nbf

Tiny, dependency-free, CLI tool for working with the Netscape Bookmark File format.

Its only function is to take a JSON object from stdin and output HTML to stdout.

Installation

Install globally with npm or yarn:

# npm:
npm install -g nbf

# yarn:
yarn global add nbf

Or run it directly with npx:

npx nbf

JSON format

nbf expects JSON items to have the following properties:

{
	/*
		Required bookmark URL.
	 */
	"uri": "https://...",

	/*
		Optional bookmark title.
	 */
	"title": "...",

	/*
		Optional bookmark description.
	 */
	"description": "...",
	
	/*
		Optional bookmark date in any format 
		parsable by JavaScript's Date() constructor. 
		When missing, the bookmark will be tagged 
		with the current timestamp.
	 */
	"dateAdded": "...",
	
	/*
		Optional tags, either as an array 
		or a comma-separated string.
	 */
	"tags": ["...", "..."],
	"tags": "..., ..., ..."
}

Folders and subfolders are possible:

{
	/*
		Folder title.
	 */
	"title": "...",

	/*
		Nested bookmarks.
	 */
	"children": [
		// ...
	]
}

Guide to exporting bookmarks

Most of these recipes use:

• curl
• jq (tutorial)
• hred

Jump to:

• Mastodon
• NetNewsWire starred articles
• GitHub starred repos
• Lobste.rs
• Safari bookmarks

Mastodon

Grab an archive of your Mastodon account with mastodon-archive.

In the example below, we use jq to look at favourites that have a card attached to them, and reshape the JSON to fit our schema:

cat mastodon.social.user.danburzo.json | \
jq '[
		.favourites[] | 
		select(.card) | 
		{ 
			dateAdded: .created_at, 
			uri: .card.url, 
			title: .card.title, 
			description: "\(.card.description)\nvia: \(.url)",
			tags: ["from:mastodon"] 
		}
	]' | \
nbf > mastodon-faves.html

NetNewsWire starred articles

On macOS, NetNewsWire keeps user data in a SQLite database. We can browse and query it, and grab a JSON of the result, using datasette:

datasette serve ~/Library/Containers/com.ranchero.NetNewsWire-Evergreen/Data/Library/Application\ Support/NetNewsWire/Accounts/OnMyMac/DB.sqlite3

(Older versions of NNW store their database under ~/Library/Application\ Support/NetNewsWire/Accounts/OnMyMac/DB.sqlite3)

Head over to http://127.0.0.1:8001/DB and run this query:

select 
  a.title as title, 
  a.summary as description, 
  coalesce(a.url, a.externalURL) as uri,
  a.datePublished * 1000 as dateAdded
from articles as a join statuses as s 
on a.articleID = s.articleID 
where s.starred = 1;

Then follow the json link, and add the _shape=array query parameter — this shapes the JSON in a way that we can use directly. We can curl -nS it in our command:

curl -nS http://127.0.0.1:8001/DB.json?_shape=array&sql=select+%0D%0A++a.title+as+title%2C+%0D%0A++a.summary+as+description%2C+%0D%0A++coalesce(a.url%2C+a.externalURL)+as+uri%2C%0D%0A++a.datePublished+*+1000+as+dateAdded%0D%0Afrom+articles+as+a+join+statuses+as+s+%0D%0Aon+a.articleID+%3D+s.articleID+%0D%0Awhere+s.starred+%3D+1%3B | \
nbf > nnw.html

GitHub starred repos

GitHub CLI (currently in beta) makes it easy to collate paginated responses from the GitHub API.

gh api user/starred \
	-H "Accept: application/vnd.github.v3.star+json" \
	-H "Accept: application/vnd.github.mercy-preview+json" \
	--paginate \
| jq '
        .[] | select(.repo.private == false) | {
                title: .repo.full_name,
                uri: .repo.html_url,
                dateAdded: .starred_at,
                description: "\(.repo.description)\n\(.repo.homepage // "")",
                tags: ["source:github"]
}' \
| jq -s '.' \
| nbf > stars.html

If you also want to include repository topics and/or language as tags — but be aware that people can go overboard with topics in an effort to make their repos more discoverable:

tags: (
    (.repo.topics // []) +
    [.repo.language // ""] +
    ["source:github"]
) | map(. | ascii_downcase) | unique

If afterwards you want to unstar the repos in bulk, use:

gh api user/starred | \
jq -r '.[] | "user/starred/\(.full_name)"' | \
xargs -L1 gh api --method=DELETE

Variations:

• plop an --interactive flag for xargs to confirm each unstar with the y key + Enter. (Is there a way to get Yes by default? 🤔)
• use the --paginate flag on gh api to go through all your starred repos, but that might get you rate-limited.

Lobste.rs

Lobste.rs offers a JSON endpoint for most things, for example lobste.rs/saved.json. Because you need the session cookie to make this request from the command-line, we go to the browser's dev tools, right-click the select and choose Copy as cURL.

curl ... | jq '[.[] | {
	title: .title,
	description: "\(.description)\nvia: \(.short_id_url)",
	uri: .url,
	tags: ((.tags // []) + ["source:lobste.rs"]),
	dateAdded: .created_at
}]' | nbf

Safari Bookmarks

Safari offers an option to Export Bookmarks... which is already in the Netscape Bookmark Format. However, we may want to run the bookmarks through nbf to:

• add source:safari as a tag;
• give nbf a chance to timestamp each bookmark.

We can extract a JSON from the Safari Bookmarks.html file with hred, then using jq to add the tags property to each object:

cat ~d/Safari\ Bookmarks.html \
| hred "a { @href => uri, @.textContent => title }" \
| jq '[ .[] | .tags = ["source:safari"] ]' \
| nbf > safari-tagged.html