cadmus

Shell Scripts to Facilitate Effective Note Taking

Introduction

This is a self-contained shell script that uses pre-existing tools (such as TMSU and recoll) to provide an interface for markdown notes.

It's command based and prints out available subcommands for any given command, these means you can use a directory of markdown files like a personal wiki, much like OneNote/Evernote/Notable, for example:

and an overview of what it looks like in the terminal

For example if you wanted to extract all the tags from your markdown notes (either #tags or from the YAML), cadmus has a tool for that:

Ultimately the idea is it is to act a menu to dispatch different scripts I already had so I could more easily share those scripts with classmates.

The real heavy lifting is done by Pandoc, Recoll, ripgrep, skim/fzf, TMSU etc.

Installation

Automatic

Copy this into your shell:

cd $(mktemp -d)
wget https://raw.githubusercontent.com/RyanGreenup/cadmus/master/install.sh
bash install.sh

Cadmus will work from within a self-contained directory and add a symlink to ~/.local/bin
- in this way it's zero lock-in, it does not modify your curr directory of Markdown files.
Installation will automatically create a config file in its directory (which is by default ~/.cadmus
The script will list any dependencies that are not satisfied.

Manual

To install manually:

satisfy the dependencies
Set up Recoll

Download cadmus and put it into the PATH

git clone https://github.com/RyanGreenup/cadmus ~/.cadmus  \
|| echo "Delete $HOME/.cadmus first"
mkdir -p $HOME/.local/bin
ln -s "$HOME/.cadmus/bin/cadmus" "$HOME/.local/bin/"

According to the SystemD Standard ~/.local/bin should be in $PATH, if you are using some other init implementation you can add this directory to "$PATH" it by doing something like this:

## Should work in bash/zsh/fish
echo $PATH | grep "$HOME/.local/bin" &> /dev/null && echo "$HOME/.local/bin in path already" || ls "$HOME/.local/bin" &> /dev/null && echo 'PATH="$PATH:$HOME/.local/bin"' >> $HOME/.profile

When first run, the script will prompt you to make a config file in the directory in which it is run.

Assumptions

It is assumed that:

Notes are:
1. Markdown files with a .md extension
2. Recoll updates it's index on the fly
  - The notes directory will need to be indexed by Recoll in order for the results to show up when using cadmus search.
SSD
- I use an SSD and so I let some scripts be pretty inefficient (for example something like grep | cut | xargs find to avoid creating a variable), I don't know if HDD performance would be great.
All Notes have Unique Names
On MacOS you'll need to define xdg-open and have GNU coreutils, so do something like:

alias xdg-open='open &>/dev/null' 
alias realpath=grealpath &>/dev/null

Configuring recoll

Currently the search just uses the default recoll config, I intend to modify this to use ~/.cadmus as a config directory so as to not interfere with the default config.

It isn't in practice an issue if ~/.recoll is indexing more than the notes because you can just modify the call to Skim (sk) in ..cadmus.. to start the call with ~/Notes/MD.

Design Philosophy

cadmus acts as a menu for scripts to acheive things
- The Actual work will be done by subscripts denoted by description.bash
  - The subscripts will take the note directory as an argument so they are portable and modular
- The Arguments will be shifted and then all passed down to subfunctions
- the script name ~~will~~ should always be printed to the terminal so the individual script can be repurposed with out fishing through code.
Subscripts ~~will~~ should take only one argument (or STDIN)
- If the first argument is either -h or --help help will be printed and then exit 0
- This might lead to some limitations but the simplicity is for sanity, modularity and extensibility.
- Will always return absolute path.
  - I played around with relative path but it got confusing when calling the script from inside a function inside a script, so instead if you want a relative path you should do scriptname './' | xargs realpath --relative-to='./'
Be a Front end to tie together different scripts and tools
Don't replicate work other people have done.
Plain Text, Free as in Speech and Beer.
try to make modular subscripts:
- Pipe in input, output goes to STDOUT
- Leave Aliases and piping to the user
  - See Recommended Aliases

Dependencies

bat
cut
fd
find
fzf
GNU realpath
grep
highlight
jq ^{Arch | brew | ubuntu}
mdcat^AUR
node
Pandoc
perl
python
ranger
recode
Recoll
rg
- Make sure to include pcre2, this comes in Arch, if using cargo:
```
cargo install ripgrep --features 'pcre2' 
```
sd
sed
skim
tmsu^AUR
xclip

Recommended for all Features

ip
- if you're on mac this stackExchange answer suggests iproute2 may work
Kitty
- I've also heard good things about iterm2
MkDocs
- MkDocs Material Extensions
- MkDocs Material Theme
nvim
tectonic
WeasyPrint

Interesting / Helpful / Recommended Generally (Not strictly necessary)

PATH

If any dependencies are installed with pip or cargo it will be necessary to add these directories to your PATH:

## bash
echo '
export PATH="$HOME/.local/bin:$PATH"
export PATH="$HOME/.cargo/bin:$PATH"
' >> ~/.bashrc

## zsh
echo '
export PATH="$HOME/.local/bin:$PATH"
export PATH="$HOME/.cargo/bin:$PATH"
' >> ~/.bashrc

## fish

echo '
set PATH $HOME/.local/bin $PATH
set PATH $HOME/bin $PATH
set PATH "$HOME/.cargo/bin $PATH
' >> ~/.config/fish/config.fish

Recommended Aliases

I wrote all this with aliases in mind, when I settle on some aliases i'll put up my fish functions. (I also wanted autocomplete)

Why / Comparison with other tools

So the boxes I needed ticked are, more or less: 𐄞

	FOSS	Offline	Linux/BSD?	terminal?	RawFiles?	Markdown	AnyEditor?
OneNote	❌	❌	❌	❌	❌	❌	❌
EverNote	❌	?	❌	❌	❌	❌	❌
Notable	❌	✅	✅	❌	✅	✅	✅
Zim	✅	✅	✅	❌	✅	✅	✅
Obsidian	❌	✅	✅	❌	✅	✅	✅
dokuwiki	✅	❌	✅	❌	✅	✅*	✅
joplin	✅	✅	✅	✅	❌	✅	❌ †
mediawiki	✅	❌	✅	❌	❌	❌	❌ ‡
Org-Mode	✅	✅	✅	✅	✅	❌	❌
Cadmus	✅	✅	✅	✅	✅	✅	✅

† You can't open the files from vim with FZF so it gets a no.
‡ Unlike dokuwiki everything is in a database so it gets a no
* With a Plugin

mego22/cadmus