdir(KETTLE_LIB_DIR) if jar.endswith(".jar")]

# 启动JVM并加载Kettle的类路径
jpype.startJVM(classpath=classpath_entries)

try:
    # 导入Kettle核心类
    KettleEnvironment = JClass("org.pentaho.di.core.KettleEnvironment")
    TransMeta = JClass("org.pentaho.di.trans.TransMeta")
    Trans = JClass("org.pentaho.di.trans.Trans")

    # 初始化Kettle环境
    KettleEnvironment.init()

    # 加载转换元数据
    trans_metadata = TransMeta(KTR_FILE_PATH)

    # 创建转换实例
    transformation = Trans(trans_metadata)

    # 可选：动态设置参数
    transformation.setParameterValue("DB_PASSWORD", "secure_password_123")

    # 执行转换
    transformation.execute([])
    # 等待转换执行完成
    transformation.waitUntilFinished()

    # 检查执行状态
    if transformation.getErrors() == 0:
        print("转换执行成功！")
        # 可选：获取完整日志内容
        # print("执行日志:\n", transformation.getLogText())
    else:
        print(f"转换执行失败，共出现 {transformation.getErrors()} 个错误。")

finally:
    # 关闭JVM释放资源
    jpype.shutdownJVM()

注意事项： #

• 务必确保Java版本和Kettle版本兼容（旧版Kettle可能只支持Java 8）
• 要把lib目录下所有jar都加入classpath，缺失jar会导致运行时错误
• 可以通过Kettle API获取更详细的执行指标和日志（比如transformation.getStatus()、transformation.getLogChannelId()）

内容的提问来源于stack exchange，提问作者Rishabh K Sharma