just-the-docs/docs/search.md
Dale Phurrough 7a01ef1a59
add options for custom lunr Liquid and JS code (#1068)
This is a prototype for review and discussion. My use and testing of this PR is on top of 6d9d41359c46882d9b64a446d5a83fac5b3e20a7. The changes are trival to rebase to `main` and I'm happy to do so if this prototype moves forward.


* Feature request details in linked issue, fixes just-the-docs/just-the-docs#1067
* I welcome feedback and all discussion
* A draft doc site of mine using this PR is at https://docs.hidale.com/

To use the prototype, the two include files need to be customized. Here are mine from the draft website 9c0d836408

Co-authored-by: Matt Wang <matt@matthewwang.me>
2023-01-13 19:08:33 -08:00

162 lines
5.4 KiB
Markdown

---
layout: default
title: Search
nav_order: 7
---
# Search
{: .no_toc }
## Table of contents
{: .no_toc .text-delta }
1. TOC
{:toc}
---
Just the Docs uses [lunr.js](http://lunrjs.com) to add a client-side search interface powered by a JSON index that Jekyll generates.
All search results are shown in an auto-complete style interface (there is no search results page).
By default, all generated HTML pages are indexed using the following data points:
- Page title
- Page content
- Page URL
## Enable search in configuration
In your site's `_config.yml`, enable search:
```yaml
# Enable or disable the site search
# Supports true (default) or false
search_enabled: true
```
### Search granularity
Pages are split into sections that can be searched individually.
The sections are defined by the headings on the page.
Each section is displayed in a separate search result.
```yaml
# Split pages into sections that can be searched individually
# Supports 1 - 6, default: 2
search.heading_level: 2
```
### Search previews
A search result can contain previews that show where the search words are found in the specific section.
```yaml
# Maximum amount of previews per search result
# Default: 3
search.previews: 3
# Maximum amount of words to display before a matched word in the preview
# Default: 5
search.preview_words_before: 5
# Maximum amount of words to display after a matched word in the preview
# Default: 10
search.preview_words_after: 10
```
### Search tokenizer
The default is for hyphens to separate tokens in search terms:
`gem-based` is equivalent to `gem based`, matching either word.
To allow search for hyphenated words:
```yaml
# Set the search token separator
# Default: /[\s\-/]+/
# Example: enable support for hyphenated search words
search.tokenizer_separator: /[\s/]+/
```
### Display URL in search results
```yaml
# Display the relative url in search results
# Supports true (default) or false
search.rel_url: false
```
### Display search button
The search button displays in the bottom right corner of the screen and triggers the search input when clicked.
```yaml
# Enable or disable the search button that appears in the bottom right corner of every page
# Supports true or false (default)
search.button: true
```
## Hiding pages from search
Sometimes you might have a page that you don't want to be indexed for the search nor to show up in search results, e.g, a 404 page.
To exclude a page from search, add the `search_exclude: true` parameter to the page's YAML front matter:
#### Example
{: .no_toc }
```yaml
---
layout: default
title: Page not found
nav_exclude: true
search_exclude: true
---
```
## Generate search index when used as a gem
If you use Just the Docs as a remote theme, you do not need the following steps.
If you use the theme as a gem, you must initialize the search by running this `rake` command that comes with `just-the-docs`:
```bash
$ bundle exec just-the-docs rake search:init
```
This command creates the `assets/js/zzzz-search-data.json` file that Jekyll uses to create your search index.
Alternatively, you can create the file manually with [this content]({{ site.github.repository_url }}/blob/main/assets/js/zzzz-search-data.json).
## Custom content for search index
By default, the search feature indexes a page's `.content`, `.title`, and *some* headers within the `.content`.
Other data (ex front matter, files in `_data`, `assets`) is not indexed. To index additional data, users can customize what `lunr` indexes.
{: .warning }
> Customizing search indices is an advanced feature that requires Javascript and Liquid knowledge.
1. First, ensure that `assets/js/zzzz-search-data.json` is up-to-date; it can be regenerated with `rake` or manually (see: ["Generate search index when used as a gem"](#generate-search-index-when-used-as-a-gem)).
2. To add Liquid/Jekyll-based data: create a new include at the path `_includes/lunr/custom-data.json`. Insert custom Liquid code that reads various data (ex: `include.page`, `site.data`, `site.static_files`) that then generates valid [JSON](https://www.json.org/json-en.html) to add to the index. Verify the fields in the generated `assets/js/search-data.json`.
3. For all custom data (Liquid, JavaScript, or external): create a new include at the path `_includes/lunr/custom-index.js` to your site. Add valid JavaScript that creates relevant fields to add to the index. You may want to inspect `assets/js/just-the-docs.js` to better understand the code structure. **This is necessary to render any relevant custom index code.**
#### Example
`_includes/lunr/custom-data.json`: this example adds each page's `usage` and `examples` front matter fields, normalizes the text, and writes the text to custom Javascript `myusage` and `myexamples` fields.
{% raw %}
```liquid
{%- capture newline %}
{% endcapture -%}
"myusage": {{ include.page.usage | markdownify | replace:newline,' ' | strip_html | normalize_whitespace | strip | jsonify }},
"myexamples": {{ include.page.examples | markdownify | replace:newline,' ' | strip_html | normalize_whitespace | strip | jsonify }},
```
{% endraw %}
`_includes/lunr/custom-index.js` custom code is within a Javascript loop. All custom
Javascript fields are accessed as fields of `docs[i]` such as `docs[i].myusage`.
Finally, append your custom fields on to the already existing `docs[i].content`.
```javascript
const content_to_merge = [docs[i].content, docs[i].myusage, docs[i].myexamples];
docs[i].content = content_to_merge.join(' ');
```