Are you tired of your favourite radio shows still not providing a podcast service? Is your favourite music only played at nighttime? radioman allows you to record just about any radio webstream that syndicates mp3 streams either directly or via m3u or pls playlists (which should be just about every station) according to a predefined schedule.
- Schedule string syntax allows for the most complicated recording schedules.
- Recovers from connection losses gracefully without interupting the recording.
- Easy systemd service integration.
Let's look at a sample schedule:
schedule: (
("1live", "m3u", "http://www.wdr.de/wdrlive/media/einslive.m3u", (
("hourly", "0M", 60),
("News", "0M", 5)
)),
("wdr2", "m3u", "http://www.wdr.de/wdrlive/media/wdr2.m3u", (
("News", "0M", 5),
("MonTalk", "(19:05 & MON)", 115),
("Zugabe", "(22:30 & FRI)", 60),
("In Concert", "(23:05 & SUN)", 55)
)),
("bbc-r1", "pls", "http://www.bbc.co.uk/worldservice/meta/live/mp3/eneuk.pls", (
("News", "0M", 5)
))
)
A schedule consists of a list of stations, which in turn consist of an identifier ("1live"), a strategy (either "direct" for direct downloads of mp3 streams or "m3u" or "pls" for playlists), an URL ("http://www.wdr.de/wdrlive/media/einslive.m3u") and a list if programmes to be recorded on this station. A programme consists of an identifier, a schedule string and a duration in minutes. The schedule string encodes the conditions which have to be met to start a recording. For example wdr2-In Concert airs every Sunday at 11:05 PM. This can simply be stated as "(23:05 & SUN)". For a more in depth look at schedule strings and their syntax, look at the configuration files located in the etc directory.
There are various listings of streaming urls available on the net. Here are a few:
- http://www.radio-browser.info/gui/#/topclick
- https://wiki.ubuntuusers.de/Internetradio/Stationen/#Radiosender-Deutschland
- http://www.thomas-oestreicher.de/radiostream.htm
- http://doc.ubuntu-fr.org/liste_radio_belgique
- http://mathewpeet.org/lists/BBC_radio_audio_streams/
- http://www.hendrikjansen.nl/henk/streaming.html
If you want to find a stream for a particular radio station it is often helpful to enter " m3u" or " mp3" into a search engine of your choice.
You know how to do this! Just git clone https://github.com/niklasfi/radioman.git.
To build radioman, all you need is a c++ compiler of your choice (clang++ and g++ come to mind) and cmake. You will find them in your distributions package manager.
Please make sure to have libconfig and boost (particularly the system, filesystem, date_time and parse sublibraries) available on your system. Don't be afraid you should be able to install them from your favourite package manager. You will probably need the respective development packages usually ending in -devel or -dev as well.
If all prerequisites are installed properly, all you need to do is to run the following three commands:
cd bin
cmake ../src
make
Edit one of the etc/config.* files, or create one of your own. The sample config files are very verbose and contain a detailed explanation about creating your own recording shedule. Ensure to set your preferred destinationPath in which all your recordings will be stored.
Running radioman is pretty simple. Just pass your config file as a parameter to the main executable
bin/radioman path/to/your/config
There is a sample systemd service file in the etc directory. You can adapt it to your needs by changing the User and Group as well as the path to the binary and config in the ExecStart setting. Once you are done, copy it to etc/systemd/system/radioman.service or create a symlink pointing to your local service file in this location. Finally you need to tell systemd to reload its configuration files by executing sudo systemctl daemon-reload.
To start the radioman service, call
sudo systemctl start radioman
You can check the status of the radioman service with
sudo systemctl status feedme core
To ensure the radioman service is run at startup you need to enable it. To do so:
sudo systemctl enable radioman
If you kept the StandardOutput setting in your service file, you can follow radioman's logs using
sudo journalctl -u radioman -f -o cat
Of course it is! It all started out with a simple recording script which was called via cron jobs at regular intervals. For reference, this is it:
#! /bin/bash
tmp=$(mktemp)
curl -s ${url} | wget -q -O "${tmp}" -i - &
id=$!
sleep ${rectime}
(kill ${id}) 2>/dev/null
chmod +r "${tmp}"
mv "${tmp}" "${dst}"
If you want to try it, put it in a file called download.sh, mark it as executable and run it with
dst=$(pwd) rectime=60 url=http://www.wdr.de/wdrlive/media/einslive.m3u ./download.sh
It turned out, that, while this is a good way to get started, it did not handle dropping connections very well; or at least not as well as I would have liked it to. So I wrote a node.js service which was a mess and never really worked and finally settled to redo it in a properly typed language with sane error handling.
In the end, the C++ solution is elegant in its own way.
- It utilizes worker threads which ensure that their assigned stream is always available.
- The use of boost_parse allows for a very concise way to formulate recording shedules as time constraints.
- It handles pls playslists and direct downloads, which the above approach does not.
- It just works.