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, manim, bw_plex, and many more.
- Can parse many broken SRT files which other SRT libraries cannot, and fix them
- Extremely lightweight, ~150 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
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:
$ 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:
$ 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.
Detailed API documentation is available, but here are the basics:
>>> # list() is needed as srt.parse creates a generator
>>> subs = list(srt.parse('''\
... 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
...
... '''))
>>> subs
[Subtitle(index=1, start=datetime.timedelta(0, 33, 843000), end=datetime.timedelta(0, 38, 97000), content='地球上只有3%的水是淡水', proprietary=''),
Subtitle(index=2, start=datetime.timedelta(0, 40, 641000), end=datetime.timedelta(0, 44, 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(0, 57, 908000), end=datetime.timedelta(0, 63, 414000), content='所有陆地生命归根结底都依赖於淡水', proprietary='')]
>>> 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
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To install the latest stable version from PyPi:
pip install -U srt
To install the latest development version directly from GitHub:
pip install -U git+https://github.com/cdown/srt.git@develop
tox