Gooseberry - a Knowledge Base for the Lazy

WIP

Gooseberry provides a command-line interface for Hypothesis (a tool to annotate the web) and lets you generate a knowledge-base wiki without you having to actually type your knowledge out.

Installation requirements

• A Hypothesis account, and a personal API token obtained as described here.
• bat to display highlighted markdown in the terminal.

Binaries

See the releases

• OSX - allow gooseberry via System Preferences (necessary in Catalina at least)
• Linux - chmod +x gooseberry
• Currently, doesn't work on Windows (waiting on this issue)

Contributing

See CONTRIBUTING.md for an in-depth explanation of how Gooseberry works and what could be improved.

Motivation

So yes, knowledge-base tools are old hat and a dime a dozen, and we really have no excuse to not have a nice big tangled folder of markdown files filled with our overflowing wisdom. But after spending all day writing code and papers and tasks, it just isn't fair that our reading time also needs to be typing time to get all that knowledge down. And if we don't type things out our fancy knowledge-base is bare, empty, and sad.

In comes Gooseberry - a tool to build a knowledge base from highlighting and annotating passages while reading news articles, blog posts, papers, what have you. Gooseberry combines the ease of annotation offered by Hypothesis, bulk tagging and organization support in the command line, and a customizable plaintext wiki with HandleBars templating.

A Typical Workflow

1. Find an article, blog post, paper etc. to read.
2. Highlight lines and facts you'd like to remember later. You can add comments and tags already if you're up for it but the focus can also be just on reading and highlighting without thinking too much about taking notes.
3. More often than not, when one gets into a topic it ends in 50 open tabs of subtopics. This is fine, keep reading and highlighting away, we'll get back to this.
4. Finally, once your thirst for knowledge has been fulfilled, fire up a terminal and run
   • gooseberry sync to download all your latest highlights and annotations.
   • gooseberry tag --from "9a.m." topic to tag everything you've read this morning with the topic you were looking into. This subcommand is super flexible. You can tag something by a website, so that all annotations from subtopic B's wikipedia page are tagged as B for instance. Or just open up search to search your annotations and add tags to everything matching a search query (or remove tags and annotations). Tags are very nestable, definitely make use of this - e.g. all annotations today may be about topic A, five of them are also subtopic B etc.
   • gooseberry make to add all this new tagged information to your knowledge base.

Here's an example. Today I read and annotated three articles about insects: this Nautilus article titled "We need insects more than they need us", this publication about honey bees and pesticides, and an Atlantic article about the "anternet" .

I synced and tagged these annotations:

Then ran gooseberry make to make an mdBook style wiki which I could then open in the browser:

Or an Obsidian style wiki, with annotations grouped into folders based on their base URI

Annotation text is just markdown so text formatting, LaTeX, pictures etc. goes too!

The annotation template is configurable, as is the folder and grouping structure. Each annotation can link back to the position in the website that you got it from, if ever you feel like you're missing context.

Some advantages

• You barely have to type while reading unless you're in the mood for taking notes.
• If you're in the mood, the note-taking won't involve window switching.
• Even without using the wiki functionality you end up with a CLI to quickly tag your Hypothesis annotations.
• Even without using the tagging functionality you end up with a pretty cool wiki listing all your annotations.
• Since it's just plaintext, and the template can be customized, you can integrate it with any knowledge base system accepting plaintext files (like Obsidian, mdBook, org-mode, etc.)

Customization

Knowledge base customization

You can set all the below options at once by running gooseberry config kb all or changing the corresponding keys in the config file (found at gooseberry config location)

gooseberry config kb annotation - Annotation template

Change the template used for rendering the annotation.

The following keys can be used inside the template

• {{ id }} - Annotation ID
• created - Date of creation. Use with the date_format helper
• updated - Date of creation. Use with the date_format helper
• {{ user }} - User account ID formatted as acct:<username>@<authority>
• {{ uri }} - Annotation URI
• {{ base_uri }} - Base website of URI, i.e just the protocol and domain.
  • e.g. https://github.com/rust-lang/cargo?asdf becomes https://github.com/
• {{ incontext }} - Link to annotation in context
• highlight - List of selected/highlighted lines from document (split by newline)
• {{ text }} - The text content of the annotation body
• tags - A list of tags associated with the annotation.
• {{ group }} - ID of Hypothesis group,
• references - List of annotation IDs for any annotations this annotation references (e.g. is a reply to)
• {{ display_name }} - Display name of annotation creator. This may not be set.

See the Handlebars Language Guide for more on templating. Some examples for using the list type keys and for formatting dates are shown below for different systems:

• mdBook

##### {{date_format "%c" (created)}} - *{{id}}*

{{#each tags}}| {{this}} {{#if @last}}|{{/if}}{{/each}}

{{#each highlight}}> {{this}}{{/each}}

{{text}}

[See in context]({{incontext}})

Renders as:

##### Sat Jan 16 11:12:49 2021 - *test*

| tag1 | tag2 |

> exact text highlighted in website

testing annotation

[See in context](https://incontext_link.com)

This makes each tag a link to a dedicated page consisting of annotations with that tag - you can set this up below by configuring the hierarchy.

• Obsidian

### {{id}}

Created: {{date_format "%c" (created)}} Tags: {{#each tags}}#{{this}}{{#unless @last}}, {{/unless}}{{/each}}

{{#each highlight}}> {{this}}{{/each}}

{{text}}

[See in context]({{incontext}})

Renders as:

### test

Created: Sat Jan 16 10:22:20 2021 Tags: #tag1, #tag2

> exact text highlighted in website

testing annotation

This uses #tags b/c Obsidian likes those.

TODO add org-mode example

gooseberry config kb link - Index link template

This configures the index file, which generally contains links to all other pages in the generated knowledge base. The template controls how each of these links are rendered.

Available keys:

• {{ name }} - file stem
• {{ relative_path }} - path relative to KB directory
• {{ absolute_path }} - full path on filesystem

Examples:

• mdBook

- [{{name}}]({{relative_path}})

• Obsidian

- [[{{name}}]]

to make internal links, or

- ![[{{name}}]]

to transclude files

• Org-mode

- [[{{relative_path}}][{{name}}]]

Index filename - gooseberry config kb index

The name of the Index file. For instance, mdBook needs this to be called SUMMARY.

File extensions - gooseberry config kb extension

e.g. "md", "org", "txt" etc. (Don't include the .)

Grouping annotations into folders and pages - gooseberry config kb hierarchy

The hierarchy defines how the folder structure of the knowledge base looks and which annotations are on what pages. This is useful for KB systems where folders correspond to topics.

The available options are:

• empty - Set hierarchy = ["empty"] to have all annotations rendered on the index page.
• tag - Groups annotations by tag
• uri - Groups annotations by their URI
• base_uri - Groups annotations by their base URI
• id - Groups annotations by annotation ID.

Multiple hierarchies make folders and sub-folders.

e.g.

hierarchy = ["base_uri", "tag"] 

would make a separate folder for each base_uri. Within each folder would be a page for each tag consisting of annotations marked with that tag.

hierarchy = ["tag"]

gives the structure in the mdBook figure above, i.e. a page for each tag.

Annotations within a page are sorted by their date of creation.

Why "Gooseberry"?

Because Discworld will never let me down when it comes to names: Dis-organizer Mark 5, the Gooseberry