Compile, archive, unpack, and load compiled modules leveraging v8 cached data.
- Status
- Installation
- Abstract
- Basic Usage
- API
- Command Line Interface
- Constraints
- See Also
- Prior Art
- License
Stable/Documentation
$ npm install tiny-module-compiler
The tiny-module-compiler
module is small toolkit for compiling
JavaScript CommonJs modules into standalone binaries that leverage that
v8 cache data format exposed by the vm
API. Compiled module objects can be archived into a single file based on the
TinyBox file format and then unpacked later to the file
system.
This toolkit allows for the compilation of an entire project into a single
compiled binary object file. Multiple binary object files and various
assets (*.node
, *.so
, etc) can be packaged into an archive and
unpacked to file system for later use making it suitable as a delivery
mechanism.
Compiled modules objects can be loaded and executed but must be in a runtime that uses the same version of v8.
const tmc = require('tiny-module-compiler')
const targets = '*.js'
const archiveName = 'modules.a'
// compile module targets into compiled module objects
tmc.compile(targets, (err, objects) => {
// archive objects into file specified by `archiveName`
tmc.archive(archiveName, objects, (err) => {
tmc.load(archiveName, (err, archive) => {
// `archive` contains a mapping of compiled module object
// filenames to loaded exports
})
})
})
The tiny-module-compiler
module exports a public API described in this
section.
Compiles a file specified at target
. The default behavior is to write
the output to a file of the same name as target
with .out
appended
to the end. This behaviour can be configured by specifying an
options.output
option or options.storage
as a custom
random-access-storage factory function.
The value of options
is optional and can be:
{
// current working directory for compilation
cwd: process.cwd(),
// output filename (single file compilation) or directory (multiple files)
output: target + '.out',
// custom storage factory function to return
// 'random-access-storage' compliant object
storage(filename) {
return require('random-access-file')(filename)
}
}
A simple compilation example that compiles a target input file to an output file.
const { compile } = require('tiny-module-compiler')
// compile this file
const target = __filename
const output = __filename + '.out'
compile(target, { output }, (err) => {
// `target` compiled to `output`
})
A simple compilation example that compiles a target input file to an in
memory random-access-memory
storage.
const { compile } = require('tiny-module-compiler')
const ram = require('random-access-memory')
// compile this file
const target = __filename
const storage = ram()
compile(target, { storage: () => storage }, (err) => {
// `target` compiled and written to `storage`
})
Archives the entries found in a given inputs
Map
into target
. The
default behaviour is to enumerate the entries in inputs
and write them
to a TinyBox instance specified at target
where the keys
and values of the entries are "put" into the TinyBox storage that lives
on the file system. This behaviour can be configured by specifying
options.storage
as a custom random-access-storage instance.
The value of options
is optional and can be:
{
// if `false`, will not truncate archive storage
truncate: true,
// custom 'random-access-storage' compliant object
storage: require('random-access-file')(target)
}
A simple example that archives the compiled objects of a compiled file.
const { compile, archive } = require('tiny-module-compiler')
// compile this file
const target = __filename
compile(target, (err, objects) => {
archive(__filename + '.a', objects, (err) => {
// `target` is compiled and then archived
})
})
A simple example that archives the compiled objects of a compiled file
into an in memory random-access-memory
storage.
const { compile, archive } = require('tiny-module-compiler')
const ram = require('random-access-memory')
// compile this file
const target = __filename
compile(target, { storage: ram }, (err) => {
const storage = ram()
archive(__filename + '.a', objects, { storage}, (err) => {
// `target` is compiled and then archived
// `storage` contains archived objects
})
})
Loads the exports from a module or the entries in archive specified at
target
. The default behaviour is to load the contents of the file
specified at target
as a compiled module and call callback(err, exports)
upon success or error with the module exports. If target
is
a path to an archive represented by a TinyBox, then the
entries are loaded and callback(err, archive)
is called upon success
or error. The storage of the compiled module or archive can be
explicitly set by specifying options.storage
as a custom
random-access-storage instance.
