Imagine you have a URL to a podcast episode. You want to download the file, edit the filename in dmenu before saving, automatically add thumbnail and albumart add a notification, maybe convert the audio to mp3. The whole process consist of functions that may or may not be useful for other links, (images, video, torrent links etc). But it gets hard to maintain if all functions are scattered around the filesystem. Not to mention, keeping notifications and menus consistent across the scripts. You also want to have the pattern matching close to the scripts so to speak, but maybe not in the scripts..
gurl tries to solve this by using a special directory for all patterns, commands and common dirthack utilities one creates to handle URLs.
The gurl function itself is just ~30 lines of bash and awk, but it sets a nice stage for managing and creating URL handlers.
If you use Arch Linux you can get gurl from AUR.
Use the make(1) to do a systemwide installation of both
the script and the manpage.
(configure the installation destination in the Makefile, if needed)
$ git clone https://github.com/budlabs/gurl.git
$ cd gurl
# make install
$ gurl -v
gurl - version: 2020.03.26.3
updated: 2020-03-26 by budRich
When gurl is executed it will match it's first
argument, (URL) against rules defined in files named match
located in MATCHERS_DIR (defaults to ~/.config/gurl)
including subdirectories. The syntax for the rules are as
follows:
ACTION: PATTERN
# blank lines and lines starting with sharp (#)
# will be ignored.
ACTION2: PATTERN2
ACTION is a command either in PATH or relative to the
match file.
PATTERN is a awk(1) compatible regular expression.
All arguments passed to gurl will be forwarded to ACTION. If ACTION is a executable file in MATCHERS_DIR, the directory of the ACTION will be the actions working directory, MATCHERS_DIR/_lib will be added to PATH.
If no match is found, ACTION is set to MATCHERS_DIR/default.
gurl will add the following filestructure the first
time it is invoked. No files or directories are mandatory,
but patterns must be declared in files named match . Be
sure to set scripts to be executable. (bash scripts are used
in the example, but any executable files will work).
~/.config/gurl
_lib/
message.sh
ynp.sh
cb.sh
github/
match
repo.sh
ghfile.sh
assets/
vars
youtube/
youtube.sh
.last
match
default
$ gurl https://github.com/budlabs/gurl - this will match
the second pattern in github/match and execute
github/repo.sh. Notice how the path to the action is
relative to match file it is declared in. arguments (the URL
in this case) passed to gurl will get saved to .last,
which can be handy when writing the patterns.
~/.config/gurl/github/match
ghfile.sh: (http[s]?://)?.*github[.]com.*/(blob|raw)/.+
repo.sh: (http[s]?://)?.*github[.]com/[^/]+/[^/]
executables in _lib/ can be executed without specifying
the path. And files (assets/vars) can be sourced using a
path relative to the scriptfile.
~/.config/gurl/github/repo.sh
#!/bin/bash
source assets/vars
[[ $1 =~ (http[s]?://)?.*github[.]com/([^/]+/[^/]+) ]] \
&& repo=${BASH_REMATCH[2]}
prompt="clone $repo to $clone_dir/${repo#*/}? "
[[ $(ynp.sh "$prompt") = yes ]] && message.sh "you answered yes"~/.config/gurl/github/assets/vars
clone_dir=$HOME/git/clones~/.config/gurl/_lib/message.sh
#!/bin/bash
notify-send "$*"~/.config/gurl/_lib/ynp.sh
#!/bin/bash
echo -e "yes\nno" | dmenu -p "$1"~/.config/gurl/_lib/cb.sh
#!/bin/bash
echo -n "$1" | xclip -sel c~/.config/gurl/github/ghfile.sh
#!/bin/bash
message.sh "$1 is a github file, download link in clipboard"
cb.sh "${1/blob/raw}"~/.config/gurl/youtube/youtube.sh
#!/bin/bash
message.sh "a youtube link!"~/.config/.last
https://github.com/budlabs/gurl
~/.config/gurl/default
#!/bin/bash
echo no pattern matching: "$1"~/.config/gurl/match
youtube/youtube.sh: (http[s]?://)?.*youtube[.]com.*
notify-send: (http[s]?://)?.*google[.]com.*
notice how commands can be used for actions (notify-send).
gurl is licensed with the BSD-2-CLAUSE license
See old releasenotes in docs/releasenotes