Implementation suggestion

• Reads every _documents.xml files
  • build the URL from the topic attribute of <documents>, and the article tag of each <document>
  • use as is article and title tags, as they are designed to be concise.
• Take a decision about what to do with index pages (some matches tools, some are list of several tools sharing the same directory or in subdirectories). Include them is acceptable by the way, they would match legitimate results and would finally lead to the right tool.
• Outputs as a JSON document.

Output example

A simple JSON schema could be an array of objects with title, description and URL properties:

[
    {
        "title": "List operations",
        "description": "Adds, subtracts or intersects lists.",
        "url": "/lists/operations"
    },
    {
        "title": "List replace",
        "description": "Performs regexp replacement in a list.",
        "url": "/lists/replace"
    }
]