Overview 

parsley is a simple language for data-extraction from XML-like documents (including HTML). parsley is:

1. Blazing fast -- Typical HTML parses are sub-50ms.
2. Easy to write and understand -- parsley uses your current knowledge of JSON, CSS, and XPath.
3. Powerful. parsley can understand full XPath, including standard and user-defined functions.

Examples

A simple script, or "parslet", looks like this:

{
  "title": "h1",
  "links(a)": [
    {
      "text": ".",
      "href": "@href"
    }
  ]
}

This returns JSON or XML output with the same structure. Applying this parslet to http://www.yelp.com/biz/amnesia-san-francisco yields either:

{
  "title": "Amnesia",
  "links": [
    {
      "href": "\/",
      "text": "Yelp"
    },
    {
      "href": "\/",
      "text": "Welcome"
    },
    {
      "href": "\/signup?return_url=%2Fuser_details",
      "text": " About Me"
    },
    .....
  ]
}

or equivalently:

<parsley:root>
  <title>Amnesia</title>
  <links>
    <parsley:group>
      <href>/</href>
      <text>Yelp</text>
    </parsley:group>
    <parsley:group>
      <href>/</href>
      <text>Welcome</text>
    </parsley:group>
    <parsley:group>
      <href>/signup?return_url=%2Fuser_details</href>
      <text> About Me</text>
    </parsley:group>
    .....
  </links>
</parsley:root>      

This parslet could also have been expressed as:

{
  "title": "h1",
  "links(a)": [
    {
      "text": ".",
      "href": "@href"
    }
  ]
}

The "a" in links(a) is a "key selector" -- an explicit grouping (with scope) for the array. You can use any XPath 1.0 or CSS3 expression as a value or a key selector. Parsley will try to be smart, and figure out which you are using. You can use CSS selectors inside XPath functions -- "substring-after(h1>a, ':')" is a valid expression.

Variables

You can use $foo to access the value of the key "foo" in the current scope (i.e. nested curly brace depth). Also available are $parent.foo, $parent.parent.foo, $root.foo, $root.foo.bar, etc.

Custom Functions

You can write custom functions in XSLT (I'd like to also support C and JavaScript). They look like:

<func:function name="user:excited">
   <xsl:param name="input" />
   <func:result select="concat($input, '!!!!!!!')" />
</func:function>

If you run:

{
  "title": "user:excited(h1)",
}

on the Yelp page, you'll get:

{
  "title": "Amnesia!!!!!!!",
}