The value of options
is optional and can be:
{
// custom 'random-access-storage' compliant object
storage: require('random-access-file')(target),
// custom cache `Map` store for loaded modules
cache: new Map(),
// current working directory to load module in
cwd: process.cwd()
}
A simple compilation and load example that compiles a target input file to an output file and then loads the exports.
file.js:
module.exports = {
hello() {
return 'world'
}
}
compile-and-load.js:
const { compile, load } = require('tiny-module-compiler')
// compile this file
const target = 'file.js'
const output = target + '.out'
compile(target, { output }, (err) => {
// `target` compiled to `output`
load(output, (err, exports) => {
// `exports` points to `module.exports` in `output`
console.log(exports.hello()) // 'world'
})
})
A simple compilation and load example that compiles a target input file to an in memory storage and then loads the exports from it.
file.js:
module.exports = {
hello() {
return 'world'
}
}
compile-and-load-in-memory.js:
const { compile, load } = require('tiny-module-compiler')
const ram = require('random-access-memory')
// compile this file
const target = 'file.js'
const storage = ram()
compile(target, { storage: () => storage }, (err) => {
// `target` compiled to `storage`
load('loaded from memory', { storage }, (err, exports) => {
// `exports` points to `module.exports` in `output`
console.log(exports.hello()) // 'world'
})
})
Unpacks target
archive entries. The default behaviour is to enumerate
the entries in the archive specified at target
and copy them to the
file system. This behaviour can be configured by specifying a options.storage
random-access-storage factory function to provide custom
storage for the archive entries. If target
is a random-access-storage
instance, it will be used instead of reading from the file system.
The value of options
is optional and can be:
{
// output path for default storage
output: process.cwd(),
// custom storage factory function to return
// 'random-access-storage' compliant object
storage(filename) {
return require('random-access-file')(filename)
}
}
A simple example to unpack an archive's entries to file system.
const { unpack } = require('tiny-module-compiler')
const archive = '/path/to/archive'
unpack(archive, (err, entries) => {
// `archive` entries unpacked to file system
console.log(entries) // array of unpacked file names
})
A simple example to unpack an archive's entries to an in memory file store.
const { unpack } = require('tiny-module-compiler')
const ram = require('random-access-memory'
const archive = '/path/to/archive'
const files = new Map()
unpack(archive, { storage: createStorage }, (err, entries) => {
// `archive` entries unpacked to file system
console.log(entries) // array of unpacked file names
for (const entry of entries) {
console.log(files[entry.filename])
}
})
function createStorage(filename) {
return files.set(filename, ram()).get(filename)
}
- Extends:
nanoresource-pool
The Compiler
class represents a container of compile targets
that can be compiled into a single binary file containing
v8 cache data and header information about the compiled output.
Creates a new Compiler
instance where options
can be:
{
// current working directory for compilation
cwd: process.cwd()
}
const compiler = new Compiler()
Array<String>
All opened targets in the compiler.
for (const target of compiler.targets) {
// handle opened `target`
}
Waits for compiler to be ready and calls callback()
upon success.
compiler.ready(() => {
// `compiler` is opened and ready
})
Creates and returns a new compile target that is added
to compiler pool calling callback(err, target)
when the target resource
is opened or an error occurs. The target will be compiled when
compiler.compiler()
is called.
The value of options
can be:
{
// the default output for a compilation target
output: target + '.out',
// custom storage factory function to return
// 'random-access-storage' compliant object
storage(filename) {
return require('random-access-file')(filename)
}
}
compiler.target('/path/to/file.js', (err, target) => {
// `target` is an opened `nanoresource`
})
Compiles all pending compile targets calling
callback(err, objects, assets)
upon success or error. Callback will be
given a Map
of compiled objects and a Map
of extracted assets that
should live with the compiled objects on the file system.
