[Feature Request]: Tags

[reakaleek]

Context

The legacy asciidocs build supported tags.
In the conf.yaml, you can see these tags applied at a book level.

These tags were also used to enhance the search experience, making it possible to filter by tags.

Goal

Add a tags feature that can be used by the search feature for filtering.

Requirements

• The user should be able to set one or more tags for a page.
• The user should be able to set a default tag for reference content.
  E.g. the logstash reference should have the "Logstash" tag by default
• The user should be able to append tags on a page, in addition to the default tag.
  Hence, if a default tag exists and the user adds tags to a page, it should contain both the default tag and the additional tags.
• Linting: Make sure we reuse existing tags.

Proposed implementation

1. Add a tags frontmatter option; these tags should then be added to HTML meta tags so that the web team can crawl and parse them.
2. Write a script to parse tags from conf.yaml and add it to frontmatter based on the prefix attribute in conf.yaml and mapped_pages frontmatter in elastic/docs-content. This means a page might have multiple tags.
3. Do a release so we can already utilize the tags for search.
4. Add a "default" tag feature for reference content. (Can we add it somehow to assembler.yml or navigation.yml?

Open Questions

• What format should the meta tags be so that the web team can parse them
• Can we add additional metadata for the search? E.g., content type, license, or version? AFAIU, right now it's only for "Products" filtering
• What's the best way to add default tags for reference content? docset.yml, toc.yml, assembler.yml or navigation.yml?

References

• https://search.elastic.co/?q=test&location%5B0%5D=Documentation
• https://github.com/elastic/docs/blob/master/conf.yaml#L185
• https://github.com/elastic/docs-content/blob/441bb97aad23441f057a8529984b252f6405b18c/deploy-manage/tools/snapshot-and-restore/ec-gcs-snapshotting.md?plain=1#L4

Stakeholders

@elastic/docs-engineering, @KOTungseth, @colleenmcginnis, @elastic/webteam

[acsnyder]

There are a few meta tags the search crawler relies on for extracting data from pages; but are missing on the new docs site. The following meta tags have dummy data to illustrate what kind of data is needed.

<meta class="elastic" name="product_version" content="8.17"/>
<meta class="elastic" name="product_name" content="Elasticsearch"/>
<meta class="elastic" name="website_area" content="documentation"/>
<meta name="DC.subject" content="Elasticsearch"/>

The DC.subject meta tag may be duplicative of product_name, we can adjust the crawler if necessary. I created a ticket last year but it might've been added to a deprecated repository.

[zumwalt]

++ to @acsnyder's comment.

Can we add additional metadata for the search? E.g. content type, license, or version?

Yes, we can support any data that you'd like to filter/facet on assuming that we can reliably extract it for all documents

[reakaleek]

Thank you for the quick response. @acsnyder In the new Information Architecture (IA), one page could apply to multiple products.

How would we handle this case?

[acsnyder]

@reakaleek We could have multiple values separated by some kind of delimiter. We'd have to update the crawler script for that. Can you give an example of a page used for multiple products?

[reakaleek]

@KOTungseth, can you please help me find an example page?

[KOTungseth]

@KOTungseth, can you please help me find an example page?

Sure can!

• Cloud Native Vulnerability Management Dashboard (source file)
• Deploy (source file)
• Elastic Cloud on Kubernetes release notes (source file)

[acsnyder]

Thanks @KOTungseth! The second item (Deploy) is the same link as the source file.

Looking at your mapped_pages frontmatter, it looks like we may need to rethink how we crawl docs pages. I'll comment later when I have some time to check locally.

[bmorelli25]

Another requirement:

• Use CI to enforce only product tags listed in https://github.com/elastic/docs-content/blob/add-product-tags/frontmatter.config.yml are used

[reakaleek]

@colleenmcginnis provided a list of product names and their kebab-case mapping value.

