Writerside Help

Algolia search

Algolia is a search-as-a-service API that Writerside supports out of the box.

Create Algolia application and search index

  1. Sign up for an Algolia account.

  2. On the Algolia applications page, click Create Application.

    Configure application parameters, select a region for your datacenter, and create the application.

  3. On the Algolia application page, go to Data sources | Indices and click Create Index.

    Provide an informative name and create the index.

  4. On the index page, go to Configuration | Facets, and click Add an attribute under Attributes for faceting.

    Add two attributes for faceting: product and version. Make sure both are searchable.

  5. Click Review and Save Settings and confirm.

Add Algolia parameters to Writerside build configuration

  1. In your documentation project, open buildprofiles.xml.

  2. Under <variables> , specify the following:

    <algolia-id>

    Algolia application ID

    <algolia-index>

    Algolia index name

    <algolia-api-key>

    Algolia Search-Only API Key

    Show the Algolia logo in search results if you are using their free plan to comply with Algolia's usage terms.

    For example:

    <variables> <algolia-id>TGW4GIW73X</algolia-id> <algolia-index>index_name</algolia-index> <algolia-api-key>29effd68f91893202311a653263f1fb8</algolia-api-key> <algolia-show-logo>true</algolia-show-logo> </variables>
  3. Commit and push these changes, then build and publish the help instance with these configuration parameters.

Now your published help instance will have a search field that will send queries to the specified Algolia index using the provided API key.

Upload your search index records to Algolia

To reply to search requests from your help, Algolia needs records in the specified search index. The Writerside builder collects them while building the help website as a separate archive named algolia-indexes-HI.zip or whatever your instance ID was instead of HI.

This archive contains JSON files, each with a JSON object representing a separate search index record. Usually, there is one record per topic, but some bigger topics can be split into multiple records. Although Algolia lets you upload these JSON files with records to the search index manually, there are usually a lot of them, and you probably want to automate it anyway. The best approach is to add a step to your build and publication process that will upload the JSON files from the archive.

  1. If you do not yet have a TeamCity configuration for building your help artifacts, set it up as described in Build on TeamCity Cloud.

  2. Create a separate build configuration for publishing search indexes to your Algolia application.

    Configure it to use the following Docker image:

    registry.jetbrains.team/p/writerside/builder/algolia-publisher:2.0.32-3

    This image will run the help-publication-agent script.

  3. Create a Command line type step with the following command:

    env "algolia-key=${'$'}ALGOLIA_KEY" java -jar /opt/builder/help-publication-agent.jar \ update-index \ --application-name ${'$'}ALGOLIA_APP_NAME \ --index-name ${'$'}ALGOLIA_INDEX_NAME \ --product ${'$'}CONFIG_JSON_PRODUCT \ --version ${'$'}CONFIG_JSON_VERSION \ --index-directory algolia-index/ \ 2>&1 | tee algolia-update-index-log.txt
  4. In General Settings, in the Artifact path field, add algolia-update-index-log.txt to write logs.

  5. In Dependencies, add artifacts produced by your help build configuration: webhelp.zip and algolia-indexes-*.zip!** => algolia-index. This action unpacks the ZIP archive with indexes to algolia-index.

  6. In Parameters, add the following environment variables:

    env.ALGOLIA_APP_NAME

    Algolia application ID

    env.ALGOLIA_INDEX_NAME

    Name of Algolia index

    env.ALGOLIA_KEY

    Algolia Admin API Key.

    This is a private key. Set the type of this parameter to Password, so that users without proper permissions will not see its value.

    env.CONFIG_JSON_PRODUCT

    Help instance ID from the tree file or the value of the web-path attribute specified in writerside.cfg if it is different from the ID

    env.CONFIG_JSON_VERSION

    Help instance version (usually the same as the branch name) specified in writerside.cfg

  7. Save the build configuration settings and click Deploy. After it finishes successfully, go to the index in your Algolia account and make sure that the index records have really been uploaded.

  1. In your repository on GitHub, go to Settings | Secrets and variables | Actions, and add a New repository secret with the name ALGOLIA_KEY and the value of the Admin API Key from your Algolia account.

  2. Set up a GitHub Actions workflow that builds your help artifacts as described in Build and publish on GitHub. More specifically, make sure that you add a publish-indexes job to the workflow as described in Upload search indexes.

After you run this workflow successfully and the publish-indexes job finishes, go to the index in your Algolia account and make sure that the index records have really been uploaded.

  1. In your repository on GitLab, go to Settings | CI/CD | Variables, and add a new CI/CD Variable called ALGOLIA_KEY with the Algolia Admin API Key from your Algolia account.

  2. Set up a GitLab CI build pipeline that builds your help artifacts as described in Build and publish on GitLab. More specifically, make sure that you add a publish-indexes job to the workflow as described in Upload search indexes.

After you run this build pipeline successfully and the publish-indexes stage finishes, go to the index in your Algolia account and make sure that the index records have really been uploaded.

After you upload the index records to Algolia, you can check that search works in your published help. You will need to upload new search index records every time you publish a new help build.

Algolia crawler

Algolia provides a hosted crawler that extracts content from your website, transforms it, and uploads as search index records to your Algolia account. You can use it instead of uploading the index records collected by the builder if this is more convenient for you, but do not use both methods at the same time.

Create crawler

  1. On the Algolia application page, go to Data sources | Crawler and click Add new crawler.

  2. Configure the crawler using the web editor.

    Implement a function that will extract an array of JSON objects and return it as the value of the recordExtractor parameter.

    Here is a simple function that creates an array of index records:

    recordExtractor: ({ $, url }) => { const records = []; const mainTitle = $("h1").text().trim(); const headers = "h1, h2, h3, h4, h5, h6"; $(headers).each((i, h) => { var urlString = url.href; const anchor = $(h).attr("id"); if (anchor) { urlString += "#" + anchor; } const content = $(h).nextUntil(headers).text().trim(); records.push({ url: urlString, mainTitle: mainTitle, pageTitle: $(h).text().trim(), content: content, product: "hi" }); }); return records; }

    The function iterates over headers on a page and creates a separate index record for each one with the following fields:

    url

    The URL where the index record should point.

    mainTitle

    The main title on the page.

    pageTitle

    The header of this specific record, like a chapter title.

    content

    The main content of the search index record.

    product

    The ID of the published instance. For example, the ID of the default Help Instance is hi. The index should have the product attribute for faceting. For information about adding facets to an index, see Create Algolia application and search index.

  3. Run the crawler to collect index records.

  4. If the published instance had Algolia parameters pointing to the correct Algolia application, index, and API key when it was built, you should now be able to search against the index from your published help instance.

Last modified: 31 January 2025