'data' URI Syntax

"data:" + valueList + "," + data

• "data:" is case-insensitve.
• valueList is a ";"-separated list of values. The first value is a percent-encoded value representing the mime type. Although "/" and "+" in the mime type can be percent-encoded as "%2F" and "%2B" repsepectively, they are not required to be.
• For the rest of the values in valueList, if the value does not contain a "=", then it is a content-encoding value (like base64). If the value does contain a "=", then the value is a name=value pair where name and value are percent-encoded representations of a name and value.
• valid names in a name=value pair are "charset", "filename" and "content-disposition"
• The mime type of the URI is the mime type value after percent-decoding it, trimming leading and trailing white-space from it and converting it to lowercase. If the value is then empty, the mime type for the URI is "text/plain".
• For content-encoding values, before determining the content encoding value, it must be percent-decoded, stripped of leading and trailing white-space and converted to lowercase. Further, the value is not used if its length is 0 and when there are multiple content-encodings in the valueList, the first non-empty (if any) one that has a supported value is used and the rest are ignored. Supported values may differ between UAs. "base64" is usually the only supported value at the moment.
• If a value is determined to be a name=value pair, the value must be split by the first "=". The value on the left-side of the "=" is the name and the value on the right-side of the first "=" is the value. Then, the name and value must each be percent-decoded, stripped of leading and trailing white-space and converted to lowercase. Further, if more than one value in the valueList contains a name=value pair with the same name, the first one with a non-empty value (if any) is used and the rest of the duplicates are ignored.
• If a charset value is not found, then it defaults to US-ASCII.
• data is a percent-encoded value representing the file's data. To get the data, it must be percent-decoded.