mapping

# I wasn't sure where to put this...
# We can delete it from here if we want to move it to docs-builder.
apm: 'APM'
apm-dotnet-agent: 'APM .NET Agent'
apm-android-agent: 'APM Android Agent'
apm-attacher: 'APM Attacher'
apm-aws-lambda-extension: 'APM AWS Lambda extension'
apm-go-agent: 'APM Go Agent'
apm-ios-agent: 'APM iOS Agent'
apm-java-agent: 'APM Java Agent'
apm-node-agent: 'APM Node.js Agent'
apm-php-agent: 'APM PHP Agent'
apm-python-agent: 'APM Python Agent'
apm-ruby-agent: 'APM Ruby Agent'
apm-rum-agent: 'APM RUM Agent'
beats-logging-plugin: 'Beats Logging plugin'
cloud-control-ecctl: 'Cloud Control ECCTL'
cloud-enterprise: 'Cloud Enterprise'
cloud-hosted: 'Cloud Hosted'
cloud-kubernetes: 'Cloud Kubernetes'
cloud-native-ingest: 'Cloud Native Ingest'
cloud-serverless: 'Cloud Serverless'
cloud-terraform: 'Cloud Terraform'
ecs-logging: 'ECS Logging'
ecs-logging-dotnet: 'ECS Logging .NET'
ecs-logging-go-logrus: 'ECS Logging Go Logrus'
ecs-logging-go-zap: 'ECS Logging Go Zap'
ecs-logging-go-zerolog: 'ECS Logging Go Zerolog'
ecs-logging-java: 'ECS Logging Java'
ecs-logging-node: 'ECS Logging Node.js'
ecs-logging-php: 'ECS Logging PHP'
ecs-logging-python: 'ECS Logging Python'
ecs-logging-ruby: 'ECS Logging Ruby'
elastic-agent: 'Elastic Agent'
ecs: 'Elastic Common Schema (ECS)'
elastic-products-platform: 'Elastic Products platform'
elastic-stack: 'Elastic Stack'
elasticsearch: 'Elasticsearch'
elasticsearch-dotnet-client: 'Elasticsearch .NET Client'
elasticsearch-apache-hadoop: 'Elasticsearch Apache Hadoop'
elasticsearch-cloud-hosted-heroku: 'Elasticsearch Cloud Hosted Heroku'
elasticsearch-community-clients: 'Elasticsearch community clients'
elasticsearch-curator: 'Elasticsearch Curator'
elasticsearch-eland-python-client: 'Elasticsearch Eland Python Client'
elasticsearch-go-client: 'Elasticsearch Go Client'
elasticsearch-groovy-client: 'Elasticsearch Groovy Client'
elasticsearch-java-client: 'Elasticsearch Java Client'
elasticsearch-java-script-client: 'Elasticsearch JavaScript Client'
elasticsearch-painless-scripting-language: 'Elasticsearch Painless scripting language'
elasticsearch-perl-client: 'Elasticsearch Perl Client'
elasticsearch-php-client: 'Elasticsearch PHP Client'
elasticsearch-plugins: 'Elasticsearch plugins'
elasticsearch-python-client: 'Elasticsearch Python Client'
elasticsearch-resiliency-status: 'Elasticsearch Resiliency Status'
elasticsearch-ruby-client: 'Elasticsearch Ruby Client'
elasticsearch-rust-client: 'Elasticsearch Rust Client'
fleet: 'Fleet'
ingest: 'Ingest'
integrations: 'Integrations'
kibana: 'Kibana'
logstash: 'Logstash'
machine-learning: 'Machine Learning'
observability: 'Observability'
reference-architectures: 'Reference Architectures'
search-ui: 'Search UI'
security: 'Security'

This list is incomplete and lacks products for some reference content.
But it's a complete list for docs-content.

[lcawl]

It is interesting to ponder whether users will eventually want to filter on content types within the "documentation" area (even if it's just "reference" and "guide" for now). I am mentioning this because we'll also be striving to add matching meta tags in the pages we're hosting on https://www.elastic.co/docs/api/ and it begs the question of whether we want to enable folks to filter those in or out of the search results (in which case it could potentially be nested values like this: "documentation" > "reference" > "API")?

[theletterf]

Is this the desired form of the product frontmatter attribute?

products:
  - cloud-hosted
  - cloud-serverless

The mapping list is missing values for EDOT distributions. I'm providing one here following the convention established in the previous comments:

Additional mappings for EDOT

edot-collector: 'Elastic Distribution of OpenTelemetry Collector'
edot-java: 'Elastic Distribution of OpenTelemetry Java'
edot-dotnet: 'Elastic Distribution of OpenTelemetry .NET'
edot-nodejs: 'Elastic Distribution of OpenTelemetry Node.js'
edot-php: 'Elastic Distribution of OpenTelemetry PHP'
edot-python: 'Elastic Distribution of OpenTelemetry Python'
edot-android: 'Elastic Distribution of OpenTelemetry Android'
edot-ios: 'Elastic Distribution of OpenTelemetry iOS'