autosuggest.md

Autosuggest and matching

Search-index offers two main strategies for autosuggest

1. Just do a normal search
2. Use the matcher functionality

Just do a normal search

Simply wire your autosuggest up to the search endpoint

Use the matcher functionality

The matcher is a fast dictionary lookup of all the terms that are in the corpus. It returns a list that is ordered by frequency in the index (most frequent first).

It accepts an options object that can contain:

var options = {
  beginsWith: "", // The beginning of the text to match
  field: "*",     // The field to use (defaults to all fields)
  threshold: 3,   // The amount of characters to ignore before returning matches (default: 3)
  limit: 10,      // The maximum amount of suggestions
  type: 'simple'  // The type of autosuggest returned- can be `simple`, `ID` or `count`
}

Basic use:

si.match(options)
  .on('data', function (match) {
    // a match
  })
  .end('end', function () {
    // all matches returned
  })

Example:

var options = {
  beginsWith: "lon",
  field: "city",
  threshold: 2
}
si.match(options)
  .on('data', function (match) {
    // a match
  })
  .end('end', function () {
    // all matches returned
  })

...might return:

'london'
'long beach'
'long island'
'lonsdale'

You can also set the type option to ID to return sets of IDs:

var options = {
  beginsWith: "lon",
  field: "city",
  threshold: 2,
  type: 'ID'
}
si.match(options)
  .on('data', function (match) {
    // a match
  })
  .end('end', function () {
    // all matches returned
  })

...which would return a list of suggestions, together with the sets of IDs for the documents that they appear in

[['london', ['1', '3', '4', '10', '12']],
 ['long beach', ['1', '5', '10']],
 ['long island', ['3']],
 ['lonsdale', ['11']]]

Setting the type option to count will return a count for every suggestion where count is a the total number of documents that the term appears in:

var options = {
  beginsWith: "lon",
  field: "city",
  threshold: 2,
  type: 'count'
}
si.match(options)
  .on('data', function (match) {
    // a match
  })
  .end('end', function () {
    // all matches returned
  })

...returns a list of suggestions, together with the sets of IDs for the documents that they appear in

[['london', 5],
 ['long beach', 3],
 ['long island', 1],
 ['lonsdale', 1]]