This library provides FFmpeg builds ported to JavaScript using Emscripten project. Builds are optimized for in-browser use: minimal size for faster loading, asm.js, performance tunings, etc. Though they work in Node as well.
This is a fork of the original FFmpeg.js project which adds a custom video filter used by the AudioMoth Filter Playground.
See documentation on Module object for the list of options that you can pass.
ffmpeg.js provides common module API, ffmpeg-webm.js
is the default module. Add its name after the slash if you need another build, e.g. require("ffmpeg.js/ffmpeg-mp4.js")
.
const ffmpeg = require("ffmpeg.js");
let stdout = "";
let stderr = "";
// Print FFmpeg's version.
ffmpeg({
arguments: ["-version"],
print: function(data) { stdout += data + "\n"; },
printErr: function(data) { stderr += data + "\n"; },
onExit: function(code) {
console.log("Process exited with code " + code);
console.log(stdout);
console.log(stderr);
},
});
Use e.g. browserify in case of Browser.
ffmpeg.js also provides wrapper for main function with Web Worker interface to offload the work to a different process. Worker sends the following messages:
{type: "ready"}
- Worker loaded and ready to accept commands.{type: "run"}
- Worker started the job.{type: "stdout", data: "<line>"}
- FFmpeg printed to stdout.{type: "stderr", data: "<line>"}
- FFmpeg printed to stderr.{type: "exit", data: "<code>"}
- FFmpeg exited.{type: "done", data: "<result>"}
- Job finished with some result.{type: "error", data: "<error description>"}
- Error occurred.{type: "abort", data: "<abort reason>"}
- FFmpeg terminated abnormally (e.g. out of memory, wasm error).
You can send the following messages to the worker:
{type: "run", ...opts}
- Start new job with provided options.
const worker = new Worker("ffmpeg-worker-webm.js");
worker.onmessage = function(e) {
const msg = e.data;
switch (msg.type) {
case "ready":
worker.postMessage({type: "run", arguments: ["-version"]});
break;
case "stdout":
console.log(msg.data);
break;
case "stderr":
console.log(msg.data);
break;
case "done":
console.log(msg.data);
break;
}
};
You can use worker_threads module in case of Node.
Empscripten supports several types of file systems. ffmpeg.js uses MEMFS to store the input/output files in FFmpeg's working directory. You need to pass Array of Object to MEMFS
option with the following keys:
- name (String) - File name, can't contain slashes.
- data (ArrayBuffer/ArrayBufferView/Array) - File data.
ffmpeg.js resulting object has MEMFS
option with the same structure and contains files which weren't passed to the input, i.e. new files created by FFmpeg.
const ffmpeg = require("ffmpeg.js");
const fs = require("fs");
const testData = new Uint8Array(fs.readFileSync("test.webm"));
// Encode test video to VP8.
const result = ffmpeg({
MEMFS: [{name: "test.webm", data: testData}],
arguments: ["-i", "test.webm", "-c:v", "libvpx", "-an", "out.webm"],
});
// Write out.webm to disk.
const out = result.MEMFS[0];
fs.writeFileSync(out.name, Buffer(out.data));
You can also mount other FS by passing Array of Object to mounts
option with the following keys:
- type (String) - Name of the file system.
- opts (Object) - Underlying file system options.
- mountpoint (String) - Mount path, must start with a slash, must not contain other slashes and also the following paths are blacklisted:
/tmp
,/home
,/dev
,/work
. Mount directory will be created automatically before mount.
See documentation of FS.mount for more details.
const ffmpeg = require("ffmpeg.js");
ffmpeg({
// Mount /data inside application to the current directory.
mounts: [{type: "NODEFS", opts: {root: "."}, mountpoint: "/data"}],
arguments: ["-i", "/data/test.webm", "-c:v", "libvpx", "-an", "-y", "/data/out.webm"],
});
// out.webm was written to the current directory.
It's recommended to use Docker to build ffmpeg.js.
-
Clone ffmpeg.js repository with submodules:
git clone https://github.com/pcprince/ffmpeg.js.git --recurse-submodules
-
Modify Makefile and/or patches if you wish to make a custom build.
-
Open in Powershell:
docker run --rm -it -v ${PWD}:/mnt -w /opt kagamihi/ffmpeg.js
-
Prepare files and install yasm:
cp -a /mnt/{.git,build,Makefile} . && source /root/emsdk/emsdk_env.sh && apt-get install yasm
-
Copy custom filter to correct location then run configure script to add custom filter:
cp -ar /mnt/customfilter/. build/ffmpeg-mp4/libavfilter/ && ./build/ffmpeg-mp4/configure
-
Build
make mp4 && cp ffmpeg*.js /mnt
That's it. ffmpeg.js modules should appear in your repository clone.
Based on the steps given in the ffmpeg docs.
-
Copy and rename an existing filter in
build\ffmpeg-mp4\libavfilter
sed 's/drawbox/audiomothanimation/g;s/DrawBox/AudioMothAnimation/g;s/DRAWBOX/AUDIOMOTHANIMATION/g' libavfilter/vf_drawbox.c > libavfilter/vf_audiomothanimation.c
Delete drawgrid code
-
Edit
build/ffmpeg-mp4/libavfilter/makefile
, adding the filter using the same pattern as the other filters -
Edit
build/ffmpeg-mp4/libavfilter/allfilters.c
, adding the filter here too -
Start docker
docker run --rm -it -v ${PWD}:/mnt -w /opt kagamihi/ffmpeg.js
-
Copy over files and install yasm
cp -a /mnt/{.git,build,Makefile} . && source /root/emsdk/emsdk_env.sh && apt-get install yasm
-
Run configure script to add custom filter
./build/ffmpeg-mp4/configure
-
Make
make mp4 && cp ffmpeg*.js /mnt
Own library code licensed under LGPL 2.1 or later.
This build uses LGPL version of FFmpeg and thus available under LGPL 2.1 or later. See here for more details and FFmpeg's license information.
Included libraries:
- libopus licensed under BSD.
- libvpx licensed under BSD.
See LICENSE.WEBM for the full text of software licenses used in this build.
This build uses GPL version of FFmpeg and thus available under GPL 2.0. It also includes patent encumbered H.264, AAC and MP3 encoders. Make sure to contact lawyer before using it in your country.
Included libraries:
- x264 licensed under GPL.
- LAME licensed under LGPL.
See LICENSE.MP4 for the full text of software licenses used in this build.