The value of options
can be:
{
// if `true`, will produce a source map stored in the `assets` map
map: false,
// if `false`, `ncc` will produce verbose output
quiet: true,
// if `true`, will use `ncc` cache for faster builds
cache: false,
// if `true`, will produce compiled javascript source debug output
debug: false,
// if `true`, will minify compiled javascript source before creating binary
optimize: false,
// an array of external modules that should _not_ be compiled
externals: [],
}
compiler.compile({ externals: ['sodium-native'] }, (err, objects, assets) => {
// `objects` is a `Map` mapping file names to compiled module objects
// `assets` is a `Map` mapping file names to assets that should be copied
})
- Extends:
nanoresource
The Archiver
class represents an abstraction for storing
compiled objects into a TinyBox
Creates a new Archiver
instance where options
can be:
{
// default custom storage factory function to return
// 'random-access-storage' compliant object used in
// `archiver.archive()` calls. This storage can be overloaded
// by supplying a storage factory function to `archiver.archive()`
storage(filename) {
return require('random-access-file')(filename)
}
}
const ram = require('random-access-memory')
// in memory archiver
const archiver = new Archiver({ storage: ram })
Waits for archiver to be ready and calling callback()
when it is.
archiver.ready(() => {
// `archiver` is opened and ready
})
Archives the entries found in a given inputs
Map
into target
. The
underlying storage for filename
can be given by options.storage
or a
new one is created by the storage
factory function given to the
Archiver
class constructor.
callback(err)
is called upon success or error.
The value of options
can be:
{
// if `false`, will not truncate archive storage
truncate: true,
// custom 'random-access-storage' compliant object
// where `inputs` are archived to
storage: require('random-access-file')(filename)
}
const extend = require('map-extend')
compiler.compile((err, objects, assets) => {
// merge `objects` and `assets` and archive
archiver.archive('/path/to/archive', extend(objects, assets), (err) => {
// inputs should be archived
})
- Extends:
nanoresource-pool
The Loader
class represents an abstraction for loading compiled
module objects and JavaScript sources as node modules.
Creates a new Loader
instance where options
can be:
{
// custom cache `Map` store for loaded modules
cache: new Map()
}
Waits for loader to be ready and calling callback()
when it is.
loader.ready(() => {
// `loader` is opened and ready
})
Loads a compiled module object or JavaScript source module
specified at filename calling callback(err, exports)
upon success
or error. Success loads will cache resulting module for subsequent
requests to load the module.
The value of options
can be:
{
// custom 'random-access-storage' compliant object
// where module is loaded from
storage: require('random-access-file')(filename)
}
loader.load('/path/to/compiled/module.js', (err, exports) => {
// `exports` points to `module.exports` of loaded module
})
- Extends:
nanoresource
The Target
class represents a nanoresource
to a target
file backed by a random-access-storage
instance.
Creates a new Target
instance where filename
is the name of the target file and options
can be:
{
// custom 'random-access-storage' compliant object
// where file is loaded from
storage: require('random-access-file')(filename)
}
Note: This class is intended for internal and advanced use. You will most likely not use this directly.
const target = new Target('/path/to/file.js')
?(Number)
The active file descriptor for the target resource. Will be null
if not opened.
Queries for stats from the underlying target storage calling
callback(err, stats)
upon success or error.
target.stat((err, stats) => {
console.log(stats.size)
})
Reads data from the underlying target storage at a specified offset
and
size
calling callback(err, buffer)
upon success or error.
taret.read(32, 64, (err, buffer) => {
console.log(buffer)
})
Waits for loader to be ready and calling callback()
when it is.
target.ready(() => {
// `target` is opened and ready
})
The tiny-module-compiler
exposes a command line interface through the
tmc
command that is suitable for compiling, archiving, and unpacking
compiled modules and their assets.
This section describes the command line interface and a few workflows
for making the best use of the tmc
command.
