Skip to main content

Capture video stream, save as screenshots.

Installation

You can install from pypi

pip install -U capture_stream

Usage

As a Module

>>> import capture_stream
>>> help(capture_stream)

Help on package capture_stream:

NAME
    capture_stream - # encoding: utf-8

PACKAGE CONTENTS
    __main__

FUNCTIONS
    cv_capture(url: str, start: int = 0, stop: None | int = None, step: int = 1, prefix: str = 'frame_', dir_to_save: str = '')
        从 url 读取视频流,保存图片到本地,jpg 图片格式
        
        :param url: 视频链接
        :param start: 开始索引(从 0 开始编号)
        :param stop:  结束索引(不含此帧),如果 stop 为 None 则到视频结束
        :param step:  索引递增步长
        :param prefix: 文件名会被保存为 f"{prefix}{frame_number}.jpg",frame_number 是帧的索引序号
        :param dir_to_save: 把生成的 jpg 保存到此目录,默认是当前工作目录
        
        从第 start 帧开始,到第 stop 帧结束,以 step 帧递增
            - 如果 step <= 0,只截取 start 那一帧
            - 如果 stop 为 None,则从 start 到视频结束
            - 否则,跳过此视频
    
    cv_capture_iter(url: str, start: int = 0, stop: None | int = None, step: int = 1) -> collections.abc.Iterator[tuple[int, numpy.ndarray]]
        从 url 读取视频流,返回帧的迭代器
        
        :param url: 视频链接
        :param start: 开始索引(从 0 开始编号)
        :param stop:  结束索引(不含此帧),如果 stop 为 None 则到视频结束
        :param step:  索引递增步长
        
        :return: (帧索引序号, 图片数据) 的元组的迭代器
        
        从第 start 帧开始,到第 stop 帧结束,以 step 帧递增
            - 如果 step <= 0,只截取 start 那一帧
            - 如果 stop 为 None,则从 start 到视频结束
            - 否则,跳过此视频

DATA
    __all__ = ['cv_capture_iter', 'cv_capture']

VERSION
    (0, 0, 1)

AUTHOR
    ChenyangGao <https://chenyanggao.github.io>

FILE
    /path/to/capture_stream/__init__.py

As a Command Line

You can use this module to capture screenshots for video steam.

$ capture_stream -h
usage: capture_stream [-h] [-s START] [-t STOP] [-st STEP] [-p PREFIX] [-d DIR_TO_SAVE] [-v] url

从 url 读取视频流,为某些帧截图

positional arguments:
  url                   视频的 url

options:
  -h, --help            show this help message and exit
  -s START, --start START
                        开始帧的索引,从 0 开始编号,默认值:0
  -t STOP, --stop STOP  结束帧的索引(不含),默认为到视频结束
  -st STEP, --step STEP
                        帧的索引递增步长,如果小于等于 0 则只截取 start 那一帧,默认值:1
  -p PREFIX, --prefix PREFIX
                        生成图片的名称的前缀,文件名会被保存为 f'{prefix}{frame_number}.jpg',frame_number 是帧的索引序号,默认无前缀
  -d DIR_TO_SAVE, --dir-to-save DIR_TO_SAVE
                        把生成的 jpg 保存到此目录,默认值:当前工作目录
  -v, --version         输出版本号

Metadata

Release files for capture_stream 0.0.1.5

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for capture_stream 0.0.1.5
File Size Uploaded
capture_stream-0.0.1.5.tar.gz 4.4 kB Details

Built distribution (wheel)

Table of built distributions (wheels) for capture_stream 0.0.1.5
File Interpreter ABI Platform
capture_stream-0.0.1.5-py3-none-any.whl Python 3 none any Details

Total release size: 11.4 kB

Release files / capture_stream-0.0.1.5.tar.gz

Download URL capture_stream-0.0.1.5.tar.gz
Size 4.4 kB
Tags Source
SHA-256 checksum
How to use checksums
3d25b93c989e33bce12129e3452aa85ddc1d8d5c029854dabaef69dfa205a1ce
BLAKE2b-256 checksum
How to use checksums
1d50b9eca22533096c2361a9e65af0bde780be92c4e1433d8f6ae4d30cbadf77
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via poetry/1.7.0 CPython/3.11.8 Darwin/23.5.0

Release files / capture_stream-0.0.1.5-py3-none-any.whl

Download URL capture_stream-0.0.1.5-py3-none-any.whl
Size 7.0 kB
Tags Python 3
SHA-256 checksum
How to use checksums
56d2808f4615d1ea280125c93a86c30b950931b91d04bfd051dd194ae954b9da
BLAKE2b-256 checksum
How to use checksums
500610e4754697fd887c9a4335d1b9b583ed6d6a76881d439f2632dcb3dc852a
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via poetry/1.7.0 CPython/3.11.8 Darwin/23.5.0

Release history Release notifications | RSS feed

This release

0.0.1.5 This release

2 release files

0.0.1

2 release files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page