This project is no longer maintained.
In the end the approach it took didn't work well in many cases and it was hard to optimize further. Therefore I won't be putting any more work into it.
See its successor instead: https://github.com/asciinema/agg
asciicast2gif
asciicast2gif is a tool for generating GIF animations from asciicast files recorded by asciinema.
How it works
Here's how the asciicast->GIF conversion is implemented.
asciicast2gif
shell script parses command line arguments and executes Node.js script
(main.js
). main.js
loads asciicast (either from remote URL or local
filesystem), generates text representation of the screen for each frame
using asciinema-player's
virtual terminal emulator, and sends it to PhantomJS-based renderer script
(renderer.js
), which saves PNG screenshots to a temporary directory. Finally,
main.js
calls ImageMagick's convert
on these PNG images to construct GIF
animation, also piping it to gifsicle
to get the final, optimized GIF file.
Note, asciicast2gif
doesn't capture screenshots at a fixed frame-rate (e.g. 30 FPS)
like alternative tools. Instead, it generates PNG files for
each screen update, and specifies delay for every image individually (convert -delay <delay-a> 0.png -delay <delay-b> 1.png -delay <delay-c> 2.png ...
). When
the screen is idle there're no screenshots generated. This saves disk space and
makes it less work for both convert
and gifsicle
, while resulting in smaller
GIF file.
Installation
You can install asciicast2gif with npm
, use it via Docker image, or build it
from source.
npm package
To install asciicast2gif using npm
run:
npm install --global asciicast2gif
Following runtime dependencies need to be also installed:
- ImageMagick
- giflossy (or gifsicle)
Docker image
You can use asciicast2gif through official asciicast2gif Docker image (automated build on Docker Hub from this repository).
Pull the image:
docker pull asciinema/asciicast2gif
Use it like this:
docker run --rm -v $PWD:/data asciinema/asciicast2gif [options and arguments...]
You need to mount some local directory at /data
so input and output files can
be accessed by the container. Mounting current working directory ($PWD
) makes
most sense in majority of cases.
For example, generating GIF from local file, with double speed and Solarized theme:
docker run --rm -v $PWD:/data asciinema/asciicast2gif -s 2 -t solarized-dark demo.json demo.gif
Running the above, long command can get old very quickly. Creating a shell alias may be a good idea:
alias asciicast2gif='docker run --rm -v $PWD:/data asciinema/asciicast2gif'
Look at general usage instructions below for all command line arguments, options etc.
Note: if you want to override gifsicle options (via GIFSICLE_OPTS
env var)
when using Docker image you need to pass it via Docker's -e
option.
Building from source
Clone the repository:
git clone --recursive https://github.com/asciinema/asciicast2gif.git
cd asciicast2gif
All further commands are assumed to be called from within the checked out directory.
Install build time dependencies
Both Node.js script (main.js
) and page script used by renderer's HTML page
(page/page.js
) need to be build from
ClojureScript source code.
You need Java 8 JDK and Leiningen.
Install runtime dependencies
Following runtime dependencies need to be installed:
- Node.js
- PhantomJS (optional, see below)
- ImageMagick
- giflossy (or gifsicle)
Install Node.js wrapper for PhantomJS:
npm install
If you don't have PhantomJS available in $PATH
at this point it will be
automatically downloaded during
phantomjs-prebuilt
package installation.
Build
To build the scripts run:
lein cljsbuild once main && lein cljsbuild once page
Usage
asciicast2gif [-t theme] [-s speed] [-S scale] [-w cols] [-h rows] <input-json-path-or-url> <output-gif-path>
Following options are supported:
-t <theme> color theme, one of: asciinema, tango, solarized-dark, solarized-light, monokai (default: asciinema)
-s <speed> animation speed (default: 1)
-S <scale> image scale / pixel density (default: 2)
-w <columns> clip terminal to specified number of columns (width)
-h <rows> clip terminal to specified number of rows (height)
Example of generating GIF from asciicast URL, with default options (normal speed, double pixel density, asciinema theme):
asciicast2gif https://asciinema.org/a/118274.json demo.gif
Example of generating GIF from local asciicast file, with Solarized Dark theme,
double speed (-s 2
), single pixel density (-S 1
):
asciicast2gif -t solarized-dark -s 2 -S 1 118274.json demo.gif
Setting custom gifsicle options
You can override default options passed to giflossy/gifsicle by setting
GIFSICLE_OPTS
environment variable.
Default options when giflossy is installed are:
-k 64 -O2 -Okeep-empty --lossy=80
Default options when gifsicle is installed are:
-k 64 -O2 -Okeep-empty
For example to generate 16 color gif, with level 3 optimizations, set it like this:
GIFSICLE_OPTS="-k 16 -O3"
Debugging
You can set DEBUG=1
environment variable to make the output of the conversion
process more verbose.
Tweaking conversion process
You can pass extra arguments to Node.js script invocation via NODE_OPTS
environment variable.
Limiting node process memory usage
Limit Node's heap size to 512 MB:
NODE_OPTS="--max-old-space-size=512" asciicast2gif ...
Alternatives
There are following alternative tools solving this problem:
- marionebl/svg-term-cli - generates animated SVG from local or asciinema.org hosted asciicast
- tav/asciinema2gif - generates animated GIF from asciinema.org hosted asciicast
- pettarin/asciicast2gif - generates animated GIF from local asciicast
- anishkny/webgif - generates animated GIF from any website:
npm i -g webgif && webgif -u https://asciinema.org/a/147023?t=0
License
Copyright © 2017 Marcin Kulik.