srt

|ghactions| |coveralls|

.. |ghactions| image:: https://img.shields.io/github/actions/workflow/status/cdown/srt/ci.yml?branch=develop :target: https://github.com/cdown/srt/actions?query=branch%3Adevelop :alt: Tests

.. |coveralls| image:: https://img.shields.io/coveralls/cdown/srt/develop.svg?label=test%20coverage :target: https://coveralls.io/github/cdown/srt?branch=develop :alt: Coverage

srt is a tiny but featureful Python library for parsing, modifying, and composing SRT files. Take a look at the quickstart for a basic overview of the library. Detailed API documentation_ is also available.

Want to see some examples of its use? Take a look at the tools shipped with the library. This library is also used internally by projects like subsync, NVIDIA RAD-TTS, manim, kinobot, bw_plex, and many more.

.. _subsync: https://github.com/smacke/subsync .. _NVIDIA RAD-TTS: https://github.com/NVIDIA/radtts .. _bw_plex: https://github.com/Hellowlol/bw_plex .. _manim: https://github.com/ManimCommunity/manim .. _kinobot: https://github.com/vitiko98/kinobot

Why choose this library?

• Can parse many broken SRT files which other SRT libraries cannot, and fix them
• Extremely lightweight, ~200 lines of code excluding docstrings
• Simple, intuitive API
• High quality test suite using Hypothesis_
• 100% test coverage_ (including branches)
• Well documented API_, at both a high and low level
• ~30% faster than pysrt on typical workloads_
• Full support for PyPy_
• No dependencies outside of the standard library
• Tolerant of many common errors found in real-world SRT files
• Support for Asian-style SRT formats (ie. "fullwidth" SRT format)
• Completely Unicode compliant
• Released under a highly permissive license (MIT)
• Real world tested — used in production to process thousands of SRT files every day
• Portable — runs on Linux, OSX, and Windows
• Tools included — contains lightweight tools to perform generic tasks with the library

.. _quickstart: http://srt.readthedocs.org/en/latest/quickstart.html .. _SRT files: https://en.wikipedia.org/wiki/SubRip#SubRip_text_file_format .. _Hypothesis: https://github.com/DRMacIver/hypothesis .. _100% test coverage: https://coveralls.io/github/cdown/srt?branch=develop .. _Well documented API: http://srt.readthedocs.org/en/latest/index.html .. _PyPy: http://pypy.org/ .. _~30% faster than pysrt on typical workloads: https://paste.pound-python.org/raw/8nQKbDW0ROWvS7bOeAb3/

Usage

Tools

There are a number of tools shipped with the library_ to manipulate, process, and fix SRT files. Here's an example using hanzidentifier_ to strip out non-Chinese lines:

.. code::

$ cat pe.srt
1
00:00:33,843 --> 00:00:38,097
Only 3% of the water on our planet is fresh.
地球上只有3%的水是淡水

2
00:00:40,641 --> 00:00:44,687
Yet, these precious waters are rich with surprise.
可是这些珍贵的淡水中却充满了惊奇

$ srt lines-matching -m hanzidentifier -f hanzidentifier.has_chinese -i pe.srt
1
00:00:33,843 --> 00:00:38,097
地球上只有3%的水是淡水

2
00:00:40,641 --> 00:00:44,687
可是这些珍贵的淡水中却充满了惊奇

These tools are easy to chain together, for example, say you have one subtitle with Chinese and English, and other with French, but you want Chinese and French only. Oh, and the Chinese one is 5 seconds later than it should be. That's easy enough to sort out:

.. code::

$ srt lines-matching -m hanzidentifier -f hanzidentifier.has_chinese -i chs+eng.srt |

srt fixed-timeshift --seconds -5 |
srt mux --input - --input fra.srt

See the srt_tools/ directory for more information.

.. _hanzidentifier: https://github.com/tsroten/hanzidentifier

Library

Detailed API documentation_ is available, but here are the basics.

Here's how you convert SRT input to Subtitle objects which you can manipulate:

.. code:: python

>>> data = '''\
1
00:00:33,843 --> 00:00:38,097
地球上只有3%的水是淡水

2
00:00:40,641 --> 00:00:44,687
可是这些珍贵的淡水中却充满了惊奇

3
00:00:57,908 --> 00:01:03,414
所有陆地生命归根结底都依赖於淡水

'''
>>> for sub in srt.parse(data):
...     print(sub)
Subtitle(index=1, start=datetime.timedelta(seconds=33, microseconds=843000), end=datetime.timedelta(seconds=38, microseconds=97000), content='地球上只有3%的水是淡水', proprietary='')
Subtitle(index=2, start=datetime.timedelta(seconds=40, microseconds=641000), end=datetime.timedelta(seconds=44, microseconds=687000), content='可是这些珍贵的淡水中却充满了惊奇', proprietary='')
Subtitle(index=3, start=datetime.timedelta(seconds=57, microseconds=908000), end=datetime.timedelta(seconds=63, microseconds=414000), content='所有陆地生命归根结底都依赖於淡水', proprietary='')

And here's how you go back from Subtitle objects to SRT output:

.. code:: python

>>> subs = list(srt.parse(data))
>>> subs[1].content = "Changing subtitle data is easy!"
>>> print(srt.compose(subs))
1
00:00:33,843 --> 00:00:38,097
地球上只有3%的水是淡水

2
00:00:40,641 --> 00:00:44,687
Changing subtitle data is easy!

3
00:00:57,908 --> 00:01:03,414
所有陆地生命归根结底都依赖於淡水

Installation

To install the latest stable version from PyPi:

.. code::

pip install -U srt

To install the latest development version directly from GitHub:

.. code::

pip install -U git+https://github.com/cdown/srt.git@develop

Testing

.. code::

tox

.. _Tox: https://tox.readthedocs.org .. _Detailed API documentation: http://srt.readthedocs.org/en/latest/api.html .. _tools shipped with the library: https://github.com/cdown/srt/tree/develop/srt_tools