This package contains tools for helping you write transforms for browserify.
Many different transforms perform certain basic functionality, such as turning the contents of a stream into a string, or loading configuration from package.json. This package contains helper methods to perform these common tasks, so you don't have to write them over and over again:
makeStringTransform()
creates a transform which consumes and returns a string, instead of using a stream.makeFalafelTransform()
parses a JS file using falafel and allows you to modify the code.makeRequireTransform()
passes you the contents of each require()
call in each script, and allows you to rewrite the require statement.make*Transform()
functions will support, then loadTransformConfig()
will load configuration for you.runTransform()
can be used to unit test your shiny new transform.Install with npm install --save browserify-transform-tools
.
Browserify transforms work on streams. This is all well and good, until you want to call a library like "falafel" which doesn't work with streams. (If you're using falafel specifically, see below for makeFalafelTransform
.)
Suppose you are writing a transform called "unbluify" which replaces all occurances of "blue" with a color loaded from a configuration:
var options = {excludeExtensions: [".json"]};
module.exports = transformTools.makeStringTransform("unbluify", options,
function (content, transformOptions, done) {
var file = transformOptions.file;
if(!transformOptions.config) {
return done(new Error("Could not find unbluify configuration."));
}
done(null, content.replace(/blue/g, transformOptions.config.newColor));
});
Notice that the color we replace "blue" with gets loaded from configuration. The configuration can be set in a variety of ways. A simple example is to set it directly in package.json:
{
"name": "myProject",
"version": "1.0.0",
...
"unbluify": {"newColor": "red"}
}
See the section on "Loading Configuration" below for details on where configuration can be loaded from.
Parameters for makeStringTransform()
:
transformFn(contents, transformOptions, done)
- Function which is called to
do the transform. contents
are the contents of the file. done(err, transformed)
is
a callback which must be called, passing the a string with the transformed contents of the
file. transformOptions
consists of:
transformOptions.file
is the name of the file (as would be passed to a normal browserify transform.)
transformOptions.config
is the configuration for your transform, loaded either from
browserify or from package.json.
transformOptions.configData
is the configuration data for the transform (see
loadTransformConfig
below for details on where this comes from.)
options.excludeExtensions
- A list of extensions which will not be processed. e.g.
"['.coffee', '.jade']"
options.includeExtensions
- A list of extensions to process. If this options is not
specified, then all extensions will be processed. If this option is specified, then
any file with an extension not in this list will skipped.
options.jsFilesOnly
- If set true, then your transform will only run on "javascript" files.
This is handy for Falafel and Require transforms, defined below. This is equivalent to
passing
includeExtensions: [".js", ".coffee", ".coffee.md", ".litcoffee", "._js", "._coffee", ".jsx", ".es", ".es6"]
.
Many transforms are based on falafel. browserify-transform-tools provides an easy way to define such transforms. Here is an example which wraps all array expressions in a call to fn()
:
var options = {};
// Wraps all array expressions in a call to fn(). e.g. '[1,2,3]' becomes 'fn([1,2,3])'.
module.exports = transformTools.makeFalafelTransform("array-fnify", options,
function (node, transformOptions, done) {
if (node.type === 'ArrayExpression') {
node.update('fn(' + node.source() + ')');
}
done();
});
makeFalafelTransform()
will be called once for every node in your JS file. You can update the node. Be sure to pass errors back via done(err)
, and call done()
when complete.
Options passed to makeFalafelTransform()
are the same as for makeStringTransform()
, as are the transformOptions passed to the transform function. You can additionally pass a options.falafelOptions
to makeFalafelTransform
- this object will be passed as an options object directly to falafel.
Many transforms are focused on transforming require()
calls. browserify-transform-tools has a solution for this:
transform = transformTools.makeRequireTransform("requireTransform",
{evaluateArguments: true},
function(args, opts, cb) {
if (args[0] === "foo") {
return cb(null, "require('bar')");
} else {
return cb();
}
});
This will take all calls to require("foo")
and transform them to require('bar')
. Note that makeRequireTransform can parse many simple expressions, so the above would succesfully parse require("f" + "oo")
, for example. Any expression involving core JavaScript, __filename
, __dirname
, path
, and join
(where join is an alias for path.join
) can be parsed. Setting the evaluateArguments
option to false will disable this behavior, in which case the source code for everything inside the ()s will be returned.
Note that makeRequireTransform
expects your function to return the complete require(...)
call. This makes it possible to write require transforms which will, for example, inline resources.
Again, all other options you can pass to makeStringTransform
are valid here, too.
These transforms are focused on transforming arbitrary function calls:
transform = transformTools.makeRequireTransform("requireTransform",
{evaluateArguments: true, functionNames: ["foobar"]},
function(functionParams, opts, cb) {
if (functionParams.args[0].value === "foo") {
return cb(null, functionParams.name + "('bar')");
} else {
return cb();
}
});
This will take all calls to foobar("foo")
and transform them to foobar('bar')
. Note that makeFunctionTransform can parse many simple expressions, so the above would succesfully parse foobar("f" + "oo")
, for example. Any expression involving core JavaScript, __filename
, __dirname
, path
, and join
(where join is an alias for path.join
) can be parsed. Setting the evaluateArguments
option to false will disable this behavior, in which case the source code for everything inside the ()s will be returned.
Note that makeFunctionTransform
expects your function to return the complete [functionName](...)
call. This makes it possible to write function transforms which will, for example, inline resources.
The option functionNames
can either be a string or an array of strings. If no functionName is provided makeFunctionTransform
fallbacks to require()
calls.
The functionParams
object which is passed to the given transform function has 2 attributes. The first one is name
which is the function name The second one is args
and is an ordered array of the function args. Each entry consists of value
and type
. Type can be one of these values: Literal, Identifier, FunctionExpression, ObjectExpression, ArrayExpression
.
Again, all other options you can pass to makeStringTransform
are valid here, too.
All make*Transform()
functions will automatically load configuration for your transform and make it available via transformOptions.config
(and through the more detailed transformOptions.configData
.) Functions are also provided for reading configuration if you are not using one of the make*Transform()
functions.
Transform configuration can be loaded from a project's package.json file, from a js or coffee file specified in package.json, or programatically. For details, see the transform configuration documentation.
If you want to unit test your transform, then runTransform()
is for you:
var myTransform = transformTools.makeFalafelTransform(...);
var dummyJsFile = path.resolve(__dirname, "../testFixtures/testWithConfig/dummy.js");
var content = "console.log('Hello World!');";
transformTools.runTransform(myTransform, dummyJsFile, {content: content},
function(err, transformed) {
// Verify transformed is what we expect...
}
);
Some of this was heavily inspired by: