Warning
You are viewing the technical documentation for Sharetribe. If you are looking for our no-code documentation, see our new help center.

Last updated

How the listing search works

Listings search can be crucial for your marketplace to thrive. Providing people with the right kind of search parameters and filters will help the users to easily find listings relevant to their needs.

Table of Contents

Listings search in Sharetribe

In Sharetribe, listings are searched by using the /listings/query endpoint. Sharetribe has a powerful listing search engine, which can find listings based on multiple criteria:

  • Geolocation. The search can be used to display only listings that are within a provided radius from certain coordinates. Read more about location search below.
  • Free text. The search can be used to find listings that have a certain keyword provided by the user. The search can find from listing title and description. You can also choose to index some public data fields (like listing category) so the search finds from them as well.
  • Price. It's possible to filter out listings with too high or too low price.
    • e.g. show only listings between $20 and $100: sdk.listings.query({ price: "2000,10000" })
  • Availability. It's possible to filter out listings that have insufficient availability during a given time range. Works together with seats so that you can filter for availability with minimum number of seats. It is also possible to filter by required minimum duration of availability for the given time range.
    • e.g. show only listings available for October 22-29, 2022: sdk.listings.query({ availability: "full", start: "Sat Oct 22 2022 01:00:00 GMT+0100", end: "Sat Oct 29 2022 01:00:00 GMT+0100" })
  • Stock. It's possible to show all listings or only listings with positive stock.
    • e.g. show only listings with at least one item in stock: sdk.listings.query({ minStock: 1 })
  • Custom filter: any value. Any number of custom filters can be added. "Any value" filters out listings that don't have the given value (or any of a set of given values) in their public data. A typical use case is filtering by category or subcategory.
    • e.g. show only listings from public data category 'used': sdk.listings.query({ pub_category: "used" })
  • Custom filter: all values. Filters out listings that don't have all values in a given set. A typical use case is choosing among "amenities": an apartment must have both balcony and floor heating in their public data to be displayed in the search.
    • e.g. show only listings that have both balcony and floor heating: sdk.listings.query({ pub_amenities: "has_all:balcony,floorHeating" })
  • Custom numeric filter. Filters out listings that don't have a certain numeric value in their public data. For example, you might want to build a slider filter for the skill level of the user, ranging from 0 to 10.
    • e.g. show only listings for user skill levels 4-7: sdk.listings.query({ pub_userSkillLevel: "4,7" })
  • Sorting. Listing sorting order can be customized per query. Sorting is supported by one or more of: listing price, listing creation time, or any numeric attribute in the listing's public data or metadata.
    • e.g. sort listings for user skill level: sdk.listings.query({ ...params, sort: "pub_userSkillLevel" })

This article approaches the search from two different perspectives: keywords and location. See the API reference documentation for a full list of search parameters. Do note that for all search parameters, you will need to create a search schema so that the data is indexed correctly for search.

The keyword search works so that it matches the keywords parameter to text content of a listing. The listing attributes that are matched in keyword search by default are title and description. Listing public data fields can also be used in the keyword search by defining them to have type text in the marketplace's listing public search schema. All text public data fields with defined schema are indexed for keyword search.

When the listing text content is indexed, in addition to indexing the actual word, it is also broken into subwords, n-grams. For words longer than 3 characters, n-grams of 3 and more characters are constructed from the beginning and end of the word. So if an indexed field contains the word local in addition to the actual word the following n-grams are indexed: loc, loca, ocal, and cal.

The search results are sorted so that a match with an actual word from the listing always weighs more than a match with an n-gram. The order in which matches in different fields of a listing effect the search score is the following: title > description > public data fields. However, the search score differences between the fields are substantially smaller than the difference between a match with an actual word in the listing and an n-gram. Listings that don't match the search keywords at all are not included in the results.

The keywords parameter is a single string that is tokenized on non-alphanumeric characters. Therefore, passing local attractions as a value for the parameter will conduct a search with two keywords: local and attractions and those are then matched against the listing content. A listing will be included in the results, in case any of the keywords match.

In order to take the keyword search into use in the Sharetribe Web Template, you need to change the configuration files.

The listings search provides two parameters to modify the search by location: origin and bounds. origin parameter takes a single location. It does not limit the results in any way but it sorts the results from closest to the furthest from the given point. The origin parameter can not be combined with keywords.

In order to combine location and keyword search, the bounds parameter can be used. It takes north-east and south-west corners of a bounding box that limits the returned results to that area. It does not affect the sorting of results. The bounds can then conveniently be used to match search results to a map view.

The Sharetribe Marketplace API does not have an endpoint for querying users. This is because listings are modeled as the focus of the marketplace. If you do, however, want to implement a search functionality for users, you have a few options.

Users as listings

When modifying your listing creation flow, you can model the listings as service provider profiles. This means that the user entity is distinct from the user's service provider entity. You can then use the default listing search to query the service provider profiles with all the query options described above.

Custom user search endpoint in the template server

The Sharetribe Integration API does have an endpoint for querying users. However, using the Integration API safely requires a trusted context, such as the template application server.

In addition, the Integration API /users/query endpoint returns not only public user information but also non-public information.

This means that if you do want to use the Integration API user query, you need to create a custom server endpoint in the template's server that calls the Integration API endpoint and only returns the user's public information – public data and metadata – back to the browser. Otherwise, you risk revealing sensitive user information to everyone who visits your marketplace site.

That said, the Integration API query has fewer filtering and sorting options than the Marketplace API listing query, so for any level of complexity in your user search we recommended using the "users as listings" approach.