The tmc
command has a command line signature of the following:
tmc [-hV] [-acu] [-vCDMO] [options] ...input
Where options
can be:
-a, --archive If present, will archive input into "tinybox" format
-c, --compile If present, will compile input into header prefixed v8 cached data
, --concurrency <jobs> An alias for '--jobs'
-C, --copy-assets If present, will copy assets to directory of output
-D, --debug If present, will enable debug output (DEBUG=tiny-module-compiler)
-e, --external <module> Specifies an external dependency that will be linked at runtime
-h, --help If present, will print this message
-l, --load If present, will load inputs
-j, --jobs <jobs> Specifies the number of concurrent jobs for batch tasks (--load, --archive, --copy-assets)
-M, --source-map If present, a source map will be generated
-o, --output <path> If present, will change the output path. Assumes directory if multiple inputs given
-O, --optimize If present, will optimize output by minifying JavaScript source prior to compilation
-u, --unpack If present, will treat input as an archive and will unpack files to path specified by '--output'
-v, --verbose If present, will emit verbose output to stdout/stderr
-V, --version If present, will print the version number
-x An alias for '--external'
The simplest use of the tmc
command is to compile a single file. By
default, the -c
(or --compile
) is assumed if -a
(or
--archive
) and -u
(or --unpack
) flags are not present.
$ tmc file.js ## will write ./file.js.out
The output of the compiled file can be configured using the -o
(or --output
)
flag to set the output name.
$ tmc file.js -o file.compiled.js
If multiple inputs were given, then the output value will be a directory and the original file names are preserved.
$ tmc *.js -o build/
After compiling modules it may be useful to archive them into a single file. This is often beneficial if the compiled module contained static assets that need to live on the file system alongside the compiled module file during runtime.
The tmc
command makes it easy to specify a number of files that
should be added to an archive by making use of the -a
(or --archive
)
flag to indicate that inputs should be archived.
$ tmc -a modules.archive *.compiled.js ## archives all `*.compiled.js` files into `modules.archive`
Any asset can be added to an archive as well.
$ tmc -a modules.archive *.node ## archives all `*.node` files into `modules.archive`
The modules.archive
file contains an index of entries added to it by
file name. The archive can be unpacked using the tmc
command as well.
Compiled modules and their assets living in an archive can be easily
unpacked to the file system. The tmc
command makes it easy to unpack
entries in an archive.
$ tmc -u modules.archive
The output of the entries can be specified with the -o
(or --output
)
flag.
$ tmc -u modules.archive -o build/
Compiled modules and packed archives can be loaded at runtime in an
application. The tmc
command provides a simple way to do this from the
command line from array of inputs by making use of the -l
(or
--load
) flag to indicate the inputs should be loaded.
$ tmc -l module.compiled.js
Archives can be loaded as well.
$ tmc -l modules.archive
Regular JavaScript files an be given as input.
$ tmc -l file.js
Multiple inputs can be loaded.
$ tmc -l modules.archive file.js module.compiled
The load concurrency can be set by specifying the -j
(or --jobs
or
--concurrency
) flag to indicate the number of concurrent loads that
can occur when loading the inputs.
$ tmc -l -j1 modules.archive file.js module.compiled ## one at a time
The tiny-module-compile
module ships with a few constrains due to the
nature of how v8 cache data works. They are described in this section.
The v8 version (process.versions.v8
) in the runtime of node
used
to compile modules to v8 cache data must match the one used to load. The
module will throw an error if the versions do not match as unexpected
behaviour may occur.
For user-defined functions,
Function.prototype.toString()
may not work as expected or not at all because the string source code
representation of the function is lost after compilation.
- https://github.com/OsamaAbbas/bytenode
- https://github.com/zertosh/v8-compile-cache
- https://github.com/v8/v8/blob/master/src/snapshot/code-serializer.h
- https://hackernoon.com/how-to-compile-node-js-code-using-bytenode-11dcba856fa9
MIT