Add documentation.

README.md

@@ -1,9 +1,84 @@
 # livepatch
 [![Build Status](https://travis-ci.org/pagarme/livepatch.svg)](https://travis-ci.org/pagarme/livepatch) [![Coverage Status](https://coveralls.io/repos/pagarme/livepatch/badge.svg?branch=master)](https://coveralls.io/r/pagarme/livepatch?branch=master)
 
-
 Patch JSON streams on the fly.
 
+# Example
+
+```js
+var fs = require('fs');
+var livepatch = require('livepatch');
+
+fs.createReadStream('test.json')
+	.pipe(livepatch(function() {
+		/* Transformations goes here */
+
+		this.rename('$.name_*', 'name'); // Renames all keys in the root starting with 'name_' to 'name'
+
+		this.remove('$._*'); // Removes all fields in the root starting with an underscore
+
+		// Advanced usage
+		this.rewrite('$.author', function(match) {
+			return {
+				rename: 'author_name',
+				value: match.value.toUpperCase()
+			};
+		});
+	}))
+	.pipe(fs.createWriteStream('output.json'));
+```
+
+# Documentation
+
+## rename(path, name)
+
+Renames all fields matched by `path` to `name`.
+
+## remove(path)
+
+Remove all fields matched by `path`.
+
+## rewrite(path, rewriteFn)
+
+Rewrite all fields matched by `path` using `rewriteFn`.
+
+The passed function receives an argument with the following information:
+
+**key:** current key name
+**path:** current path(as an array)
+**value:** current key value(if available)
+
+It should return a JSON object with atleast one of the following actions:
+
+**rename:** Renames the fields with the given value
+**remove:** If `true`, removes the key
+**value:** Changes the key value
+
+## path
+
+All paths used in transformations are based on the JSONPath spec.
+
+`$` denotes the root of the object
+`.` denotes the end of a key name
+`*` denotes any range of characters
+
+### Examples
+
+`$.name`
+Matches the `name` field.
+
+`$.books[*].author`
+Matches all books authors.
+
+`$.books[*].*_name`
+Matches all fields in a book that ends with `_name`.
+
+# How it works
+
+LivePatch works by reading a JSON stream with [clarinet](https://github.com/dscape/clarinet) and doing live modifications based on the current path in the stream. After patching, it immediatly serializes to the next stream the resulting JSON.
+
+It's very useful when dealing with very large objects which came from a stream(e.g.: ElasticSearch results) that need to be modified and streamed to another place(e.g.: network). As it doesn't hold the entire data on memory you don't need to worry about loading large files.
+
 # License
 
 Check [here](LICENSE).