Your job is to parse a partipants description of an image containing colored shapes.
The set of possible shapes is: <triangle cloud cross umbrella heart star circle square>. 
The set of possible colors is: <red green blue gold purple saddlebrown gray black darkorange cyan>.

Your response should take the form of a json describing the color and shape of each object referred to in the description. 
When the participant describes the image, they may not use exactly the correct shape or color, so you must infer which valid color/shape is closest to the ones they describe. 
Pay close attention to whether the shapes and colors in your response are spelled correctly and are in lowercase.

For example, if the participants description were: 
"There are four objects in the image, a brown triangle, a blue cloud, and two green crosses"

The JSON output would be:
[
    {
      "shape": "triangle",
      "color": "saddlebrown"
    },
    {
      "shape": "cloud",
      "color": "blue"
    },
    {
      "shape": "cross",
      "color": "green"
    },
    {
      "shape": "cross",
      "color": "green"
    }
]
Even if the participant already describes the image in json format, you should still parse it and return the correct json. 
Pay close attention to whether the shapes and colors in your response are spelled exactly as they are spelled in the description above.
Return only the json object with abolutely no other text or delimiting characters.
Parse the following description: 