描述 #

视频格式转换处理器，将各种视频格式转换为MP4

核心功能 #

• 支持多种视频格式转换为MP4
• 自动选择第一个音轨
• 视频质量和编码参数自定义
• 支持视频高度限制和缩放
• 音频编码参数精细控制

格式支持 #

• 输入：AVI、MOV、MKV、FLV、WMV、3GP等主流视频格式
• 输出：MP4 (.mp4)
• 视频编解码器：H.264/H.265等
• 音频编解码器：AAC、MP3等

算子参数 #

输入 #

输出 #

包含转换结果路径的数组，成功返回输出路径，失败返回None

参数 #

如参数没有默认值，则为必填参数

调用示例 #

下面的代码展示了如何使用 Daft（适用于分布式）运行算子对视频进行MP4格式转换。

from __future__ import annotations

import os

import daft
from daft import col
from daft.las.functions.udf import las_udf
from daft.las.functions.video import VideoConvertToMp4

if __name__ == "__main__":
    # 转换后的视频会保存到指定的TOS路径下，因此，需要设置好环境变量以保证有权限写入TOS，包括：ACCESS_KEY,SECRET_KEY,TOS_ENDPOINT,TOS_REGION,TOS_TEST_DIR
    TOS_TEST_DIR_URL = os.getenv("TOS_TEST_DIR_URL", "las-cn-beijing-public-online.tos-cn-beijing.volces.com")
    TOS_TEST_DIR = os.getenv("TOS_TEST_DIR", "tos_bucket")

    if os.getenv("DAFT_RUNNER", "native") == "ray":
        import logging

        import ray

        def configure_logging():
            logging.basicConfig(
                level=logging.INFO,
                format="%(asctime)s - %(name)s - %(levelname)s - %(message)s",
                datefmt="%Y-%m-%d %H:%M:%S",
            )
            logging.getLogger("tracing.span").setLevel(logging.WARNING)
            logging.getLogger("daft_io.stats").setLevel(logging.WARNING)
            logging.getLogger("DaftStatisticsManager").setLevel(logging.WARNING)
            logging.getLogger("DaftFlotillaScheduler").setLevel(logging.WARNING)
            logging.getLogger("DaftFlotillaDispatcher").setLevel(logging.WARNING)

        ray.init(dashboard_host="0.0.0.0", runtime_env={"worker_process_setup_hook": configure_logging})
        daft.context.set_runner_ray()

    daft.set_execution_config(actor_udf_ready_timeout=600)
    daft.set_execution_config(min_cpu_per_task=0)

    samples = {
        "input_path": [f"https://{TOS_TEST_DIR_URL}/public/archive/video_convert_to_mp4/sample.mp4"],
        "output_path": [f"tos://{TOS_TEST_DIR}/video_convert_to_mp4/sample_converted.mp4"],
    }
    ds = daft.from_pydict(samples)

    # Using Daft to convert video to MP4 format
    constructor_kwargs = {
        "video_codec": "libx264",
        "crf": 23,
        "preset": "medium",
        "max_height": 240,
        "audio_codec": "aac",
        "audio_bitrate": "192k",
        "select_audio": "auto",
        "extra_params": ["-movflags", "+faststart"],
    }

    ds = ds.with_column(
        "convert_result",
        las_udf(
            VideoConvertToMp4,
            construct_args=constructor_kwargs,
            num_cpus=1,
            concurrency=1,
            batch_size=1,
        )(col("input_path"), col("output_path")),
    )

    ds.show()
    # ╭────────────────────────────────┬────────────────────────────────┬────────────────────────────────╮
    # │ input_path                     ┆ output_path                    ┆ convert_result                 │
    # │ ---                            ┆ ---                            ┆ ---                            │
    # │ String                         ┆ String                         ┆ String                         │
    # ╞════════════════════════════════╪════════════════════════════════╪════════════════════════════════╡
    # │ https://las-public-data-qa.to… ┆ tos://tos_bucket/video_conver… ┆ tos://tos_bucket/video_conver… │
    # ╰────────────────────────────────┴────────────────────────────────┴────────────────────────────────╯