FR → LISEZMOI.md

Demo

??

Presentation

signet.sh

Signet.sh is a shell script parsing a bookmark database to a webpage. It uses awk inside a here-doc declaration redirected to an html file.

I have created this script because I found the bookmarks manager from firefox not great to use, to the point where I did not have the habit to bookmark my browsing. Firefox stores bookmarks in a sqlite format and I wanted something text based

Bookmarks database format

Only a Url is required, the rest of the fields are optional : URL : ... Name : grabs from the bookmarked page Description : ... Tags : comma separated keywords Date : posix time of the bookmarked link Color : css color (name, hex, rgb etc...)

For example, in a file called BOOKMARKS by default:

URL: https://rosettacode.org/wiki/Rosetta_Code
Name: Rosetta Code
Description: 
Tags: literacy, read
Date: 1704675057
Color: Purple

URL: https://www.emigre.com/TypeSpecimens
Name: Emigre: Type Specimens
Description: 
Tags: emigre, type
Date: 1704680644

URL: https://web.archive.org/web/20211025182257/http://len.falken.ink/
Name: Wayback Machine
Description: 
Tags: read
Date: 1704712747

Other bookmark managers:

• nb https://xwmx.github.io/nb/ (AGPL-3.0)
• ??

Dependency :

To add a link via the proposed interface, we will need to install dmenu and htmlq.

• dmenu https://tools.suckless.org/dmenu/ (MIT/X)
• htmlq https://github.com/mgdm/htmlq (MIT) replaced by one awk command

dmenu

Dmenu is an interactive menu that allows us to select and write values in a menu. These values can come from a program provided as input from a pipe "|", for example:

ls | dmenu

displays a drop-down menu with the files in my directory. In our script, to store the given tags in a variable, we can do:

tags=$(echo "" | dmenu -p "Enter comma-separated tags:")

htmlq

Htmlq is an HTML parser written in Go. It doesn't matter which parser we uses, it seems that each language has its own. We give to the program an html string and it filters through it using css selectors returning the found html elements Using javascript, retrieving all the <h1> children of a <section> can be done with :

document.querySelectorAll('section h1')

In shell, it's more complicated, htmlq is made for that:

cat fichier.html | htmlq 'section h1'

And to output only the text -- the javascript equivalent of .innerHTML:

cat fichier.html | htmlq 'section h1' --text

We use it to retrieve the title tag of the page to add:

curl page.html | htmlq 'title' --text

Pre-release patch:

curl $url | awk -v RS='</title>' \
    '/<title>/ {gsub(/.*<title>/, ""); print}' |\
    tr -d '\n'

For the moment this step is blocking. In case internet cuts while bookmarking a link, you will have to wait for the end of the curl attempt to move to the next field in the script :/ sorry!

Add a bookmark :

To add a link, I select the URL of the page with Ctrl + l, the copy and run add.sh with Super + i. If anyone knows how to retrieve the url of the current browser page without having to copy it, this would save two steps out of three. En résumé : Ctrl + l, Ctrl + c, Super + i (ou plus court : Ctrl + l + i, Super + i)

Super + i because in my window manager configuration file located in ~/.config/awesome/rc.lua, I have the following lines:

awful.key({modkey}, "i", function()
awful.util.spawn_with_shell("add.sh") end, 
{description = "Enregistre le lien copié dans signet.sh"}),

Structure du répertoire

• BOOKMARKS → A textual database of bookmarks
• add.sh → A script to add a link to the database - dmenu - ./signet.sh
• bookmark.sh → The shell script itself - It generates a new html page from the link database (default index.html)
• style.css → Stylesheet for index.html
• script.js → A bit of javascript for: - search in the <textarea> - add background colors to entries that have them - if the description field is empty, do not display it - format posix time dates to YY-MM-DD format /!\ Soon /!
  starred.sh → curl from the github user api and format it in the same way as signet.sh without using jq https://api.github.com/users/[user]/starred

Credits

Junicode (OFL-1.1) https://psb1558.github.io/Junicode-font/ Dmenu License (MIT/X) https://tools.suckless.org/dmenu/